<!DOCTYPE html
PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" lang="en-us" xml:lang="en-us">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8"></meta>
<meta http-equiv="X-UA-Compatible" content="IE=edge"></meta>
<meta name="copyright" content="(C) Copyright 2005"></meta>
<meta name="DC.rights.owner" content="(C) Copyright 2005"></meta>
<meta name="DC.Type" content="concept"></meta>
<meta name="DC.Title" content="nvJPEG Library"></meta>
<meta name="abstract" content="A GPU-accelerated JPEG decoding library."></meta>
<meta name="description" content="A GPU-accelerated JPEG decoding library."></meta>
<meta name="DC.Coverage" content="CUDA API References"></meta>
<meta name="DC.subject" content="nvJPEG, CUDA nvJPEG"></meta>
<meta name="keywords" content="nvJPEG, CUDA nvJPEG"></meta>
<meta name="DC.Format" content="XHTML"></meta>
<meta name="DC.Identifier" content="abstract"></meta>
<link rel="stylesheet" type="text/css" href="../common/formatting/commonltr.css"></link>
<link rel="stylesheet" type="text/css" href="../common/formatting/site.css"></link>
<title>nvJPEG :: CUDA Toolkit Documentation</title>
<!--[if lt IE 9]>
<script src="../common/formatting/html5shiv-printshiv.min.js"></script>
<![endif]-->
<script type="text/javascript" charset="utf-8" src="//assets.adobedtm.com/b92787824f2e0e9b68dc2e993f9bd995339fe417/satelliteLib-7ba51e58dc61bcb0e9311aadd02a0108ab24cc6c.js"></script>
<script type="text/javascript" charset="utf-8" src="../common/formatting/jquery.min.js"></script>
<script type="text/javascript" charset="utf-8" src="../common/formatting/jquery.ba-hashchange.min.js"></script>
<script type="text/javascript" charset="utf-8" src="../common/formatting/jquery.scrollintoview.min.js"></script>
<script type="text/javascript" src="../search/htmlFileList.js"></script>
<script type="text/javascript" src="../search/htmlFileInfoList.js"></script>
<script type="text/javascript" src="../search/nwSearchFnt.min.js"></script>
<script type="text/javascript" src="../search/stemmers/en_stemmer.min.js"></script>
<script type="text/javascript" src="../search/index-1.js"></script>
<script type="text/javascript" src="../search/index-2.js"></script>
<script type="text/javascript" src="../search/index-3.js"></script>
<link rel="canonical" href="http://docs.nvidia.com/cuda/nvjpeg/index.html"></link>
<link rel="stylesheet" type="text/css" href="../common/formatting/qwcode.highlight.css"></link>
</head>
<body>
<header id="header"><span id="company">NVIDIA</span><span id="site-title">CUDA Toolkit Documentation</span><form id="search" method="get" action="search">
<input type="text" name="search-text"></input><fieldset id="search-location">
<legend>Search In:</legend>
<label><input type="radio" name="search-type" value="site"></input>Entire Site</label>
<label><input type="radio" name="search-type" value="document"></input>Just This Document</label></fieldset>
<button type="reset">clear search</button>
<button id="submit" type="submit">search</button></form>
</header>
<div id="site-content">
<nav id="site-nav">
<div class="category closed"><a href="../index.html" title="The root of the site.">CUDA Toolkit
v10.1.168</a></div>
<div class="category"><a href="index.html" title="nvJPEG ">nvJPEG </a></div>
<ul>
<li>
<div class="section-link"><a href="#introduction">1. Introduction</a></div>
<ul>
<li>
<div class="section-link"><a href="#nvjpeg-decoder-introduction">1.1. nvJPEG Decoder</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-encoder-introduction">1.2. nvJPEG Encoder</a></div>
</li>
</ul>
</li>
<li>
<div class="section-link"><a href="#title-nvjpeg-decode">2. JPEG Decoding</a></div>
<ul>
<li>
<div class="section-link"><a href="#using-nvjpeg-api">2.1. Using JPEG Decoding</a></div>
<ul>
<li>
<div class="section-link"><a href="#nvjpeg-single-image-decoding">2.1.1. Single Image Decoding</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-decode-by-phases">2.1.2. Decode by Phases</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-batched-image-decoding">2.1.3. Batched Image Decoding</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-batched-single-phase">2.1.4. Single Phase</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-batched-multiple-phases">2.1.5. Multiple Phases</a></div>
</li>
</ul>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-type-declarations">2.2. nvJPEG Type Declarations</a></div>
<ul>
<li>
<div class="section-link"><a href="#nvjpeg-memory-allocator-interface">2.2.1. nvJPEG Device Memory Allocator Interface</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-host-pinned-memory-allocator-interface">2.2.2. nvJPEG Host Pinned Memory Allocator Interface</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-opaque-lib-handle-struct">2.2.3. nvJPEG Opaque Library Handle Struct</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-opaque-jpeg-decoding-state-handle">2.2.4. nvJPEG Opaque JPEG Decoding State Handle</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-output-pointer-struct">2.2.5. nvJPEG Output Pointer Struct</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-backend">2.2.6. nvJPEG Backend</a></div>
</li>
</ul>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-api-reference">2.3. nvJPEG API Reference</a></div>
<ul>
<li>
<div class="section-link"><a href="#nvjpeg-helper-api-reference">2.3.1. nvJPEG Helper API Reference</a></div>
<ul>
<li>
<div class="section-link"><a href="#nvjpeg-get-property">2.3.1.1. nvjpegGetProperty()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-create">2.3.1.2. nvjpegCreate()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-create-simple">2.3.1.3. nvjpegCreateSimple()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-create-ex">2.3.1.4. nvjpegCreateEx()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpegDestroy">2.3.1.5. nvjpegDestroy()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-set-device-mem-padding">2.3.1.6. nvjpegSetDeviceMemoryPadding()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-get-device-mem-padding">2.3.1.7. nvjpegGetDeviceMemoryPadding()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-set-pinned-mem-padding">2.3.1.8. nvjpegSetPinnedMemoryPadding()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-get-pinned-mem-padding">2.3.1.9. nvjpegGetPinnedMemoryPadding()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpegJpegStateCreate">2.3.1.10. nvjpegJpegStateCreate()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpegJpegStateDestroy">2.3.1.11. nvjpegJpegStateDestroy()</a></div>
</li>
</ul>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-retrieve-image-info-api">2.3.2. Retrieve Encoded Image Information API</a></div>
<ul>
<li>
<div class="section-link"><a href="#nvjpegGetImageInfo">2.3.2.1. nvjpegGetImageInfo()</a></div>
</li>
</ul>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-decode-api-single-phase">2.3.3. Decode API -- Single Phase</a></div>
<ul>
<li>
<div class="section-link"><a href="#nvjpeg-decode">2.3.3.1. nvjpegDecode()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-decode-batched-init">2.3.3.2. nvjpegDecodeBatchedInitialize()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-decode-batched">2.3.3.3. nvjpegDecodeBatched()</a></div>
</li>
</ul>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-decode-multiple-phase-api">2.3.4. Decode API -- Multiple Phases</a></div>
<ul>
<li>
<div class="section-link"><a href="#nvjpeg-decode-phaseone">2.3.4.1. nvjpegDecodePhaseOne()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-decode-phasetwo">2.3.4.2. nvjpegDecodePhaseTwo()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-decode-phasethree">2.3.4.3. nvjpegDecodePhaseThree()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-decode-batched-phaseone">2.3.4.4. nvjpegDecodeBatchedPhaseOne()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-decode-batched-phasetwo">2.3.4.5. nvjpegDecodeBatchedPhaseTwo()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-decode-batched-phasethree">2.3.4.6.
nvjpegDecodeBatchedPhaseThree()</a></div>
</li>
</ul>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-api-return-codes">2.3.5. nvjpeg-api-return-codes</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-chroma-subsampling">2.3.6. nvjpeg-chroma-subsampling</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-reference-docs">2.3.7. Reference Documents</a></div>
</li>
</ul>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-examples">2.4. Examples of nvJPEG</a></div>
</li>
</ul>
</li>
<li>
<div class="section-link"><a href="#title-nvjpeg-encode">3. JPEG Encoding</a></div>
<ul>
<li>
<div class="section-link"><a href="#using-encoder">3.1. Using the Encoder</a></div>
<ul>
<li>
<div class="section-link"><a href="#encoding-parameters">3.1.1. Encoding the Parameters</a></div>
</li>
<li>
<div class="section-link"><a href="#encoding-state">3.1.2. Encoding the State</a></div>
</li>
<li>
<div class="section-link"><a href="#encoding-the-image">3.1.3. Encoding the Image</a></div>
<ul>
<li>
<div class="section-link"><a href="#using-nvjpegEncodeYUV">3.1.3.1. nvjpegEncodeYUV</a></div>
</li>
<li>
<div class="section-link"><a href="#using-nvjpegEncodeImage">3.1.3.2. nvjpegEncodeImage</a></div>
</li>
</ul>
</li>
<li>
<div class="section-link"><a href="#retrieving-compressed-stream">3.1.4. Retrieving the Compressed Stream </a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-encode-examples">3.1.5. JPEG Encoding Example</a></div>
</li>
</ul>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-encoder-type-declarations">3.2. nvJPEG Encoder Type Declarations</a></div>
<ul>
<li>
<div class="section-link"><a href="#nvjpeg-inputformat">3.2.1. nvjpegInputFormat_t</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-encoder-state">3.2.2. nvjpegEncoderState_t</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-encoder-params">3.2.3. nvjpegEncoderParams_t</a></div>
</li>
</ul>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-encoder-helper-api-reference">3.3. nvJPEG Encoder Helper API Reference</a></div>
<ul>
<li>
<div class="section-link"><a href="#nvjpeg-encoder-state-create">3.3.1. nvjpegEncoderStateCreate()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-encoder-state-destroy">3.3.2. nvjpegEncoderStateDestroy()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-encoder-params-create">3.3.3. nvjpegEncoderParamsCreate()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-encoder-params-destroy">3.3.4. nvjpegEncoderParamsDestroy()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-encoder-params-set-quality">3.3.5. nvjpegEncoderParamsSetQuality()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-encoder-params-set-optimized-huffman">3.3.6. nvjpegEncoderParamsSetOptimizedHuffman()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-encoder-params-set-sampling-factors">3.3.7. nvjpegEncoderParamsSetSamplingFactors()</a></div>
</li>
</ul>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-encoder-api-reference">3.4. nvJPEG Encoder API Reference</a></div>
<ul>
<li>
<div class="section-link"><a href="#nvjpeg-encoder-get-buffer-size">3.4.1. nvjpegEncodeGetBufferSize()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-encode-yuv">3.4.2. nvjpegEncodeYUV()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-encode-image">3.4.3. nvjpegEncodeImage()</a></div>
</li>
<li>
<div class="section-link"><a href="#nvjpeg-encode-retrieve-bitstream">3.4.4. nvjpegEncodeRetrieveBitstream()</a></div>
</li>
</ul>
</li>
</ul>
</li>
</ul>
</nav>
<div id="resize-nav"></div>
<nav id="search-results">
<h2>Search Results</h2>
<ol></ol>
</nav>
<div id="contents-container">
<div id="breadcrumbs-container">
<div id="release-info">nvJPEG
(<a href="../../pdf/nvJPEG.pdf">PDF</a>)
-
v10.1.168
(<a href="https://developer.nvidia.com/cuda-toolkit-archive">older</a>)
-
Last updated April 24, 2019
-
<a href="mailto:CUDAIssues@nvidia.com?subject=CUDA Toolkit Documentation Feedback: nvJPEG ">Send Feedback</a></div>
</div>
<article id="contents">
<div class="topic nested0" id="abstract"><a name="abstract" shape="rect">
<!-- --></a><h2 class="title topictitle1"><a href="#abstract" name="abstract" shape="rect">nvJPEG Library</a></h2>
<div class="body conbody">
<p class="shortdesc">A GPU-accelerated JPEG decoding library.</p>
</div>
</div>
<div class="topic concept nested0" xml:lang="en-us" id="introduction"><a name="introduction" shape="rect">
<!-- --></a><h2 class="title topictitle1"><a href="#introduction" name="introduction" shape="rect">1. Introduction</a></h2>
<div class="body conbody"></div>
<div class="topic concept nested1" xml:lang="en-us" id="nvjpeg-decoder-introduction"><a name="nvjpeg-decoder-introduction" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-decoder-introduction" name="nvjpeg-decoder-introduction" shape="rect">1.1. nvJPEG Decoder</a></h3>
<div class="body conbody">
<p class="p">The nvJPEG 1.0 library provides high-performance, GPU accelerated JPEG decoding
functionality for image formats commonly used in deep learning and hyperscale multimedia
applications. The library offers single and batched JPEG decoding capabilities which
efficiently utilize the available GPU resources for optimum performance; and the flexibility
for users to manage the memory allocation needed for decoding.
</p>
<p class="p">The nvJPEG library enables the following functions: use the JPEG image data stream as
input; retrieve the width and height of the image from the data stream, and use this
retrieved information to manage the GPU memory allocation and the decoding. A dedicated
API is provided for retrieving the image information from the raw JPEG image data
stream.
</p>
<div class="note tip"><span class="tiptitle">Tip:</span> Throughout this document, the terms “CPU” and “Host” are used synonymously. Similarly,
the terms “GPU” and “Device” are synonymous.
</div>
<p class="p">The nvJPEG library supports the following:</p>
<p class="p"><strong class="ph b">JPEG options:</strong></p>
<ul class="ul">
<li dir="ltr" class="li">
<p dir="ltr" class="p">Baseline and Progressive JPEG decoding</p>
</li>
<li dir="ltr" class="li">
<p dir="ltr" class="p">8 bits per pixel</p>
</li>
<li dir="ltr" class="li">
<p dir="ltr" class="p">Huffman bitstream decoding</p>
</li>
<li dir="ltr" class="li">
<p dir="ltr" class="p">3 color channels (YCbCr) or 1 color channel (Grayscale)</p>
</li>
<li dir="ltr" class="li">
<p dir="ltr" class="p">8- and 16-bit quantization tables</p>
</li>
<li dir="ltr" class="li">
<p dir="ltr" class="p">The following chroma subsampling for the 3 color channels Y,
Cb, Cr (Y, U, V):
</p><a name="nvjpeg-decoder-introduction__ul_ld2_qxs_d2b" shape="rect">
<!-- --></a><ul class="ul" id="nvjpeg-decoder-introduction__ul_ld2_qxs_d2b">
<li dir="ltr" class="li">
<p dir="ltr" class="p">4:4:4 </p>
</li>
<li dir="ltr" class="li">
<p dir="ltr" class="p">4:2:2 </p>
</li>
<li dir="ltr" class="li">
<p dir="ltr" class="p">4:2:0 </p>
</li>
<li dir="ltr" class="li">
<p dir="ltr" class="p">4:4:0 </p>
</li>
<li dir="ltr" class="li">
<p dir="ltr" class="p">4:1:1 and </p>
</li>
<li dir="ltr" class="li">
<p dir="ltr" class="p">4:1:0</p>
</li>
</ul>
</li>
</ul>
<p class="p"><strong class="ph b">Features:</strong></p>
<div class="p"><a name="nvjpeg-decoder-introduction__ul_pkt_c1t_d2b" shape="rect">
<!-- --></a><ul class="ul" id="nvjpeg-decoder-introduction__ul_pkt_c1t_d2b">
<li dir="ltr" class="li">
<p dir="ltr" class="p">Hybrid decoding using both the CPU (i.e., host) and the GPU (i.e.,
device).
</p>
</li>
<li dir="ltr" class="li">
<p dir="ltr" class="p">Input to the library is in the host memory, and the output is in
the GPU memory.
</p>
</li>
<li dir="ltr" class="li">
<p dir="ltr" class="p">Single image and batched image decoding. </p>
</li>
<li dir="ltr" class="li">
<p dir="ltr" class="p">Single phase and multiple phases decoding.</p>
</li>
<li dir="ltr" class="li">
<p dir="ltr" class="p">Color space conversion.</p>
</li>
<li dir="ltr" class="li">
<p dir="ltr" class="p">User-provided memory manager for the device allocations.</p>
</li>
</ul>
</div>
</div>
</div>
<div class="topic concept nested1" xml:lang="en-us" id="nvjpeg-encoder-introduction"><a name="nvjpeg-encoder-introduction" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-encoder-introduction" name="nvjpeg-encoder-introduction" shape="rect">1.2. nvJPEG Encoder</a></h3>
<div class="body conbody">
<p class="p">The encoding functions of the nvJPEG library perform GPU-accelerated compression of user's
image data to the JPEG bitstream. User can provide input data in a number of formats and
colorspaces, and control the encoding process with parameters. Encoding functionality will
allocate temporary buffers using user-provided memory allocator.
</p>
<p class="p">Before calling the encoding functions the user should perform a few prerequisite steps
using the helper functions described in <a class="xref" href="index.html#nvjpeg-encoder-helper-api-reference" shape="rect">nvJPEG Encoder Helper API Reference</a>.
</p>
</div>
</div>
</div>
<div class="topic concept nested0" xml:lang="en-us" id="title-nvjpeg-decode"><a name="title-nvjpeg-decode" shape="rect">
<!-- --></a><h2 class="title topictitle1"><a href="#title-nvjpeg-decode" name="title-nvjpeg-decode" shape="rect">2. JPEG Decoding</a></h2>
<div class="body conbody"></div>
<div class="topic concept nested1" xml:lang="en-us" id="using-nvjpeg-api"><a name="using-nvjpeg-api" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#using-nvjpeg-api" name="using-nvjpeg-api" shape="rect">2.1. Using JPEG Decoding</a></h3>
<div class="body conbody">
<p class="p">The nvJPEG library provides functions for both the decoding of a single image, and batched
decoding of multiple images.
</p>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-single-image-decoding"><a name="nvjpeg-single-image-decoding" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-single-image-decoding" name="nvjpeg-single-image-decoding" shape="rect">2.1.1. Single Image Decoding</a></h3>
<div class="body conbody">
<p class="p">For single-image decoding you provide the data size and a pointer to the file data, and the
decoded image is placed in the output buffer.
</p>
<p class="p">To use the nvJPEG library, start by calling the helper functions for initialization. </p><a name="nvjpeg-single-image-decoding__ul_w1k_qp5_d2b" shape="rect">
<!-- --></a><ol class="ol" id="nvjpeg-single-image-decoding__ul_w1k_qp5_d2b">
<li class="li">Create nvJPEG library handle with one of the helper functions <samp class="ph codeph">nvjpegCreateSimple()
or nvjpegCreateEx()</samp>.
</li>
<li dir="ltr" class="li">
<p class="p">Create JPEG state with the helper function <samp class="ph codeph">nvjpegJpegStateCreate()</samp>. See
<a class="xref" href="index.html#nvjpeg-type-declarations" shape="rect">nvJPEG Type Declarations</a> and <samp class="ph codeph"><a class="xref" href="index.html#nvjpegJpegStateCreate" shape="rect">nvjpegJpegStateCreate()</a></samp>.
</p>
<p class="p">Below is the list of helper functions available in the nvJPEG library:</p>
<div class="p"><a name="nvjpeg-single-image-decoding__ul_y1k_qp5_d2b" shape="rect">
<!-- --></a><ul class="ul" id="nvjpeg-single-image-decoding__ul_y1k_qp5_d2b">
<li dir="ltr" class="li"><samp class="ph codeph"> nvjpegStatus_t nvjpegGetProperty(libraryPropertyType
type, int *value); </samp></li>
<li dir="ltr" class="li"><samp class="ph codeph"> [DEPRECATED] nvjpegStatus_t nvjpegCreate(nvjpegBackend_t backend,
nvjpegHandle_t *handle , nvjpeg_dev_allocator allocator); </samp></li>
<li class="li"><samp class="ph codeph">nvjpegStatus_t nvjpegCreateSimple(nvjpegHandle_t *handle); </samp></li>
<li class="li"><samp class="ph codeph">nvjpegStatus_t nvjpegCreateEx(nvjpegBackend_t backend, nvjpegDevAllocator_t
*dev_allocator, nvjpegPinnedAllocator_t *pinned_allocator, unsigned int flags,
nvjpegHandle_t *handle); </samp></li>
<li dir="ltr" class="li"><samp class="ph codeph"> nvjpegStatus_t nvjpegDestroy(nvjpegHandle_t handle);
</samp></li>
<li dir="ltr" class="li"><samp class="ph codeph"> nvjpegStatus_t nvjpegJpegStateCreate(nvjpegHandle_t
handle, nvjpegJpegState_t *jpeg_handle); </samp></li>
<li dir="ltr" class="li"><samp class="ph codeph"> nvjpegStatus_t nvjpegJpegStateDestroy(nvjpegJpegState
handle); </samp></li>
<li class="li">Other helper functions such as <samp class="ph codeph">nvjpegSet*()</samp> and
<samp class="ph codeph">nvjpegGet*()</samp> can be used to configure the library functionality on
per-handle basis. Refer to the <a class="xref" href="index.html#nvjpeg-helper-api-reference" shape="rect">helper API
reference</a> for more details.
</li>
</ul>
</div>
</li>
<li dir="ltr" class="li">
<p class="p">Retrieve the width and height information from the JPEG-encoded image by using the
<samp class="ph codeph">nvjpegGetImageInfo()</samp> function. See also <samp class="ph codeph"><a class="xref" href="index.html#nvjpegGetImageInfo" shape="rect">nvjpegGetImageInfo()</a></samp>.
</p>
<p class="p">Below is the signature of <samp class="ph codeph">nvjpegGetImageInfo() </samp>function:
</p>
<div class="p"><pre xml:space="preserve">
nvjpegStatus_t nvjpegGetImageInfo(
nvjpegHandle_t handle,
const unsigned char *data,
size_t length,
int *nComponents,
nvjpegChromaSubsampling_t *subsampling,
int *widths,
int *heights); </pre></div>
<p class="p">For each image to be decoded, pass the JPEG data pointer and data length to the above
function. The <samp class="ph codeph">nvjpegGetImageInfo()</samp> function is thread safe.
</p>
</li>
<li dir="ltr" class="li">
<p class="p">One of the outputs of the above <samp class="ph codeph">nvjpegGetImageInfo()</samp> function is
<samp class="ph codeph">nvjpegChromaSubsampling_t</samp>. This parameter is an enum type,
and its enumerator list is composed of the chroma subsampling property retrieved
from the JPEG image. See <a class="xref" href="index.html#nvjpeg-chroma-subsampling" shape="rect">nvJPEG Chroma
Subsampling</a>.
</p>
</li>
<li dir="ltr" class="li">
<p class="p">Use the <samp class="ph codeph">nvjpegDecode()</samp> function in the nvJPEG library to decode this
single JPEG image. See the signature of this function below:
</p>
<div class="p"><pre xml:space="preserve">
nvjpegStatus_t nvjpegDecode(
nvjpegHandle_t handle,
nvjpegJpegState_t jpeg_handle,
const unsigned char *data,
size_t length,
nvjpegOutputFormat_t output_format,
nvjpegImage_t *destination,
cudaStream_t stream);</pre></div>
<p dir="ltr" class="p">In the above <samp class="ph codeph">nvjpegDecode()</samp> function, the parameters
<samp class="ph codeph">nvjpegOutputFormat_t</samp>, <samp class="ph codeph">nvjpegImage_t</samp>, and
<samp class="ph codeph">cudaStream_t</samp> can be used to set the output behavior of the
<samp class="ph codeph">nvjpegDecode() </samp>function. You provide the <samp class="ph codeph">cudaStream_t</samp>
parameter to indicate the stream to which your asynchronous tasks are submitted.
</p>
</li>
<li dir="ltr" class="li">
<p class="p"><strong class="ph b">The <samp class="ph codeph">nvjpegOutputFormat_t</samp> parameter:</strong></p>
<p class="p">The <samp class="ph codeph">nvjpegOutputFormat_t</samp> parameter can be set to one of the
<samp class="ph codeph">output_format</samp> settings below:
</p>
<div class="tablenoborder"><a name="nvjpeg-single-image-decoding__table_rbk_qp5_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-single-image-decoding__table_rbk_qp5_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="42.016806722689076%" rowspan="1" colspan="1"><strong class="ph b">output_format</strong></td>
<td class="entry" dir="ltr" valign="top" width="57.98319327731092%" rowspan="1" colspan="1"><strong class="ph b">Meaning</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="42.016806722689076%" rowspan="1" colspan="1"><samp class="ph codeph">NVJPEG_OUTPUT_UNCHANGED</samp></td>
<td class="entry" dir="ltr" valign="top" width="57.98319327731092%" rowspan="1" colspan="1">Return the decoded image planar format.</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="42.016806722689076%" rowspan="1" colspan="1"><samp class="ph codeph">NVJPEG_OUTPUT_RGB</samp></td>
<td class="entry" dir="ltr" valign="top" width="57.98319327731092%" rowspan="1" colspan="1">Convert to planar RGB.</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="42.016806722689076%" rowspan="1" colspan="1"><samp class="ph codeph">NVJPEG_OUTPUT_BGR</samp></td>
<td class="entry" dir="ltr" valign="top" width="57.98319327731092%" rowspan="1" colspan="1">Convert to planar BGR.</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="42.016806722689076%" rowspan="1" colspan="1"><samp class="ph codeph">NVJPEG_OUTPUT_RGBI </samp></td>
<td class="entry" dir="ltr" valign="top" width="57.98319327731092%" rowspan="1" colspan="1">Convert to interleaved RGB.</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="42.016806722689076%" rowspan="1" colspan="1"><samp class="ph codeph">NVJPEG_OUTPUT_BGRI</samp></td>
<td class="entry" dir="ltr" valign="top" width="57.98319327731092%" rowspan="1" colspan="1">Convert to interleaved BGR.</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="42.016806722689076%" rowspan="1" colspan="1"><samp class="ph codeph">NVJPEG_OUTPUT_Y</samp></td>
<td class="entry" dir="ltr" valign="top" width="57.98319327731092%" rowspan="1" colspan="1">Return the Y component only.</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="42.016806722689076%" rowspan="1" colspan="1"><samp class="ph codeph">NVJPEG_OUTPUT_YUV</samp></td>
<td class="entry" dir="ltr" valign="top" width="57.98319327731092%" rowspan="1" colspan="1">Return in the YUV planar format.</td>
</tr>
</tbody>
</table>
</div>
<p class="p">For example, if the output_format is set to <samp class="ph codeph">NVJPEG_OUTPUT_Y</samp> or
<samp class="ph codeph">NVJPEG_OUTPUT_RGBI</samp>, or <samp class="ph codeph">NVJPEG_OUTPUT_BGRI</samp>
then the output is written only to channel[0], and the other channels are not
touched.
</p>
<p class="p">Alternately, in the case of planar output, the data is written to the corresponding channels of
the <samp class="ph codeph">nvjpegImage_t</samp> destination structure.
</p>
<p class="p">Finally, in the case of grayscale JPEG and RGB output, the luminance is used to create the
grayscale RGB.
</p>
</li>
<li dir="ltr" class="li">
<p class="p">As mentioned above, an important benefit of the
<samp class="ph codeph">nvjpegGetImageInfo()</samp>function is the ability to utilize the image
information retrieved from the the input JPEG image to allocate proper GPU
memory for your decoding operation.
</p>
<p class="p">The <samp class="ph codeph"><a class="xref" href="index.html#nvjpegGetImageInfo" shape="rect">nvjpegGetImageInfo()</a></samp>
function returns the <samp class="ph codeph">widths</samp>, <samp class="ph codeph">heights</samp> and
<samp class="ph codeph">nComponents</samp> parameters.
</p>
<div class="p"><pre xml:space="preserve">nvjpegStatus_t nvjpegGetImageInfo(
nvjpegHandle_t handle,
const unsigned char *data,
size_t length,
int *nComponents,
nvjpegChromaSubsampling_t *subsampling,
int *widths,
int *heights); </pre></div>
<p class="p">You can use the retrieved parameters, <samp class="ph codeph">widths</samp>, <samp class="ph codeph">heights</samp>
and <samp class="ph codeph">nComponents</samp>, to calculate the required size for the output
buffers, either for a single decoded JPEG, or for every decoded JPEG in a batch.
</p>
<p class="p">To optimally set the <samp class="ph codeph">destination</samp> parameter for the
<samp class="ph codeph">nvjpegDecode()</samp> function, use the following guidelines:
</p>
<div class="tablenoborder"><a name="nvjpeg-single-image-decoding__table_tbk_qp5_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-single-image-decoding__table_tbk_qp5_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" valign="top" width="33.850931677018636%" rowspan="1" colspan="1">
<p dir="ltr" class="p"><strong class="ph b">For the </strong><strong class="ph b">output_format:</strong></p>
<p dir="ltr" class="p">NVJPEG_OUTPUT_Y </p>
</td>
<td class="entry" dir="ltr" valign="top" width="35.09316770186335%" rowspan="1" colspan="1"><strong class="ph b">destination.pitch[0] should be at least: </strong>width[0]
</td>
<td class="entry" dir="ltr" valign="top" width="31.055900621118017%" rowspan="1" colspan="1"><strong class="ph b">destination.channel[0] should be at least of size:
</strong>destination.pitch[0]*height[0]
</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="33.850931677018636%" rowspan="1" colspan="1"><strong class="ph b">For the </strong><strong class="ph b">output_format</strong></td>
<td class="entry" dir="ltr" valign="top" width="35.09316770186335%" rowspan="1" colspan="1"><strong class="ph b">destination.pitch[c] should be at least:</strong></td>
<td class="entry" dir="ltr" valign="top" width="31.055900621118017%" rowspan="1" colspan="1"><strong class="ph b">destination.channel[c] should be at least of size:</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="33.850931677018636%" rowspan="1" colspan="1">NVJPEG_OUTPUT_YUV </td>
<td class="entry" dir="ltr" valign="top" width="35.09316770186335%" rowspan="1" colspan="1">width[c] for c = 0, 1, 2</td>
<td class="entry" dir="ltr" valign="top" width="31.055900621118017%" rowspan="1" colspan="1">destination.pitch[c]*height[c] for c = 0, 1, 2</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="33.850931677018636%" rowspan="1" colspan="1">NVJPEG_OUTPUT_RGB and NVJPEG_OUTPUT_BGR</td>
<td class="entry" dir="ltr" valign="top" width="35.09316770186335%" rowspan="1" colspan="1">width[0] for c = 0, 1, 2</td>
<td class="entry" dir="ltr" valign="top" width="31.055900621118017%" rowspan="1" colspan="1">destination.pitch[0]*height[0] for c = 0, 1, 2</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="33.850931677018636%" rowspan="1" colspan="1">NVJPEG_OUTPUT_RGBI and NVJPEG_OUTPUT_BGRI</td>
<td class="entry" valign="top" width="35.09316770186335%" rowspan="1" colspan="1">width[0]*3</td>
<td class="entry" dir="ltr" valign="top" width="31.055900621118017%" rowspan="1" colspan="1">destination.pitch[0]*height[0]</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="33.850931677018636%" rowspan="1" colspan="1">NVJPEG_OUTPUT_UNCHANGED</td>
<td class="entry" dir="ltr" valign="top" width="35.09316770186335%" rowspan="1" colspan="1">width[c] for c = [ 0, nComponents - 1 ]</td>
<td class="entry" dir="ltr" valign="top" width="31.055900621118017%" rowspan="1" colspan="1">destination.pitch[c]*height[c] for c = [ 0, nComponents -
1]
</td>
</tr>
</tbody>
</table>
</div>
</li>
<li dir="ltr" class="li">
<p class="p">Ensure that the <samp class="ph codeph">nvjpegImage_t</samp> structure (or structures, in the case of
batched decode) is filled with the pointers and pitches of allocated buffers.
The <samp class="ph codeph">nvjpegImage_t</samp> structure that holds the output pointers is
defined as follows:
</p>
<div class="p"><pre xml:space="preserve">typedef struct
{
unsigned char * channel[NVJPEG_MAX_COMPONENT];
unsigned int pitch[NVJPEG_MAX_COMPONENT];
} nvjpegImage_t;</pre></div>
<p dir="ltr" class="p">NVJPEG_MAX_COMPONENT is the maximum number of color components the
nvJPEG library supports in the current release. For generic images, this is the
maximum number of encoded channels that the library is able to decompress.
</p>
</li>
<li class="li">
<p class="p">Finally, when you call the <samp class="ph codeph">nvjpegDecode()</samp> function with the parameters as
described above, the <samp class="ph codeph">nvjpegDecode()</samp> function fills the output
buffers with the decoded data.
</p>
</li>
</ol>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-decode-by-phases"><a name="nvjpeg-decode-by-phases" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-decode-by-phases" name="nvjpeg-decode-by-phases" shape="rect">2.1.2. <strong class="ph b" id="docs-internal-guid-1f19484d-00d7-8d33-84a5-17a946e41d29"><a name="docs-internal-guid-1f19484d-00d7-8d33-84a5-17a946e41d29" shape="rect">
<!-- --></a>Decode by
Phases</strong></a></h3>
<div class="body conbody">
<p class="p">Alternately, you can decode a single image in multiple phases. This gives you flexibility in
controlling the flow, and optimizing the decoding process.
</p>
<p class="p">To decode an image in multiple phases, follow these steps:</p><a name="nvjpeg-decode-by-phases__ul_w1k_qp5_d2b" shape="rect">
<!-- --></a><ol class="ol" id="nvjpeg-decode-by-phases__ul_w1k_qp5_d2b">
<li dir="ltr" class="li">Just as when you are decoding in a single phase, create the JPEG state with
the helper function <samp class="ph codeph">nvjpegJpegStateCreate()</samp>.
</li>
<li dir="ltr" class="li">
<div class="p">Next, call the functions in the sequence below (see <a class="xref" href="index.html#nvjpeg-decode-multiple-phase-api" shape="rect">Decode API -- Multiple
Phases.</a>) <a name="nvjpeg-decode-by-phases__ul_crh_31v_d2b" shape="rect">
<!-- --></a><ul class="ul" id="nvjpeg-decode-by-phases__ul_crh_31v_d2b">
<li class="li"><samp class="ph codeph">nvjpegDecodePhaseOne()</samp></li>
<li class="li"><samp class="ph codeph">nvjpegDecodePhaseTwo()</samp></li>
<li class="li"><samp class="ph codeph">nvjpegDecodePhaseThree()</samp></li>
</ul>
</div>
</li>
<li dir="ltr" class="li">
<p class="p"> At the conclusion of the third phase, the <samp class="ph codeph">nvjpegDecodePhaseThree()</samp>
function writes the decoded output at the memory location pointed to by its
<samp class="ph codeph">*destination</samp> parameter.
</p>
</li>
</ol>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-batched-image-decoding"><a name="nvjpeg-batched-image-decoding" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-batched-image-decoding" name="nvjpeg-batched-image-decoding" shape="rect">2.1.3. Batched Image Decoding</a></h3>
<div class="body conbody">
<p class="p">For the batched image decoding you provide pointers to multiple file data in the memory, and
also provide the buffer sizes for each file data. The nvJPEG library will decode these
multiple images, and will place the decoded data in the output buffers that you specified in
the parameters.
</p>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-batched-single-phase"><a name="nvjpeg-batched-single-phase" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-batched-single-phase" name="nvjpeg-batched-single-phase" shape="rect">2.1.4. Single Phase</a></h3>
<div class="body conbody">
<p class="p"> For batched image decoding in single phase, follow these steps:</p><a name="nvjpeg-batched-single-phase__ul_ykh_nbv_d2b" shape="rect">
<!-- --></a><ol class="ol" id="nvjpeg-batched-single-phase__ul_ykh_nbv_d2b">
<li dir="ltr" class="li">
<p dir="ltr" class="p">Call <samp class="ph codeph">nvjpegDecodeBatchedInitialize()</samp> function to
initialize the batched decoder. Specify the batch size in the
<samp class="ph codeph">batch_size</samp> parameter. See <samp class="ph codeph"><a class="xref" href="index.html#nvjpeg-decode-batched-init" shape="rect">nvjpegDecodeBatchedInitialize()</a></samp>.
</p>
</li>
<li dir="ltr" class="li">
<p dir="ltr" class="p">Next, call <samp class="ph codeph">nvjpegDecodeBatched()</samp> for each new batch.
Make sure to pass the parameters that are correct to the specific batch of
images. If the size of the batch changes, or if the batch decoding fails, then
call the <samp class="ph codeph">nvjpegDecodeBatchedInitialize()</samp> function again.
</p>
</li>
</ol>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-batched-multiple-phases"><a name="nvjpeg-batched-multiple-phases" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-batched-multiple-phases" name="nvjpeg-batched-multiple-phases" shape="rect">2.1.5. Multiple Phases</a></h3>
<div class="body conbody">
<p class="p">To decode a batch of images in multiple phases, follow these steps: </p>
<div class="note note"><span class="notetitle">Note:</span><p class="p">This is the only case where the JPEG state could be used by multiple threads at the same
time.
</p>
</div><a name="nvjpeg-batched-multiple-phases__ul_nrh_kcv_d2b" shape="rect">
<!-- --></a><ol class="ol" id="nvjpeg-batched-multiple-phases__ul_nrh_kcv_d2b">
<li class="li">Create the JPEG state with the helper function <samp class="ph codeph">nvjpegJpegStateCreate()</samp>.
</li>
<li class="li">Call the <samp class="ph codeph">nvjpegDecodeBatchedInitialize()</samp> function to initialize the
batched decoder. Specify the batch size in the <samp class="ph codeph">batch_size</samp>
parameter, and specify the <samp class="ph codeph">max_cpu_threads</samp> parameter to set the
maximum number of CPU threads that work on single batch.
</li>
<li class="li">Batched processing is done by calling the functions for the specific phases in
sequence:<a name="nvjpeg-batched-multiple-phases__ul_y44_ncv_d2b" shape="rect">
<!-- --></a><ul class="ul" id="nvjpeg-batched-multiple-phases__ul_y44_ncv_d2b">
<li class="li">In the first phase, call <samp class="ph codeph">nvjpegDecodePhaseOne()</samp> for each
image in the batch, according to the index of the image in the batch. Note
that this could be done using multiple threads. If multiple threads are used
then the thread index in the range [0, max_cpu_threads-1] should be provided
to the <samp class="ph codeph">nvjpegDecodeBatchedPhaseOne()</samp> function. Before
proceeding to the next phase, ensure that the
<samp class="ph codeph">nvjpegDecodePhaseOne()</samp> calls for every image have
finished.
</li>
<li class="li">Next, call <samp class="ph codeph">nvjpegDecodePhaseTwo().</samp>.
</li>
<li class="li">Finally, call <samp class="ph codeph">nvjpegDecodePhaseThree().</samp>.
</li>
</ul>
</li>
<li class="li">If you have another batch of images of the same size to process, then repeat from 3. </li>
</ol>
</div>
</div>
</div>
<div class="topic concept nested1" xml:lang="en-us" id="nvjpeg-type-declarations"><a name="nvjpeg-type-declarations" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-type-declarations" name="nvjpeg-type-declarations" shape="rect">2.2. nvJPEG Type Declarations</a></h3>
<div class="body conbody"></div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-memory-allocator-interface"><a name="nvjpeg-memory-allocator-interface" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-memory-allocator-interface" name="nvjpeg-memory-allocator-interface" shape="rect">2.2.1. nvJPEG Device Memory Allocator Interface</a></h3>
<div class="body conbody">
<div class="p"><pre xml:space="preserve">typedef int (*tDevMalloc)(void**, size_t);
typedef int (*tDevFree)(void*);
typedef struct
{
tDevMalloc dev_malloc;
tDevFree dev_free;
} nvjpegDevAllocator_t;</pre></div>
<p class="p">Users can tell the library to use their own device memory allocator. The function prototypes
for the memory allocation and memory freeing functions are similar to the
<samp class="ph codeph">cudaMalloc() </samp>and <samp class="ph codeph">cudaFree()</samp> functions. They should
return 0 in case of success, and non-zero otherwise. A pointer to the
<samp class="ph codeph">nvjpegDevAllocator_t</samp> structure, with properly filled fields, should
be provided to the <samp class="ph codeph">nvjpegCreate()</samp> function. NULL is accepted, in which
case the default memory allocation functions <samp class="ph codeph">cudaMalloc()</samp> and
<samp class="ph codeph">cudaFree()</samp> is used.
</p>
<p class="p">When the <samp class="ph codeph">nvjpegDevAllocator_t *allocator</samp> parameter in the
<samp class="ph codeph">nvjpegCreate()</samp> or <samp class="ph codeph">nvjpegCreateEx()</samp> function is set as a
pointer to the above <samp class="ph codeph">nvjpegDevAllocator_t</samp> structure, then this structure is
used for allocating and releasing the device memory. The function prototypes for the memory
allocation and memory freeing functions are similar to the <samp class="ph codeph">cudaMalloc()</samp> and
<samp class="ph codeph">cudaFree()</samp> functions. They should return 0 in case of success, and non-zero
otherwise.
</p>
<p class="p">However, if the <samp class="ph codeph">nvjpegDevAllocator_t *allocator</samp> parameter in the
<samp class="ph codeph">nvjpegCreate()</samp> or <samp class="ph codeph">nvjpegCreateEx()</samp> function is set to
NULL, then the default memory allocation functions <samp class="ph codeph">cudaMalloc()</samp> and
<samp class="ph codeph">cudaFree()</samp> will be used. When using <samp class="ph codeph">nvjpegCreateSimple()</samp>
function to create library handle the default device memory allocator will be used.
</p>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-host-pinned-memory-allocator-interface"><a name="nvjpeg-host-pinned-memory-allocator-interface" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-host-pinned-memory-allocator-interface" name="nvjpeg-host-pinned-memory-allocator-interface" shape="rect">2.2.2. nvJPEG Host Pinned Memory Allocator Interface</a></h3>
<div class="body conbody">
<div class="p"><pre xml:space="preserve">typedef int (*tPinnedMalloc)(void**, size_t, unsigned int flags);
typedef int (*tPinnedFree)(void*);
typedef struct
{
tPinnedMalloc pinned_malloc;
tPinnedFree pinned_free;
} nvjpegPinnedAllocator_t; </pre></div>
<p class="p">When the <samp class="ph codeph">nvjpegPinnedAllocator_t *allocator</samp> parameter in the
<samp class="ph codeph">nvjpegCreateEx()</samp> function is set as a pointer to the above
<samp class="ph codeph">nvjpegPinnedAllocator_t</samp> structure, then this structure will be used for
allocating and releasing host pinned memory for copying data to/from device. The function
prototypes for the memory allocation and memory freeing functions are similar to the
<samp class="ph codeph">cudaHostAlloc()</samp> and <samp class="ph codeph">cudaFreeHost()</samp> functions. They will
return 0 in case of success, and non-zero otherwise.
</p>
<p class="p">However, if the <samp class="ph codeph">nvjpegPinnedAllocator_t *allocator</samp> parameter in the
<samp class="ph codeph">nvjpegCreateEx()</samp> function is set to NULL, then the default memory
allocation functions <samp class="ph codeph">cudaHostAlloc()</samp> and <samp class="ph codeph">cudaFreeHost()</samp> will
be used. When using <samp class="ph codeph">nvjpegCreate()</samp> or <samp class="ph codeph">nvjpegCreateSimple()</samp>
function to create library handle, the default host pinned memory allocator will be used.
</p>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-opaque-lib-handle-struct"><a name="nvjpeg-opaque-lib-handle-struct" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-opaque-lib-handle-struct" name="nvjpeg-opaque-lib-handle-struct" shape="rect">2.2.3. nvJPEG Opaque Library Handle Struct</a></h3>
<div class="body conbody">
<div class="p"><pre xml:space="preserve">struct nvjpegHandle;
typedef struct nvjpegHandle* nvjpegHandle_t;</pre></div>
<p class="p">The library handle is used in any consecutive nvJPEG library calls, and should be initialized
first.
</p>
<p class="p">The library handle is thread safe, and can be used by multiple threads simultaneously. </p>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-opaque-jpeg-decoding-state-handle"><a name="nvjpeg-opaque-jpeg-decoding-state-handle" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-opaque-jpeg-decoding-state-handle" name="nvjpeg-opaque-jpeg-decoding-state-handle" shape="rect">2.2.4. nvJPEG Opaque JPEG Decoding State Handle</a></h3>
<div class="body conbody">
<div class="p"><pre xml:space="preserve">struct nvjpegJpegState;
typedef struct nvjpegJpegState* nvjpegJpegState_t;</pre></div>
<p class="p">The <samp class="ph codeph">nvjpegJpegState</samp> structure stores the temporary JPEG information. It
should be initialized before any usage. This JPEG state handle can be reused after being used
in another decoding. The same JPEG handle should be used across the decoding phases for the
same image or batch. Multiple threads are allowed to share the JPEG state handle only when
processing same batch during first phase (<samp class="ph codeph">nvjpegDecodePhaseOne</samp>) .
</p>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-output-pointer-struct"><a name="nvjpeg-output-pointer-struct" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-output-pointer-struct" name="nvjpeg-output-pointer-struct" shape="rect">2.2.5. nvJPEG Output Pointer Struct</a></h3>
<div class="body conbody">
<div class="p"><pre xml:space="preserve">typedef struct
{
unsigned char * channel[NVJPEG_MAX_COMPONENT];
unsigned int pitch[NVJPEG_MAX_COMPONENT];
} nvjpegImage_t;</pre></div>
<p class="p">The <samp class="ph codeph">nvjpegImage_t </samp>struct holds the pointers to the output buffers, and holds
the corresponding strides of those buffers for the image decoding.
</p>
<p class="p">See <a class="xref" href="index.html#nvjpeg-single-image-decoding" shape="rect">Single Image Decoding</a> on how to set up
the <samp class="ph codeph">nvjpegImage_t</samp> struct.
</p>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-backend"><a name="nvjpeg-backend" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-backend" name="nvjpeg-backend" shape="rect">2.2.6. nvJPEG Backend</a></h3>
<div class="body conbody">
<div class="p"><pre xml:space="preserve">typedef enum {
NVJPEG_BACKEND_DEFAULT = 0,
NVJPEG_BACKEND_HYBRID = 1,
NVJPEG_BACKEND_GPU_HYBRID = 2
} nvjpegBackend_t; </pre></div>
<p class="p">The <samp class="ph codeph">nvjpegBackend_t</samp> enum is used to select either default back-end by
default, or use GPU decoding for baseline JPEG images, or use CPU for Huffman decoding.
</p>
<div class="p">
<div class="tablenoborder"><a name="nvjpeg-backend__table_ogq_y1x_bgb" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-backend__table_ogq_y1x_bgb" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" valign="top" rowspan="1" colspan="1"><strong class="ph b">Member</strong></td>
<td class="entry" valign="top" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" rowspan="1" colspan="1">NVJPEG_BACKEND_DEFAULT</td>
<td class="entry" valign="top" rowspan="1" colspan="1">Default back-end is selected internally</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" rowspan="1" colspan="1">NVJPEG_BACKEND_HYBRID</td>
<td class="entry" dir="ltr" valign="top" rowspan="1" colspan="1">Uses CPU for Huffman decoding</td>
</tr>
<tr class="row">
<td class="entry" valign="top" rowspan="1" colspan="1">NVJPEG_BACKEND_GPU_HYBRID</td>
<td class="entry" valign="top" rowspan="1" colspan="1">The function <samp class="ph codeph">nvjpegDecodeBatched</samp> will use GPU decoding for the
baseline JPEG images with interleaved scan when batch size is greater than 100. For
other JPEG types it uses CPU. Other nvJPEG decode APIs will continue to use CPU for
Huffman decode.
</td>
</tr>
</tbody>
</table>
</div>
</div>
</div>
</div>
</div>
<div class="topic concept nested1" xml:lang="en-us" id="nvjpeg-api-reference"><a name="nvjpeg-api-reference" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-api-reference" name="nvjpeg-api-reference" shape="rect">2.3. nvJPEG API Reference</a></h3>
<div class="body conbody">
<p class="p">This section describes the nvJPEG API. </p>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-helper-api-reference"><a name="nvjpeg-helper-api-reference" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-helper-api-reference" name="nvjpeg-helper-api-reference" shape="rect">2.3.1. nvJPEG Helper API Reference</a></h3>
<div class="body conbody">
<p class="p">The nvJPEG helper functions are used for initializing. </p>
</div>
<div class="topic concept nested3" xml:lang="en-us" id="nvjpeg-get-property"><a name="nvjpeg-get-property" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-get-property" name="nvjpeg-get-property" shape="rect">2.3.1.1. nvjpegGetProperty()</a></h3>
<div class="body conbody">
<p class="p">Gets the numeric value for the major or minor version, or the patch level, of the nvJPEG
library.
</p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">
nvjpegStatus_t nvjpegGetProperty(
libraryPropertyType type,
int *value);</pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-get-property__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-get-property__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">libraryPropertyType type</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">One of the supported <samp class="ph codeph">libraryPropertyType</samp> values, that is,
MAJOR_VERSION, MINOR_VERSION or PATCH_LEVEL.
</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">int *value</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Output</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">The numeric value corresponding to the specific
<samp class="ph codeph">libraryPropertyType</samp> requested.
</td>
</tr>
</tbody>
</table>
</div>
<p dir="ltr" class="p"><strong class="ph b">Returns:</strong></p>
<p dir="ltr" class="p"><samp class="ph codeph">nvjpegStatus_t</samp> - An error code as specified in <a class="xref" href="index.html#nvjpeg-api-return-codes" shape="rect">nvJPEG API Return Codes</a>.
</p>
</div>
</div>
<div class="topic concept nested3" xml:lang="en-us" id="nvjpeg-create"><a name="nvjpeg-create" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-create" name="nvjpeg-create" shape="rect">2.3.1.2. nvjpegCreate()</a></h3>
<div class="body conbody">
<p class="p">Allocates and initializes the library handle. </p>
<div class="p" dir="ltr" id="nvjpeg-create__docs-internal-guid-676727d0-7fff-7e38-82db-42dee025468d"><a name="nvjpeg-create__docs-internal-guid-676727d0-7fff-7e38-82db-42dee025468d" shape="rect">
<!-- --></a><div class="note note"><span class="notetitle">Note:</span> This function is deprecated. Use either <samp class="ph codeph">nvjpegCreateSimple()</samp> or
<samp class="ph codeph">nvjpegCreateEx()</samp> functions to create the library handle.
</div>
</div>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">
nvjpegStatus_t nvjpegCreate(
nvjpegBackend_t backend,
nvjpegDevAllocator_t *allocator,
nvjpegHandle_t *handle);</pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-create__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-create__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegBackend_t backend</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">A backend parameter for the library. This backend will be used for all the
functions called with this handle. If this is set to DEFAULT then it
automatically chooses one of the underlying algorithms.
</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegDevAllocator_t *allocator</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">Device memory allocator. See <samp class="ph codeph">nvjpegDevAllocator_t </samp>structure
description. If NULL is provided, then the default CUDA runtime
<samp class="ph codeph">cudaMalloc() </samp>and <samp class="ph codeph">cudaFree()</samp>
functions will be used.
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegHandle_t *handle</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">The library handle.</td>
</tr>
</tbody>
</table>
</div>
<p class="p">The <samp class="ph codeph">nvjpegBackend_t</samp> parameter is an <samp class="ph codeph">enum</samp> type, with the below
enumerated list values:
</p>
<div class="p"><pre xml:space="preserve">
typedef enum {
NVJPEG_BACKEND_DEFAULT = 0,
NVJPEG_BACKEND_HYBRID = 1,
} nvjpegBackend_t; </pre></div>
<p dir="ltr" class="p"><strong class="ph b">Returns:</strong></p>
<p dir="ltr" class="p"><samp class="ph codeph">nvjpegStatus_t</samp> - An error code as specified in <a class="xref" href="index.html#nvjpeg-api-return-codes" shape="rect">nvJPEG API Return Codes</a>.
</p>
</div>
</div>
<div class="topic concept nested3" xml:lang="en-us" id="nvjpeg-create-simple"><a name="nvjpeg-create-simple" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-create-simple" name="nvjpeg-create-simple" shape="rect">2.3.1.3. nvjpegCreateSimple()</a></h3>
<div class="body conbody">
<p class="p">Allocates and initializes the library handle, with default codec implementations selected
by library and default memory allocators.
</p>
<p dir="ltr" class="p" id="nvjpeg-create-simple__docs-internal-guid-676727d0-7fff-7e38-82db-42dee025468d"><a name="nvjpeg-create-simple__docs-internal-guid-676727d0-7fff-7e38-82db-42dee025468d" shape="rect">
<!-- --></a></p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">nvjpegStatus_t nvjpegCreateSimple(nvjpegHandle_t *handle);</pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-create-simple__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-create-simple__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegHandle_t *handle</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">The library handle.</td>
</tr>
</tbody>
</table>
</div>
<p class="p"></p>
<p dir="ltr" class="p"><strong class="ph b">Returns:</strong></p>
<p dir="ltr" class="p"><samp class="ph codeph">nvjpegStatus_t</samp> - An error code as specified in <a class="xref" href="index.html#nvjpeg-api-return-codes" shape="rect">nvJPEG API Return Codes</a>.
</p>
</div>
</div>
<div class="topic concept nested3" xml:lang="en-us" id="nvjpeg-create-ex"><a name="nvjpeg-create-ex" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-create-ex" name="nvjpeg-create-ex" shape="rect">2.3.1.4. nvjpegCreateEx()</a></h3>
<div class="body conbody">
<p dir="ltr" class="p" id="nvjpeg-create-ex__docs-internal-guid-2060f851-7fff-fdbe-5643-5e1376af1004"><a name="nvjpeg-create-ex__docs-internal-guid-2060f851-7fff-fdbe-5643-5e1376af1004" shape="rect">
<!-- --></a>Allocates and
initializes the library handle using the provided arguments.
</p>
<p dir="ltr" class="p" id="nvjpeg-create-ex__docs-internal-guid-676727d0-7fff-7e38-82db-42dee025468d"><a name="nvjpeg-create-ex__docs-internal-guid-676727d0-7fff-7e38-82db-42dee025468d" shape="rect">
<!-- --></a></p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">nvjpegStatus_t nvjpegCreateEx(nvjpegBackend_t backend,
nvjpegDevAllocator_t *dev_allocator,
nvjpegPinnedAllocator_t *pinned_allocator,
unsigned int flags,
nvjpegHandle_t *handle);</pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-create-ex__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-create-ex__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegBackend_t backend</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input </td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">A backend parameter for the library. This backend will be
used for all of the functions called with this handle. If this is set to
“DEFAULT" then it will choose one of underlying algorithms
automatically.
</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegDevAllocator_t
*dev_allocator</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">Device memory allocator. See <samp class="ph codeph"><a class="xref" href="index.html#nvjpeg-memory-allocator-interface" shape="rect">nvjpegDevAllocator</a>_t</samp> structure description. If
NULL is provided, then the default CUDA runtime functions
<samp class="ph codeph">cudaMalloc()</samp> and <samp class="ph codeph">cudaFree()</samp> will
be used.
</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegPinnedAllocator_t
*pinned_allocator</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">Pinned host memory allocator. See <samp class="ph codeph"><a class="xref" href="index.html#nvjpeg-host-pinned-memory-allocator-interface" shape="rect">nvjpegPinnedAllocator_t</a></samp>structure description. If NULL is provided, then the default
CUDA runtime functions <samp class="ph codeph">cudaHostAlloc()</samp> and
<samp class="ph codeph">cudaFreeHost()</samp> will be used.
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegHandle_t *handle</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">The library handle.</td>
</tr>
</tbody>
</table>
</div>
<p class="p"></p>
<p dir="ltr" class="p"><strong class="ph b">Returns:</strong></p>
<p dir="ltr" class="p"><samp class="ph codeph">nvjpegStatus_t</samp> - An error code as specified in <a class="xref" href="index.html#nvjpeg-api-return-codes" shape="rect">nvJPEG API Return Codes</a>.
</p>
</div>
</div>
<div class="topic concept nested3" xml:lang="en-us" id="nvjpegDestroy"><a name="nvjpegDestroy" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpegDestroy" name="nvjpegDestroy" shape="rect">2.3.1.5. nvjpegDestroy()</a></h3>
<div class="body conbody">
<p class="p">Releases the library handle.</p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">nvjpegStatus_t nvjpegDestroy(nvjpegHandle_t handle);</pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpegDestroy__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpegDestroy__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegHandle_t handle</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">The library handle to release.</td>
</tr>
</tbody>
</table>
</div>
<p dir="ltr" class="p"><strong class="ph b">Returns:</strong></p>
<p dir="ltr" class="p"><samp class="ph codeph">nvjpegStatus_t</samp> - An error code as specified in <a class="xref" href="index.html#nvjpeg-api-return-codes" shape="rect">nvJPEG API Return Codes</a>.
</p>
</div>
</div>
<div class="topic concept nested3" xml:lang="en-us" id="nvjpeg-set-device-mem-padding"><a name="nvjpeg-set-device-mem-padding" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-set-device-mem-padding" name="nvjpeg-set-device-mem-padding" shape="rect">2.3.1.6. nvjpegSetDeviceMemoryPadding()</a></h3>
<div class="body conbody">
<p dir="ltr" class="p" id="nvjpeg-set-device-mem-padding__docs-internal-guid-8dde49d6-7fff-344a-2e88-36e95e842af8"><a name="nvjpeg-set-device-mem-padding__docs-internal-guid-8dde49d6-7fff-344a-2e88-36e95e842af8" shape="rect">
<!-- --></a>Use the provided
padding for all device memory allocations with specified library handle. A large number
will help to amortize the need for device memory reallocations when needed.
</p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">nvjpegStatus_t nvjpegSetDeviceMemoryPadding(
size_t padding,
nvjpegHandle_t handle);</pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-set-device-mem-padding__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-set-device-mem-padding__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">size_t padding</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">Device memory padding to use for all further device memory
allocations.
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegHandle_t *handle</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">The library handle.</td>
</tr>
</tbody>
</table>
</div>
<p dir="ltr" class="p"><strong class="ph b">Returns:</strong></p>
<p dir="ltr" class="p"><samp class="ph codeph">nvjpegStatus_t</samp> - An error code as specified in <a class="xref" href="index.html#nvjpeg-api-return-codes" shape="rect">nvJPEG API Return Codes</a>.
</p>
</div>
</div>
<div class="topic concept nested3" xml:lang="en-us" id="nvjpeg-get-device-mem-padding"><a name="nvjpeg-get-device-mem-padding" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-get-device-mem-padding" name="nvjpeg-get-device-mem-padding" shape="rect">2.3.1.7. nvjpegGetDeviceMemoryPadding()</a></h3>
<div class="body conbody">
<p dir="ltr" class="p" id="nvjpeg-get-device-mem-padding__docs-internal-guid-1402e43d-7fff-008e-d28c-5b66e027a024"><a name="nvjpeg-get-device-mem-padding__docs-internal-guid-1402e43d-7fff-008e-d28c-5b66e027a024" shape="rect">
<!-- --></a>Retrieve the
device memory padding that is currently used for the specified library handle.
</p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">nvjpegStatus_t nvjpegGetDeviceMemoryPadding(
size_t *padding,
nvjpegHandle_t handle);</pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-get-device-mem-padding__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-get-device-mem-padding__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">size_t *padding</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Output</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">Device memory padding that is currently used for device memory
allocations.
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegHandle_t *handle</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">The library handle.</td>
</tr>
</tbody>
</table>
</div>
<p dir="ltr" class="p"><strong class="ph b">Returns:</strong></p>
<p dir="ltr" class="p"><samp class="ph codeph">nvjpegStatus_t</samp> - An error code as specified in <a class="xref" href="index.html#nvjpeg-api-return-codes" shape="rect">nvJPEG API Return Codes</a>.
</p>
</div>
</div>
<div class="topic concept nested3" xml:lang="en-us" id="nvjpeg-set-pinned-mem-padding"><a name="nvjpeg-set-pinned-mem-padding" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-set-pinned-mem-padding" name="nvjpeg-set-pinned-mem-padding" shape="rect">2.3.1.8. nvjpegSetPinnedMemoryPadding()</a></h3>
<div class="body conbody">
<p dir="ltr" class="p" id="nvjpeg-set-pinned-mem-padding__docs-internal-guid-3335cc3f-7fff-2518-f4da-e363e823c042"><a name="nvjpeg-set-pinned-mem-padding__docs-internal-guid-3335cc3f-7fff-2518-f4da-e363e823c042" shape="rect">
<!-- --></a>Use the provided
padding for all pinned host memory allocations with specified library handle. A large
number will help to amortize the need for pinned host memory reallocations when
needed.
</p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">nvjpegStatus_t nvjpegSetPinnedMemoryPadding(
size_t padding,
nvjpegHandle_t handle);</pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-set-pinned-mem-padding__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-set-pinned-mem-padding__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">size_t padding</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">Pinned host memory padding to use for all further pinned host memory
allocations.
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegHandle_t handle</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">The library handle.</td>
</tr>
</tbody>
</table>
</div>
<p dir="ltr" class="p"><strong class="ph b">Returns:</strong></p>
<p dir="ltr" class="p"><samp class="ph codeph">nvjpegStatus_t</samp> - An error code as specified in <a class="xref" href="index.html#nvjpeg-api-return-codes" shape="rect">nvJPEG API Return Codes</a>.
</p>
</div>
</div>
<div class="topic concept nested3" xml:lang="en-us" id="nvjpeg-get-pinned-mem-padding"><a name="nvjpeg-get-pinned-mem-padding" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-get-pinned-mem-padding" name="nvjpeg-get-pinned-mem-padding" shape="rect">2.3.1.9. nvjpegGetPinnedMemoryPadding()</a></h3>
<div class="body conbody">
<p dir="ltr" class="p" id="nvjpeg-get-pinned-mem-padding__docs-internal-guid-ce0aef02-7fff-10a3-aa77-5eee0839d6f5"><a name="nvjpeg-get-pinned-mem-padding__docs-internal-guid-ce0aef02-7fff-10a3-aa77-5eee0839d6f5" shape="rect">
<!-- --></a>Retrieve the
pinned host memory padding that is currently used for specified library handle.
</p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">nvjpegStatus_t nvjpegGetPinnedMemoryPadding(
size_t *padding,
nvjpegHandle_t handle);</pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-get-pinned-mem-padding__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-get-pinned-mem-padding__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">size_t *padding</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Output</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">Pinned host memory padding that is currently used for pinned host memory
allocations.
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegHandle_t handle</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">The library handle.</td>
</tr>
</tbody>
</table>
</div>
<p dir="ltr" class="p"><strong class="ph b">Returns:</strong></p>
<p dir="ltr" class="p"><samp class="ph codeph">nvjpegStatus_t</samp> - An error code as specified in <a class="xref" href="index.html#nvjpeg-api-return-codes" shape="rect">nvJPEG API Return Codes</a>.
</p>
</div>
</div>
<div class="topic concept nested3" xml:lang="en-us" id="nvjpegJpegStateCreate"><a name="nvjpegJpegStateCreate" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpegJpegStateCreate" name="nvjpegJpegStateCreate" shape="rect">2.3.1.10. nvjpegJpegStateCreate()</a></h3>
<div class="body conbody">
<p class="p">Allocates and initializes the internal structure required for the JPEG processing. </p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">
nvjpegStatus_t nvjpegJpegStateCreate(
nvjpegHandle_t handle,
nvjpegJpegState_t *jpeg_handle);</pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpegJpegStateCreate__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpegJpegStateCreate__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="36.65086887835703%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="21.32701421800948%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="15.797788309636651%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="26.224328593996837%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="36.65086887835703%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegHandle_t handle</samp></td>
<td class="entry" dir="ltr" valign="top" width="21.32701421800948%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="15.797788309636651%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="26.224328593996837%" rowspan="1" colspan="1">The library handle.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.65086887835703%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegJpegState_t *jpeg_handle</samp></td>
<td class="entry" valign="top" width="21.32701421800948%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" valign="top" width="15.797788309636651%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="26.224328593996837%" rowspan="1" colspan="1">The image state handle.</td>
</tr>
</tbody>
</table>
</div>
<p dir="ltr" class="p"><strong class="ph b">Returns:</strong></p>
<p dir="ltr" class="p"><samp class="ph codeph">nvjpegStatus_t</samp> - An error code as specified in <a class="xref" href="index.html#nvjpeg-api-return-codes" shape="rect">nvJPEG API Return Codes</a>.
</p>
</div>
</div>
<div class="topic concept nested3" xml:lang="en-us" id="nvjpegJpegStateDestroy"><a name="nvjpegJpegStateDestroy" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpegJpegStateDestroy" name="nvjpegJpegStateDestroy" shape="rect">2.3.1.11. nvjpegJpegStateDestroy()</a></h3>
<div class="body conbody">
<p class="p">Releases the image internal structure. </p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">nvjpegStatus_t nvjpegJpegStateDestroy(nvjpegJpegState handle); </pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpegJpegStateDestroy__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpegJpegStateDestroy__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="36.65086887835703%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="21.32701421800948%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="15.797788309636651%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="26.224328593996837%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="36.65086887835703%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegJpegState handle</samp></td>
<td class="entry" dir="ltr" valign="top" width="21.32701421800948%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" dir="ltr" valign="top" width="15.797788309636651%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="26.224328593996837%" rowspan="1" colspan="1">The image state handle.</td>
</tr>
</tbody>
</table>
</div>
<p dir="ltr" class="p"><strong class="ph b">Returns:</strong></p>
<p dir="ltr" class="p"><samp class="ph codeph">nvjpegStatus_t</samp> - An error code as specified in <a class="xref" href="index.html#nvjpeg-api-return-codes" shape="rect">nvJPEG API Return Codes</a>.
</p>
</div>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-retrieve-image-info-api"><a name="nvjpeg-retrieve-image-info-api" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-retrieve-image-info-api" name="nvjpeg-retrieve-image-info-api" shape="rect">2.3.2. Retrieve Encoded Image Information API</a></h3>
<div class="body conbody">
<p class="p">The helper functions for retrieving the encoded image information. </p>
</div>
<div class="topic concept nested3" xml:lang="en-us" id="nvjpegGetImageInfo"><a name="nvjpegGetImageInfo" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpegGetImageInfo" name="nvjpegGetImageInfo" shape="rect">2.3.2.1. nvjpegGetImageInfo()</a></h3>
<div class="body conbody">
<p class="p">Decodes the JPEG header and retrieves the basic information about the image. </p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">
nvjpegStatus_t nvjpegGetImageInfo(
nvjpegHandle_t handle,
const unsigned char *data,
size_t length,
int *nComponents,
nvjpegChromaSubsampling_t *subsampling,
int *widths,
int *heights); </pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpegGetImageInfo__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpegGetImageInfo__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="16.29901960784314%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="12.254901960784313%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.80392156862745%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegHandle_t handle</samp></td>
<td class="entry" dir="ltr" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">The library handle.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">const unsigned char *data</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">Pointer to the encoded data.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">size_t length</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">Size of the encoded data in bytes.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">int *nComponents</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Output</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">Chroma subsampling for the 1- or 3- channel encoding.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">int *widths</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Output</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">Pointer to the first element of array of size NVJPEG_MAX_COMPONENT, where the width of
each channel (up to NVJPEG_MAX_COMPONENT) will be saved. If the channel
is not encoded, then the corresponding value would be zero.
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">int *heights</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Output</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">Pointer to the first element of array of size NVJPEG_MAX_COMPONENT, where the height of
each channel (up to NVJPEG_MAX_COMPONENT) will be saved. If the channel
is not encoded, then the corresponding value would be zero.
</td>
</tr>
</tbody>
</table>
</div>
<p dir="ltr" class="p"><strong class="ph b">Returns:</strong></p>
<p dir="ltr" class="p"><samp class="ph codeph">nvjpegStatus_t</samp> - An error code as specified in <a class="xref" href="index.html#nvjpeg-api-return-codes" shape="rect">nvJPEG API Return Codes</a>.
</p>
</div>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-decode-api-single-phase"><a name="nvjpeg-decode-api-single-phase" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-decode-api-single-phase" name="nvjpeg-decode-api-single-phase" shape="rect">2.3.3. Decode API -- Single Phase</a></h3>
<div class="body conbody">
<p class="p">Functions for decoding single image or batched images in a single phase.
</p>
</div>
<div class="topic concept nested3" xml:lang="en-us" id="nvjpeg-decode"><a name="nvjpeg-decode" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-decode" name="nvjpeg-decode" shape="rect">2.3.3.1. nvjpegDecode()</a></h3>
<div class="body conbody">
<p class="p">Decodes a single image, and writes the decoded image in the desired format to the output
buffers. This function is asynchronous with respect to the host. All GPU tasks for this
function will be submitted to the provided stream.
</p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">
nvjpegStatus_t nvjpegDecode(
nvjpegHandle_t handle,
nvjpegJpegState_t jpeg_handle,
const unsigned char *data,
size_t length,
nvjpegOutputFormat_t output_format,
nvjpegImage_t *destination,
cudaStream_t stream); </pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-decode__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-decode__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="16.29901960784314%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="12.254901960784313%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.80392156862745%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegHandle_t handle</samp></td>
<td class="entry" dir="ltr" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">The library handle.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegJpegState_t jpeg_handle</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">The image state handle.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">const unsigned char *data</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">Pointer to the encoded data.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">size_t length</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">Size of the encoded data in bytes.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegOutputFormat_t output_format</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">Format in which the decoded output will be saved.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegImage_t *destination</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host/Device</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">Pointer to the structure that describes the output destination. This structure should
be on the host (CPU), but the pointers in this structure should be
pointing to the device (i.e., GPU) memory. See <samp class="ph codeph"><a class="xref" href="index.html#nvjpegGetImageInfo" shape="rect">nvjpegImage_t.</a></samp></td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">cudaStream_t stream</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">The CUDA stream where all of the GPU work will be submitted.</td>
</tr>
</tbody>
</table>
</div>
<p dir="ltr" class="p"><strong class="ph b">Returns:</strong></p>
<p dir="ltr" class="p"><samp class="ph codeph">nvjpegStatus_t</samp> - An error code as specified in <a class="xref" href="index.html#nvjpeg-api-return-codes" shape="rect">nvJPEG API Return Codes</a>.
</p>
</div>
</div>
<div class="topic concept nested3" xml:lang="en-us" id="nvjpeg-decode-batched-init"><a name="nvjpeg-decode-batched-init" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-decode-batched-init" name="nvjpeg-decode-batched-init" shape="rect">2.3.3.2. nvjpegDecodeBatchedInitialize()</a></h3>
<div class="body conbody">
<p class="p">This function initializes the batched decoder state. The initialization parameters include the
batch size, the maximum number of CPU threads, and the specific output format in which
the decoded image will be saved. This function should be called once, prior to decoding
the batches of images. Any currently running batched decoding should be finished before
calling this function.
</p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">
nvjpegStatus_t nvjpegDecodeBatchedInitialize(
nvjpegHandle_t handle,
nvjpegJpegState_t jpeg_handle,
int batch_size,
int max_cpu_threads,
nvjpegOutputFormat_t output_format); </pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-decode-batched-init__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-decode-batched-init__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="16.29901960784314%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="12.254901960784313%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.80392156862745%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegHandle_t handle</samp></td>
<td class="entry" dir="ltr" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">The library handle.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegJpegState_t jpeg_handle</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">The image state handle.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">int batch_size</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">Batch size.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">int max_cpu_threads</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">Maximum number of CPU threads that can participate in decoding a batch.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegOutputFormat_t output_format</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">Format in which the decoded output will be saved.</td>
</tr>
</tbody>
</table>
</div>
<p dir="ltr" class="p"><strong class="ph b">Returns:</strong></p>
<p dir="ltr" class="p"><samp class="ph codeph">nvjpegStatus_t</samp> - An error code as specified in <a class="xref" href="index.html#nvjpeg-api-return-codes" shape="rect">nvJPEG API Return Codes</a>.
</p>
</div>
</div>
<div class="topic concept nested3" xml:lang="en-us" id="nvjpeg-decode-batched"><a name="nvjpeg-decode-batched" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-decode-batched" name="nvjpeg-decode-batched" shape="rect">2.3.3.3. nvjpegDecodeBatched()</a></h3>
<div class="body conbody">
<p class="p">Decodes the batch of images, and writes them to the buffers described in the
<samp class="ph codeph">destination</samp> parameter in a format provided to
<samp class="ph codeph">nvjpegDecodeBatchedInitialize()</samp> function. This function is
asynchronous with respect to the host. All GPU tasks for this function will be submitted
to the provided stream.
</p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">
nvjpegStatus_t nvjpegDecodeBatched(
nvjpegHandle_t handle,
nvjpegJpegState_t jpeg_handle,
const unsigned char *const *data,
const size_t *lengths,
nvjpegImage_t *destinations,
cudaStream_t stream); </pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-decode-batched__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-decode-batched__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="28.681177976952625%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="12.93213828425096%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="12.804097311139564%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="45.58258642765685%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="28.681177976952625%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegHandle_t handle</samp></td>
<td class="entry" dir="ltr" valign="top" width="12.93213828425096%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="12.804097311139564%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="45.58258642765685%" rowspan="1" colspan="1">The library handle.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="28.681177976952625%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegJpegState_t jpeg_handle</samp></td>
<td class="entry" valign="top" width="12.93213828425096%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.804097311139564%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="45.58258642765685%" rowspan="1" colspan="1">The image state handle.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="28.681177976952625%" rowspan="1" colspan="1"><samp class="ph codeph">const unsigned char *const *data</samp></td>
<td class="entry" valign="top" width="12.93213828425096%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.804097311139564%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="45.58258642765685%" rowspan="1" colspan="1">Pointer to the first element of array of the input data. The size of the array is
assumed to be batch_size provided to
<samp class="ph codeph">nvjpegDecodeBatchedInitialize()</samp> batch
initialization function.
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="28.681177976952625%" rowspan="1" colspan="1"><samp class="ph codeph">const size_t *lengths</samp></td>
<td class="entry" valign="top" width="12.93213828425096%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.804097311139564%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="45.58258642765685%" rowspan="1" colspan="1">Pointer to the first element of array of input sizes. Size of array is assumed to be
batch_size provided to <samp class="ph codeph">nvjpegDecodeBatchedInitialize()</samp>,
the batch initialization function.
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="28.681177976952625%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegImage_t *destinations</samp></td>
<td class="entry" valign="top" width="12.93213828425096%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" valign="top" width="12.804097311139564%" rowspan="1" colspan="1">Host/Device</td>
<td class="entry" valign="top" width="45.58258642765685%" rowspan="1" colspan="1">Pointer to the first element of array of output descriptors. The size of array is
assumed to be batch_size provided to
<samp class="ph codeph">nvjpegDecodeBatchedInitialize(),</samp> the batch
initialization function. See also <samp class="ph codeph"><a class="xref" href="index.html#nvjpegGetImageInfo" shape="rect">nvjpegImage_t</a></samp>.
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="28.681177976952625%" rowspan="1" colspan="1"><samp class="ph codeph">
cudaStream_t stream
</samp></td>
<td class="entry" valign="top" width="12.93213828425096%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.804097311139564%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="45.58258642765685%" rowspan="1" colspan="1">
<p dir="ltr" class="p">The CUDA stream where all the GPU work will be
submitted.
</p>
</td>
</tr>
</tbody>
</table>
</div>
<p dir="ltr" class="p"><strong class="ph b">Returns:</strong></p>
<p dir="ltr" class="p"><samp class="ph codeph">nvjpegStatus_t</samp> - An error code as specified in <a class="xref" href="index.html#nvjpeg-api-return-codes" shape="rect">nvJPEG API Return Codes</a>.
</p>
</div>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-decode-multiple-phase-api"><a name="nvjpeg-decode-multiple-phase-api" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-decode-multiple-phase-api" name="nvjpeg-decode-multiple-phase-api" shape="rect">2.3.4. Decode API -- Multiple Phases</a></h3>
<div class="body conbody">
<p class="p">The nvJPEG library provides an ability to control the decoding process in phases. In the
simple case of a single-image decode you can split the decoding into phases. For
decoding multiple images, you can overlap the decoding phases of separate images within
a single thread. Finally, for the batched decode you can use multiple threads to split
the host tasks. Synchronization between phases should be handled with CUDA events and
CUDA stream synchronization mechanisms, by the user.
</p>
<div class="p">
<div class="note note"><span class="notetitle">Note:</span> Note that first phases are synchronous with the respect to the host, while the second and
third phases are asynchronous--for both single image and batched decode.
</div>
</div>
</div>
<div class="topic concept nested3" xml:lang="en-us" id="nvjpeg-decode-phaseone"><a name="nvjpeg-decode-phaseone" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-decode-phaseone" name="nvjpeg-decode-phaseone" shape="rect">2.3.4.1. nvjpegDecodePhaseOne()</a></h3>
<div class="body conbody">
<p class="p">The first phase of a single-image decode. You provide all the inputs, and the nvJPEG library
performs any required preprocessing on the host. Any previous calls to
<samp class="ph codeph">nvjpegDecodePhaseOne()</samp> and <samp class="ph codeph">nvjpegDecodePhaseTwo()</samp>
with the same <samp class="ph codeph">nvjpeg_handle</samp> parameter should be finished prior to this
call.
</p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">
nvjpegStatus_t nvjpegDecodePhaseOne(
nvjpegHandle_t handle,
nvjpegJpegState_t jpeg_handle,
const unsigned char *data,
size_t length,
nvjpegOutputFormat_t output_format,
cudaStream_t stream); </pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-decode-phaseone__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-decode-phaseone__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="16.29901960784314%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="12.254901960784313%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.80392156862745%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegHandle_t handle</samp></td>
<td class="entry" dir="ltr" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">The library handle.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegJpegState_t jpeg_handle</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">The image state handle.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">const unsigned char *data</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">Pointer to the encoded stream. </td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">size_t length</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">Size of the encoded stream. </td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegOutputFormat_t output_format</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">Format in which the decoded image will be saved.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">cudaStream_t stream </samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">
<p dir="ltr" class="p">The CUDA stream where all the GPU work will be
submitted.
</p>
</td>
</tr>
</tbody>
</table>
</div>
<p dir="ltr" class="p"><strong class="ph b">Returns:</strong></p>
<p dir="ltr" class="p"><samp class="ph codeph">nvjpegStatus_t</samp> - An error code as specified in <a class="xref" href="index.html#nvjpeg-api-return-codes" shape="rect">nvJPEG API Return Codes</a>.
</p>
</div>
</div>
<div class="topic concept nested3" xml:lang="en-us" id="nvjpeg-decode-phasetwo"><a name="nvjpeg-decode-phasetwo" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-decode-phasetwo" name="nvjpeg-decode-phasetwo" shape="rect">2.3.4.2. nvjpegDecodePhaseTwo()</a></h3>
<div class="body conbody">
<p class="p">In this second phase of the decoding process, the GPU (that is, the device) is involved. The
decoding task is transferred to the device memory. Any required preprocessing is
performed on the device. Any previous calls to <samp class="ph codeph">nvjpegDecodePhaseTwo()</samp>
and <samp class="ph codeph">nvjpegDecodePhaseThree()</samp> with the same <samp class="ph codeph">jpeg_handle</samp>
parameter should be finished prior to this call.
</p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">
nvjpegStatus_t nvjpegDecodePhaseTwo(
nvjpegHandle_t handle,
nvjpegJpegState_t jpeg_handle,
cudaStream_t stream); </pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-decode-phasetwo__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-decode-phasetwo__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="16.29901960784314%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="12.254901960784313%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.80392156862745%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegHandle_t handle</samp></td>
<td class="entry" dir="ltr" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">The library handle.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegJpegState_t jpeg_handle</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">The image state handle.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">cudaStream_t stream </samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">
<p dir="ltr" class="p">The CUDA stream where all the GPU work will be
submitted.
</p>
</td>
</tr>
</tbody>
</table>
</div>
<p dir="ltr" class="p"><strong class="ph b">Returns:</strong></p>
<p dir="ltr" class="p"><samp class="ph codeph">nvjpegStatus_t</samp> - An error code as specified in <a class="xref" href="index.html#nvjpeg-api-return-codes" shape="rect">nvJPEG API Return Codes</a>.
</p>
</div>
</div>
<div class="topic concept nested3" xml:lang="en-us" id="nvjpeg-decode-phasethree"><a name="nvjpeg-decode-phasethree" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-decode-phasethree" name="nvjpeg-decode-phasethree" shape="rect">2.3.4.3. nvjpegDecodePhaseThree()</a></h3>
<div class="body conbody">
<p class="p">In this third phase of the decoding process, the decoded image is written to the output, in the
specified decoding format.
</p>
<div class="p">
<div class="note note"><span class="notetitle">Note:</span> If the same <samp class="ph codeph">jpeg_handle</samp> is shared for decoding multiple images
simultaneously, then these multiple images should be of the same
<samp class="ph codeph">output_format</samp>.
</div>
</div>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">
nvjpegStatus_t nvjpegDecodePhaseThree(
nvjpegHandle_t handle,
nvjpegJpegState_t jpeg_handle,
nvjpegImage_t *destination,
cudaStream_t stream); </pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-decode-phasethree__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-decode-phasethree__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="16.29901960784314%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="12.254901960784313%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.80392156862745%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegHandle_t handle</samp></td>
<td class="entry" dir="ltr" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">The library handle.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegJpegState_t jpeg_handle</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">The image state handle.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegImage_t *destination</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host/Device</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">Pointer to the structure that describes the output destination. This
structure should be on host, but the pointers in this structure should
be pointing to the device memory. See <samp class="ph codeph">nvjpegImage_t</samp>
description for details.
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">cudaStream_t stream </samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">
<p dir="ltr" class="p">The CUDA stream where all the GPU work will be
submitted.
</p>
</td>
</tr>
</tbody>
</table>
</div>
<p dir="ltr" class="p"><strong class="ph b">Returns:</strong></p>
<p dir="ltr" class="p"><samp class="ph codeph">nvjpegStatus_t</samp> - An error code as specified in <a class="xref" href="index.html#nvjpeg-api-return-codes" shape="rect">nvJPEG API Return Codes</a>.
</p>
</div>
</div>
<div class="topic concept nested3" xml:lang="en-us" id="nvjpeg-decode-batched-phaseone"><a name="nvjpeg-decode-batched-phaseone" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-decode-batched-phaseone" name="nvjpeg-decode-batched-phaseone" shape="rect">2.3.4.4. nvjpegDecodeBatchedPhaseOne()</a></h3>
<div class="body conbody">
<p class="p">This first phase of the batched decoding should be called separately for each image in
the batch. The batch initialization API, with appropriate batch parameters, should be
called prior to starting the task with the batch.
</p>
<p class="p">If the batch parameters (batch size, number of threads, output format) did not change,
then there is no need to initialize the batch again before starting the task.
</p>
<p class="p">It is possible to use multiple threads to split this first phase of the task. In which
case, each thread should have a unique index. Provide the index of the image in the
batch, and use the same JPEG decoding state parameter.
</p>
<p class="p">The thread index for the batch should be in the range of <samp class="ph codeph">[0,
max_cpu_threads-1]</samp>. The image index should be in the range of <samp class="ph codeph">[0,
batch_size-1]</samp>. Any previous calls to
<samp class="ph codeph">nvjpegDecodeBatchedPhaseOne()</samp> and
<samp class="ph codeph">nvjpegDecodeBatchedPhaseTwo() </samp>on a different batch with the same
JPEG state handle parameter should be completed prior to this call.
</p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">
nvjpegStatus_t nvjpegDecodeBatchedPhaseOne(
nvjpegHandle_t handle,
nvjpegJpegState_t jpeg_handle,
const unsigned char *data,
size_t length,
int image_idx,
int thread_idx,
cudaStream_t stream); </pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-decode-batched-phaseone__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-decode-batched-phaseone__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="16.29901960784314%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="12.254901960784313%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.80392156862745%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegHandle_t handle</samp></td>
<td class="entry" dir="ltr" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">The library handle.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegJpegState_t jpeg_handle</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">The image state handle.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">const unsigned char *data</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">Pointer to the encoded stream. </td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">size_t length </samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">
<p dir="ltr" class="p">Size of the encoded stream.</p>
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">int image_idx</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">Image index in the batch. Should be in the range from 0 to
<samp class="ph codeph">batch_size-1</samp>.
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">int thread_idx</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">Thread index that calls this phase. Should be in the range from 0 to
<samp class="ph codeph">max_cpu_threads-1</samp>.
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="36.642156862745104%" rowspan="1" colspan="1"><samp class="ph codeph">cudaStream_t stream</samp></td>
<td class="entry" valign="top" width="16.29901960784314%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="12.254901960784313%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.80392156862745%" rowspan="1" colspan="1">The CUDA stream where all the GPU work will be submitted.</td>
</tr>
</tbody>
</table>
</div>
<p dir="ltr" class="p"><strong class="ph b">Returns:</strong></p>
<p dir="ltr" class="p"><samp class="ph codeph">nvjpegStatus_t</samp> - An error code as specified in <a class="xref" href="index.html#nvjpeg-api-return-codes" shape="rect">nvJPEG API Return Codes</a>.
</p>
</div>
</div>
<div class="topic concept nested3" xml:lang="en-us" id="nvjpeg-decode-batched-phasetwo"><a name="nvjpeg-decode-batched-phasetwo" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-decode-batched-phasetwo" name="nvjpeg-decode-batched-phasetwo" shape="rect">2.3.4.5. <strong class="ph b" id="docs-internal-guid-22afa100-fff2-adb6-ae15-313aec425dbb"><a name="docs-internal-guid-22afa100-fff2-adb6-ae15-313aec425dbb" shape="rect">
<!-- --></a>nvjpegDecodeBatchedPhaseTwo()</strong></a></h3>
<div class="body conbody">
<p class="p">This phase should be called once per batch. It should be called only after the
<samp class="ph codeph">nvjpegDecodeBatchedPhaseOne()</samp> calls for every image in the batch
have finished. Any prior calls to <samp class="ph codeph">nvjpegDecodeBatchedPhaseTwo()</samp>
and<samp class="ph codeph"> nvjpegDecodeBatchedPhaseThree()</samp> for other batches with the same
JPEG state handle parameter should be finished prior this call.
</p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p>
<div class="p"><pre xml:space="preserve">
nvjpegStatus_t nvjpegDecodeBatchedPhaseTwo(
nvjpegHandle_t handle,
nvjpegJpegState_t jpeg_handle,
cudaStream_t stream);</pre></div>
<p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-decode-batched-phasetwo__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-decode-batched-phasetwo__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegHandle_t handle</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">The library handle.</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegJpegState_t jpeg_handle</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">The image state handle.</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">cudaStream_t stream</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">The CUDA stream where all the GPU work will be
submitted.
</td>
</tr>
</tbody>
</table>
</div>
<p dir="ltr" class="p"><strong class="ph b">Returns:</strong></p>
<p dir="ltr" class="p"><samp class="ph codeph">nvjpegStatus_t</samp> - An error code as specified in <a class="xref" href="index.html#nvjpeg-api-return-codes" shape="rect">nvJPEG API Return Codes</a>.
</p>
</div>
</div>
<div class="topic concept nested3" xml:lang="en-us" id="nvjpeg-decode-batched-phasethree"><a name="nvjpeg-decode-batched-phasethree" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-decode-batched-phasethree" name="nvjpeg-decode-batched-phasethree" shape="rect">2.3.4.6. <strong class="ph b" id="docs-internal-guid-7dffeecd-ff83-e295-4423-5cb8355b98ab"><a name="docs-internal-guid-7dffeecd-ff83-e295-4423-5cb8355b98ab" shape="rect">
<!-- --></a>
nvjpegDecodeBatchedPhaseThree()</strong></a></h3>
<div class="body conbody">
<p class="p">This phase should be called once per batch. It should be called only after the
<samp class="ph codeph">nvjpegDecodeBatchedPhaseTwo()</samp> call for the same batch has
finished.
</p>
<p class="p">Between a call to <samp class="ph codeph">nvjpegDecodeBatchedPhaseTwo()</samp> and a call to this function, no
calls are allowed to <samp class="ph codeph"> nvjpegDecodeBatchedPhaseTwo()</samp> or <samp class="ph codeph">nvjpegDecodeBatchedPhaseThree()</samp>
for any other batch with the same JPEG state handle parameter.
</p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">
nvjpegStatus_t nvjpegDecodeBatchedPhaseThree(
nvjpegHandle_t handle,
nvjpegJpegState_t jpeg_handle,
nvjpegImage_t *destinations,
cudaStream_t stream);</pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-decode-batched-phasethree__table_c1m_1ft_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-decode-batched-phasethree__table_c1m_1ft_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="33.51409978308026%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="15.075921908893712%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="10.84598698481562%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="40.563991323210416%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="33.51409978308026%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegHandle_t handle</samp></td>
<td class="entry" dir="ltr" valign="top" width="15.075921908893712%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="10.84598698481562%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="40.563991323210416%" rowspan="1" colspan="1">The library handle.</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="33.51409978308026%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegJpegState_t jpeg_handle</samp></td>
<td class="entry" dir="ltr" valign="top" width="15.075921908893712%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="10.84598698481562%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="40.563991323210416%" rowspan="1" colspan="1">The image state handle.</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="33.51409978308026%" rowspan="1" colspan="1"><samp class="ph codeph">nvjpegImage_t *destinations</samp></td>
<td class="entry" dir="ltr" valign="top" width="15.075921908893712%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" dir="ltr" valign="top" width="10.84598698481562%" rowspan="1" colspan="1">Host/Device</td>
<td class="entry" valign="top" width="40.563991323210416%" rowspan="1" colspan="1">
<p dir="ltr" class="p">Pointer to the first element of the array of output descriptors. The size of the
array is assumed to be the <samp class="ph codeph">batch_size</samp> parameter
that was provided to the batch initialization function. See
<samp class="ph codeph"><a class="xref" href="index.html#nvjpeg-output-pointer-struct" shape="rect">nvjpegImage_t</a></samp> description for details.
</p>
</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="33.51409978308026%" rowspan="1" colspan="1"><samp class="ph codeph">cudaStream_t stream</samp></td>
<td class="entry" dir="ltr" valign="top" width="15.075921908893712%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="10.84598698481562%" rowspan="1" colspan="1">Host</td>
<td class="entry" dir="ltr" valign="top" width="40.563991323210416%" rowspan="1" colspan="1">The CUDA stream to which all the GPU tasks will be
submitted.
</td>
</tr>
</tbody>
</table>
</div>
<p class="p"><strong class="ph b">Returns:</strong></p>
<p dir="ltr" class="p"><samp class="ph codeph">nvjpegStatus_t</samp> - An error code as specified in <a class="xref" href="index.html#nvjpeg-api-return-codes" shape="rect">nvJPEG API Return Codes</a>.
</p>
</div>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-api-return-codes"><a name="nvjpeg-api-return-codes" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-api-return-codes" name="nvjpeg-api-return-codes" shape="rect">2.3.5. nvjpeg-api-return-codes</a></h3>
<div class="body conbody">
<p class="p">The nvJPEG API adheres to the following return codes and their indicators: </p><pre xml:space="preserve">typedef enum
{
NVJPEG_STATUS_SUCCESS = 0,
NVJPEG_STATUS_NOT_INITIALIZED = 1,
NVJPEG_STATUS_INVALID_PARAMETER = 2,
NVJPEG_STATUS_BAD_JPEG = 3,
NVJPEG_STATUS_JPEG_NOT_SUPPORTED = 4,
NVJPEG_STATUS_ALLOCATOR_FAILURE = 5,
NVJPEG_STATUS_EXECUTION_FAILED = 6,
NVJPEG_STATUS_ARCH_MISMATCH = 7,
NVJPEG_STATUS_INTERNAL_ERROR = 8,
} nvjpegStatus_t; </pre><p dir="ltr" class="p"><strong class="ph b">Description of the returned error codes:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-api-return-codes__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-api-return-codes__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="51.28644939965695%" rowspan="1" colspan="1"><strong class="ph b">Returned Error (Returned Code)</strong></td>
<td class="entry" dir="ltr" valign="top" width="48.71355060034305%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="51.28644939965695%" rowspan="1" colspan="1"><samp class="ph codeph">NVJPEG_STATUS_SUCCESS (0)</samp></td>
<td class="entry" dir="ltr" valign="top" width="48.71355060034305%" rowspan="1" colspan="1">The API call has finished successfully. Note that many of the calls are
asynchronous and some of the errors may be seen only after
synchronization.
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="51.28644939965695%" rowspan="1" colspan="1"><samp class="ph codeph">NVJPEG_STATUS_NOT_INITIALIZED (1)</samp></td>
<td class="entry" valign="top" width="48.71355060034305%" rowspan="1" colspan="1">The library handle was not initialized. A call to
<samp class="ph codeph">nvjpegCreate()</samp> is required to initialize the
handle.
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="51.28644939965695%" rowspan="1" colspan="1"><samp class="ph codeph">NVJPEG_STATUS_INVALID_PARAMETER (2)</samp></td>
<td class="entry" valign="top" width="48.71355060034305%" rowspan="1" colspan="1">Wrong parameter was passed. For example, a null pointer as input data, or an image
index not in the allowed range.
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="51.28644939965695%" rowspan="1" colspan="1"><samp class="ph codeph">NVJPEG_STATUS_BAD_JPEG (3) </samp></td>
<td class="entry" valign="top" width="48.71355060034305%" rowspan="1" colspan="1">Cannot parse the JPEG stream. Check that the encoded JPEG stream and
its size parameters are correct.
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="51.28644939965695%" rowspan="1" colspan="1"><samp class="ph codeph">NVJPEG_STATUS_JPEG_NOT_SUPPORTED (4)</samp></td>
<td class="entry" valign="top" width="48.71355060034305%" rowspan="1" colspan="1">Attempting to decode a JPEG stream that is not supported by the
nvJPEG library.
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="51.28644939965695%" rowspan="1" colspan="1"><samp class="ph codeph">NVJPEG_STATUS_ALLOCATOR_FAILURE (5)</samp></td>
<td class="entry" valign="top" width="48.71355060034305%" rowspan="1" colspan="1">The user-provided allocator functions, for either memory allocation
or for releasing the memory, returned a non-zero code.
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="51.28644939965695%" rowspan="1" colspan="1"><samp class="ph codeph">NVJPEG_STATUS_EXECUTION_FAILED (6)</samp></td>
<td class="entry" valign="top" width="48.71355060034305%" rowspan="1" colspan="1">Error during the execution of the device tasks.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="51.28644939965695%" rowspan="1" colspan="1"><samp class="ph codeph">NVJPEG_STATUS_ARCH_MISMATCH (7)</samp></td>
<td class="entry" valign="top" width="48.71355060034305%" rowspan="1" colspan="1">The device capabilities are not enough for the set of input parameters provided (input
parameters such as backend, encoded stream parameters, output
format).
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="51.28644939965695%" rowspan="1" colspan="1"><samp class="ph codeph">NVJPEG_STATUS_INTERNAL_ERROR (8)</samp></td>
<td class="entry" valign="top" width="48.71355060034305%" rowspan="1" colspan="1">Error during the execution of the device tasks.</td>
</tr>
</tbody>
</table>
</div>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-chroma-subsampling"><a name="nvjpeg-chroma-subsampling" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-chroma-subsampling" name="nvjpeg-chroma-subsampling" shape="rect">2.3.6. nvjpeg-chroma-subsampling</a></h3>
<div class="body conbody">
<p class="p">One of the outputs of the <samp class="ph codeph">nvjpegGetImageInfo()</samp> API is
<samp class="ph codeph">nvjpegChromaSubsampling_t</samp>. This parameter is an
<samp class="ph codeph">enum</samp> type, and its enumerator list comprises of the chroma
subsampling property retrieved from the encoded JPEG image. Below are the chroma
subsampling types the <samp class="ph codeph">nvjpegGetImageInfo()</samp> function currently supports:
</p><pre xml:space="preserve">typedef enum
{
NVJPEG_CSS_444,
NVJPEG_CSS_422,
NVJPEG_CSS_420,
NVJPEG_CSS_440,
NVJPEG_CSS_411,
NVJPEG_CSS_410,
NVJPEG_CSS_GRAY,
NVJPEG_CSS_UNKNOWN
} nvjpegChromaSubsampling_t; </pre></div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-reference-docs"><a name="nvjpeg-reference-docs" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-reference-docs" name="nvjpeg-reference-docs" shape="rect">2.3.7. Reference Documents</a></h3>
<div class="body conbody">
<p class="p">Refer to the JPEG standard: <a class="xref" href="https://jpeg.org/jpeg/" target="_blank" shape="rect">https://jpeg.org/jpeg/ </a></p>
</div>
</div>
</div>
<div class="topic concept nested1" xml:lang="en-us" id="nvjpeg-examples"><a name="nvjpeg-examples" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-examples" name="nvjpeg-examples" shape="rect">Examples of nvJPEG</a></h3>
<div class="body conbody">
<p class="p">This package contains the library header and a set of libraries—static and shared. Shared
libraries (the <samp class="ph codeph">libnvjpeg.so</samp> and the respective versioned libraries)
have all of the CUDA toolkit dependencies statically linked. However, if you want to
link against the static library (<samp class="ph codeph">libnvjpeg_static.a</samp>) you also need to
link the other dependencies—for example dl, rt and thread libraries.
</p>
<p class="p"><strong class="ph b">Example of linking shared library:</strong></p>
<p class="p"><samp class="ph codeph">g++ -Icuda-linux64-nvjpeg/include -lnvjpeg -Lcuda-linux64-nvjpeg/lib64
my_example.cpp -o my_example </samp></p>
<p class="p"><strong class="ph b">Example of linking static library: </strong></p>
<p class="p"><samp class="ph codeph">g++ -Icuda-linux64-nvjpeg/include -lnvjpeg_static -ldl -lrt -pthread
-Lcuda-linux64-nvjpeg/lib64 my_example.cpp -o my_example </samp></p>
<p class="p"><strong class="ph b">Example</strong></p>
<p class="p">Below example shows how to use the various nvJPEG APIs. </p>
<p class="p">Compile with the following command from the <samp class="ph codeph">examples</samp> folder, assuming
CUDA 9.0 is installed at the path <samp class="ph codeph">/usr/local/cuda-9.0</samp>:
</p>
<p class="p"><samp class="ph codeph">g++ -O3 -m64 nvjpeg_example.cpp -I../include -lnvjpeg -L../lib64
-I/usr/local/cuda-9.0/include -ldl -lrt -pthread -lcudart
-L/usr/local/cuda-9.0/lib64 -Wl,-rpath=../lib64 -Wl,-rpath=/usr/local/cuda-9.0/lib64
-o nvjpeg_example</samp></p>
<p class="p">The below examples show how to decode the JPEG files using either single or batched API,
and write the decoded files as BMP images.
</p>
<p class="p"><strong class="ph b">To decode a single image:</strong></p>
<p class="p"><samp class="ph codeph">./nvjpeg_example -i /tmp/my_image.jpg -fmt rgb -o /tmp</samp></p>
<p class="p"><strong class="ph b">To decode multiple images in the folder using the batched API in separate phases:
</strong></p>
<p class="p"><samp class="ph codeph">./nvjpeg_example -i /tmp/my_images/ -fmt rgb -b 32 -pipelined -batched -o
/tmp</samp></p>
<p class="p"> Run the command <samp class="ph codeph">./nvjpeg_example -h</samp> for the description of the
parameters.
</p>
</div>
</div>
</div>
<div class="topic concept nested0" xml:lang="en-us" id="title-nvjpeg-encode"><a name="title-nvjpeg-encode" shape="rect">
<!-- --></a><h2 class="title topictitle1"><a href="#title-nvjpeg-encode" name="title-nvjpeg-encode" shape="rect">3. JPEG Encoding</a></h2>
<div class="body conbody">
<p class="p"> This section describes the encoding functions of the nvJPEG Library.</p>
</div>
<div class="topic concept nested1" xml:lang="en-US" id="using-encoder"><a name="using-encoder" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#using-encoder" name="using-encoder" shape="rect">3.1. Using the Encoder</a></h3>
<div class="body conbody">
<p class="p">The user should perform the below prerequisite steps before calling the nvJPEG encoding
functions. See also <a class="xref" href="index.html#nvjpeg-encoder-helper-api-reference" shape="rect">nvJPEG Encoder Helper API Reference</a>.
</p>
</div>
<div class="topic concept nested2" xml:lang="en-US" id="encoding-parameters"><a name="encoding-parameters" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#encoding-parameters" name="encoding-parameters" shape="rect">3.1.1. Encoding the Parameters</a></h3>
<div class="body conbody">
<p class="p">The user should create an encoding parameters structure with <samp class="ph codeph"><a class="xref" href="index.html#nvjpeg-encoder-params-create" shape="rect">nvjpegEncoderParamsCreate()</a></samp>
function. The function will be initialized with default parameters. User can use an
appropriate <samp class="ph codeph">nvjpegEncoderParamsSet*()</samp> function to set a specific parameter.
</p>
<p class="p">The quality parameter can be set, using <samp class="ph codeph"><a class="xref" href="index.html#nvjpeg-encoder-params-set-quality" shape="rect">nvjpegEncoderParamsSetQuality()</a></samp>
function, to an integer value between 1 and 100, and this quality parameter will be used as
a base for generating the JPEG quantization tables.
</p>
<p class="p">The parameters structure should be passed to compression functions. </p>
<div class="p">
<div class="note note"><span class="notetitle">Note:</span> The encoding parameters structure can be reused to compress multiple images
simultaneously, but no changes to the parameters should be made during the ongoing
encoding, or encoding result will be undefined.
</div>
</div>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-US" id="encoding-state"><a name="encoding-state" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#encoding-state" name="encoding-state" shape="rect">3.1.2. Encoding the State</a></h3>
<div class="body conbody">
<p class="p">The user should create the encoding state structure using
<samp class="ph codeph"><a class="xref" href="index.html#nvjpeg-encoder-state-create" shape="rect">nvjpegEncoderStateCreate()</a></samp> function. This function will hold intermediate
buffers for the encoding process. This state should be passed to the compression functions.
</p>
<div class="p">
<div class="note note"><span class="notetitle">Note:</span> The encoding state structure can be reused to encode a series of images, but no
encoding should be performed on multiple images with the same encoding state at the same
time - otherwise result of the encodings will be undefined.
</div>
</div>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-US" id="encoding-the-image"><a name="encoding-the-image" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#encoding-the-image" name="encoding-the-image" shape="rect">3.1.3. Encoding the Image</a></h3>
<div class="body conbody">
<p class="p">The nvJPEG library provides a few interfaces for compressing the image in different formats
and colorspaces. See below.
</p>
</div>
<div class="topic concept nested3" xml:lang="en-US" id="using-nvjpegEncodeYUV"><a name="using-nvjpegEncodeYUV" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#using-nvjpegEncodeYUV" name="using-nvjpegEncodeYUV" shape="rect">3.1.3.1. nvjpegEncodeYUV</a></h3>
<div class="body conbody">
<p class="p">Input for this function is an image in YUV colorspace. See <a class="xref" href="index.html#nvjpeg-encode-yuv" shape="rect">nvjpegEncodeYUV()</a>.
The <samp class="ph codeph">source</samp> argument should be filled with the corresponding YUV planar
data. The <samp class="ph codeph">chroma_subsampling</samp> argument should have the chroma subsampling of
the input data. If the chroma subsampling in the encoding parameters is the same as input
chroma subsampling, then the user's input data will be directly used in the JPEG
compression. Otherwise chroma will be resampled to match the chroma subsampling of the
encoding parameters.
</p>
<p class="p">Input data should be provided with respect to the subsampling factors. That is, the
chrominance image planes should have sizes aligned to the corresponding subsamplings. For
example:
</p><a name="using-nvjpegEncodeYUV__ul_b1t_wyw_lgb" shape="rect">
<!-- --></a><ul class="ul" id="using-nvjpegEncodeYUV__ul_b1t_wyw_lgb">
<li class="li">Image dimensions: 123x321</li>
<li class="li">Input chroma subsampling is: NVJPEG_CSS_410</li>
<li class="li">Chroma subsampling factor for this chroma subsampling: is 4x2</li>
<li class="li">Given the above, the encoder library expects the user to provide:<a name="using-nvjpegEncodeYUV__ul_c1t_wyw_lgb" shape="rect">
<!-- --></a><ul class="ul" id="using-nvjpegEncodeYUV__ul_c1t_wyw_lgb">
<li class="li">Y plane with size: 123 x 321</li>
<li class="li">Cb and Cr plane with size: 31 x 161</li>
</ul>
</li>
</ul>
</div>
</div>
<div class="topic concept nested3" xml:lang="en-US" id="using-nvjpegEncodeImage"><a name="using-nvjpegEncodeImage" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#using-nvjpegEncodeImage" name="using-nvjpegEncodeImage" shape="rect">3.1.3.2. nvjpegEncodeImage</a></h3>
<div class="body conbody">
<p class="p">See <a class="xref" href="index.html#nvjpeg-encode-image" shape="rect">nvjpegEncodeImage()</a>. Input for this function, i.e., how data should be
provided in the <samp class="ph codeph">source</samp> argument, is determined by the
<samp class="ph codeph">input_format</samp> argument. For the interleaved formats (ending with <strong class="ph b">I</strong>)
only the first channel is used. For the non-interleaved formats, all the channels in the
input format are used.
</p>
<p class="p">For example, if the user has interleaved the RGB image of size <samp class="ph codeph">W x H</samp>,
stored continuously, and the pointer to it is <samp class="ph codeph">pImage</samp>, then
<samp class="ph codeph">source</samp> should be:
</p><a name="using-nvjpegEncodeImage__ul_e5p_g1x_lgb" shape="rect">
<!-- --></a><ul class="ul" id="using-nvjpegEncodeImage__ul_e5p_g1x_lgb">
<li class="li"><samp class="ph codeph">source.channel[0] = pImage</samp></li>
<li class="li"><samp class="ph codeph">source.pitch[0] = W*3</samp></li>
</ul>
<p class="p">When the same the image is stored in planar format, with image planes pointers stored
continuously in the array <samp class="ph codeph">pImage[3]</samp>, then <samp class="ph codeph">source</samp> should
be:
</p><a name="using-nvjpegEncodeImage__ul_f5p_g1x_lgb" shape="rect">
<!-- --></a><ul class="ul" id="using-nvjpegEncodeImage__ul_f5p_g1x_lgb">
<li class="li"><samp class="ph codeph">source.channel[0] = pImage[0]</samp></li>
<li class="li"><samp class="ph codeph">source.channel[1] = pImage[1]</samp></li>
<li class="li"><samp class="ph codeph">source.channel[2] = pImage[2]</samp></li>
</ul>
<p class="p">The <samp class="ph codeph">pitch</samp> values for each channel in the <samp class="ph codeph">source</samp> parameter
should be set accordingly to the data layout.
</p>
<p class="p">The nvJPEG library will perform the color transformation to the YCbCr, and will compress
the result.
</p>
</div>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-US" id="retrieving-compressed-stream"><a name="retrieving-compressed-stream" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#retrieving-compressed-stream" name="retrieving-compressed-stream" shape="rect">3.1.4. Retrieving the Compressed Stream </a></h3>
<div class="body conbody">
<p class="p">Often it is not feasible to accurately predict the final compressed data size of the final
JPEG stream for any input data and parameters. The nvJPEG library, while encoding, will
calculate the size of the final stream, allocate temporary buffer in the encoder state and
save the compressed data in the encoding state's buffer. In order to get final compressed
JPEG stream, the user should provide the memory buffer large enough to store this compressed
data. There are two options for how to do this:
</p><a name="retrieving-compressed-stream__ol_slv_tbx_lgb" shape="rect">
<!-- --></a><ol class="ol" id="retrieving-compressed-stream__ol_slv_tbx_lgb">
<li class="li">Use the upper bound on compressed JPEG stream size for the given parameters and image
dimensions:<a name="retrieving-compressed-stream__ol_tlv_tbx_lgb" shape="rect">
<!-- --></a><ol class="ol" type="a" id="retrieving-compressed-stream__ol_tlv_tbx_lgb">
<li class="li">Use the <samp class="ph codeph"><a class="xref" href="index.html#nvjpeg-encode-retrieve-bitstream" shape="rect">nvjpegEncodeRetrieveBitstream()</a></samp> function to retrieve the
maximum possible JPEG stream size at any given time.
</li>
<li class="li">Allocate the memory buffer at any given time. </li>
<li class="li">Encode the image using one of the encoding functions.</li>
<li class="li">Retrieve the compressed JPEG stream from the encoder state after successful
encoding, using the <samp class="ph codeph">nvjpegEncodeRetrieveBitstream()</samp> and the
allocated buffer.
</li>
</ol>
</li>
<li class="li">Wait for the encoding to complete, and retrieve the exact size of required buffer, as
below:<a name="retrieving-compressed-stream__ol_ulv_tbx_lgb" shape="rect">
<!-- --></a><ol class="ol" type="a" id="retrieving-compressed-stream__ol_ulv_tbx_lgb">
<li class="li">Encode the image using one of the encoding functions.</li>
<li class="li">Use the <samp class="ph codeph">nvjpegEncodeRetrieveBitstream()</samp> function to retrieve the
size in bytes of the compressed JPEG stream.
</li>
<li class="li">Allocate the memory buffer of at least this size.</li>
<li class="li">Use the <samp class="ph codeph">nvjpegEncodeRetrieveBitstream()</samp> function to populate your
buffer with the compressed JPEG stream.
</li>
</ol>
</li>
</ol>
<div class="p">
<div class="note note"><span class="notetitle">Note:</span> As the same encoding image state can be reused to compress a series of images, the
<samp class="ph codeph">nvjpegEncodeRetrieveBitstream()</samp> function will return the result for the
last compressed image.
</div>
</div>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-US" id="nvjpeg-encode-examples"><a name="nvjpeg-encode-examples" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-encode-examples" name="nvjpeg-encode-examples" shape="rect">3.1.5. JPEG Encoding Example</a></h3>
<div class="body conbody">
<p class="p">See below the example code, and the block diagram shown in <a class="xref" href="index.html#nvjpeg-encode-examples__fig-nvjpeg-encode-example" shape="rect">Figure 1</a>, for encoding with nvJPEG Encoder.
</p>
<div class="fig fignone" id="nvjpeg-encode-examples__fig-nvjpeg-encode-example"><a name="nvjpeg-encode-examples__fig-nvjpeg-encode-example" shape="rect">
<!-- --></a><span class="figcap">Figure 1. JPEG Encoding Using nvJPEG Encoder</span><br clear="none"></br><div class="imagecenter"><img class="image imagecenter" width="800" src="graphics/nvjpeg-encoding-flow.png" alt="nvJPEG Encoder Example"></img></div><br clear="none"></br></div><pre xml:space="preserve">nvjpegHandle_t nv_handle;
nvjpegEncoderState_t nv_enc_state;
nvjpegEncoderParams_t nv_enc_params;
cudaStream_t stream;
<span xmlns:xslthl="http://xslthl.sf.net" class="xslthl-comment">// initialize nvjpeg structures</span>
nvjpegCreateSimple(&nv_handle);
nvjpegEncoderStateCreate(nv_handle, &nv_enc_state, stream);
nvjpegEncoderParamsCreate(nv_handle, &nv_enc_params, stream);
nvjpegImage_t nv_image;
<span xmlns:xslthl="http://xslthl.sf.net" class="xslthl-comment">// Fill nv_image with image data, let’s say 640x480 image in RGB format</span>
<span xmlns:xslthl="http://xslthl.sf.net" class="xslthl-comment">// Compress image</span>
nvjpegEncodeImage(nv_handle, nv_enc_state, nv_enc_params,
&nv_image, NVJPEG_INPUT_RGB, 640, 480, stream);
<span xmlns:xslthl="http://xslthl.sf.net" class="xslthl-comment">// get compressed stream size</span>
size_t length;
nvjpegEncodeRetrieveBitstream(nv_handle, nv_enc_state, NULL, &length, stream);
<span xmlns:xslthl="http://xslthl.sf.net" class="xslthl-comment">// get stream itself</span>
cudaStreamSynchronize(stream);
std::vector<<span xmlns:xslthl="http://xslthl.sf.net" class="xslthl-keyword">char</span>> jpeg(length);
nvjpegEncodeRetrieveBitstream(nv_handle, nv_enc_state, jpeg.data(), &length, 0);
<span xmlns:xslthl="http://xslthl.sf.net" class="xslthl-comment">// write stream to file</span>
cudaStreamSynchronize(stream);
std::ofstream output_file(“test.jpg”, std::ios::out | std::ios::binary);
output_file.write(jpeg.data(), length);
output_file.close();</pre></div>
</div>
</div>
<div class="topic concept nested1" xml:lang="en-us" id="nvjpeg-encoder-type-declarations"><a name="nvjpeg-encoder-type-declarations" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-encoder-type-declarations" name="nvjpeg-encoder-type-declarations" shape="rect">3.2. nvJPEG Encoder Type Declarations</a></h3>
<div class="body conbody">
<p class="p">This section describes the nvJPEG Encoder Type Declarations.</p>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-inputformat"><a name="nvjpeg-inputformat" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-inputformat" name="nvjpeg-inputformat" shape="rect">nvjpegInputFormat_t</a></h3>
<div class="body conbody">
<div class="p"><pre xml:space="preserve">typedef enum
{
NVJPEG_INPUT_RGB = 3,
NVJPEG_INPUT_BGR = 4,
NVJPEG_INPUT_RGBI = 5,
NVJPEG_INPUT_BGRI = 6
} nvjpegInputFormat_t;</pre></div>
<p class="p">The <samp class="ph codeph">nvjpegInputFormat_t</samp> enum is used to select the color model and pixel
format of the input image. It is used for conversion to YCbCr during encoding.
</p>
<div class="p">
<div class="tablenoborder"><a name="nvjpeg-inputformat__table_ogq_y1x_bgb" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-inputformat__table_ogq_y1x_bgb" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" valign="top" width="33.557046979865774%" rowspan="1" colspan="1"><strong class="ph b">Member</strong></td>
<td class="entry" valign="top" width="66.44295302013423%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="33.557046979865774%" rowspan="1" colspan="1">NVJPEG_INPUT_RGB</td>
<td class="entry" valign="top" width="66.44295302013423%" rowspan="1" colspan="1">Input image is in RGB color model. Pixel format is RGB.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="33.557046979865774%" rowspan="1" colspan="1">NVJPEG_INPUT_BGR</td>
<td class="entry" valign="top" width="66.44295302013423%" rowspan="1" colspan="1">Input image is in RGB color model. Pixel format is BGR.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="33.557046979865774%" rowspan="1" colspan="1">NVJPEG_INPUT_RGBI</td>
<td class="entry" valign="top" width="66.44295302013423%" rowspan="1" colspan="1">Input image is in RGB color model. Pixel format is interleaved RGB.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="33.557046979865774%" rowspan="1" colspan="1">NVJPEG_INPUT_BGRI</td>
<td class="entry" valign="top" width="66.44295302013423%" rowspan="1" colspan="1">Input image is in RGB color model. Pixel format is interleaved BGR.</td>
</tr>
</tbody>
</table>
</div>
</div>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-encoder-state"><a name="nvjpeg-encoder-state" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-encoder-state" name="nvjpeg-encoder-state" shape="rect">3.2.2. nvjpegEncoderState_t</a></h3>
<div class="body conbody">
<p class="p">The <samp class="ph codeph">nvjpegEncoderState_t</samp> is a structure that stores intermdiate buffers and
variables used for compression.
</p>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-encoder-params"><a name="nvjpeg-encoder-params" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-encoder-params" name="nvjpeg-encoder-params" shape="rect">3.2.3. nvjpegEncoderParams_t</a></h3>
<div class="body conbody">
<p class="p">The <samp class="ph codeph">nvjpegEncoderParams_t</samp> is a structure that stores JPEG encode
parameters.
</p>
</div>
</div>
</div>
<div class="topic concept nested1" xml:lang="en-us" id="nvjpeg-encoder-helper-api-reference"><a name="nvjpeg-encoder-helper-api-reference" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-encoder-helper-api-reference" name="nvjpeg-encoder-helper-api-reference" shape="rect">3.3. nvJPEG Encoder Helper API Reference</a></h3>
<div class="body conbody">
<p class="p">The nvJPEG Encoder helper functions are used for initializing. </p>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-encoder-state-create"><a name="nvjpeg-encoder-state-create" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-encoder-state-create" name="nvjpeg-encoder-state-create" shape="rect">3.3.1. nvjpegEncoderStateCreate()</a></h3>
<div class="body conbody">
<p class="p">Creates encoder state that stores intermediate buffers used in compression.</p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">nvjpegStatus_t nvjpegEncoderStateCreate(
nvjpegHandle_t handle,
nvjpegEncoderState_t *encoder_state,
cudaStream_t stream);</pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-encoder-state-create__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-encoder-state-create__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">handle</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">Library handle</p>
</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">encoder_state</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Output</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">Pointer to the encoder state structure, where the new state will be
placed.
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">stream</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Inputt</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">CUDA stream where all the required device operations will be placed.</p>
</td>
</tr>
</tbody>
</table>
</div>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-encoder-state-destroy"><a name="nvjpeg-encoder-state-destroy" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-encoder-state-destroy" name="nvjpeg-encoder-state-destroy" shape="rect">3.3.2. nvjpegEncoderStateDestroy()</a></h3>
<div class="body conbody">
<p class="p">Destroys the encoder state.</p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">nvjpegStatus_t nvjpegEncoderStateDestroy(
nvjpegEncoderState_t encoder_state);</pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-encoder-state-destroy__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-encoder-state-destroy__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">encoder_state</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">Encoder state structure that will be released.</td>
</tr>
</tbody>
</table>
</div>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-encoder-params-create"><a name="nvjpeg-encoder-params-create" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-encoder-params-create" name="nvjpeg-encoder-params-create" shape="rect">3.3.3. nvjpegEncoderParamsCreate()</a></h3>
<div class="body conbody">
<p class="p">Creates the structure that holds the compression parameters.</p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">nvjpegStatus_t nvjpegEncoderParamsCreate(
nvjpegHandle_t handle,
nvjpegEncoderParams_t *encoder_params,
cudaStream_t stream);</pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-encoder-params-create__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-encoder-params-create__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">handle</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">Library handle</p>
</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">encoder_params</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Output</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">Pointer to the location where the new parameters structure will be placed.</p>
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">stream</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Inputt</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">CUDA stream where all the required device operations will be placed.</p>
</td>
</tr>
</tbody>
</table>
</div>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-encoder-params-destroy"><a name="nvjpeg-encoder-params-destroy" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-encoder-params-destroy" name="nvjpeg-encoder-params-destroy" shape="rect">3.3.4. nvjpegEncoderParamsDestroy()</a></h3>
<div class="body conbody">
<p class="p">Destroys the encoder parameters structure.</p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">nvjpegEncoderParamsDestroy(
nvjpegEncoderParams_t encoder_params);</pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-encoder-params-destroy__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-encoder-params-destroy__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">encoder_params</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">Encoder params structure that will be released.</p>
</td>
</tr>
</tbody>
</table>
</div>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-encoder-params-set-quality"><a name="nvjpeg-encoder-params-set-quality" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-encoder-params-set-quality" name="nvjpeg-encoder-params-set-quality" shape="rect">3.3.5. nvjpegEncoderParamsSetQuality()</a></h3>
<div class="body conbody">
<p class="p">Sets the parameter quality in the encoder parameters structure.</p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">nvjpegStatus_t nvjpegEncoderParamsSetQuality(
nvjpegEncoderParams_t encoder_params,
const int quality,
cudaStream_t stream);</pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-encoder-params-set-quality__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-encoder-params-set-quality__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">encoder_params</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">Encoder params structure handle.</p>
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">quality</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">Integer value of quality between 1 and 100, where 100 is the highest quality.
Default value is 70.
</p>
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">stream</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Inputt</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">CUDA stream where all the required device operations will be placed.</p>
</td>
</tr>
</tbody>
</table>
</div>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-encoder-params-set-optimized-huffman"><a name="nvjpeg-encoder-params-set-optimized-huffman" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-encoder-params-set-optimized-huffman" name="nvjpeg-encoder-params-set-optimized-huffman" shape="rect">3.3.6. nvjpegEncoderParamsSetOptimizedHuffman()</a></h3>
<div class="body conbody">
<p class="p">Sets whether or not to use optimized Huffman. Using optimized Huffman produces smaller JPEG
bitstream sizes with the same quality, but with slower performance.
</p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">nvjpegStatus_t nvjpegEncoderParamsSetOptimizedHuffman(
nvjpegEncoderParams_t encoder_params,
const int optimized,
cudaStream_t stream);</pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-encoder-params-set-optimized-huffman__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-encoder-params-set-optimized-huffman__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">encoder_params</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">Encoder params structure handle.</p>
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">optimized</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">If this value is 0 then non-optimized Huffman will be used. Otherwise optimized version
will be used. Default value is 0.
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">stream</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Inputt</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">CUDA stream where all the required device operations will be placed.</p>
</td>
</tr>
</tbody>
</table>
</div>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-encoder-params-set-sampling-factors"><a name="nvjpeg-encoder-params-set-sampling-factors" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-encoder-params-set-sampling-factors" name="nvjpeg-encoder-params-set-sampling-factors" shape="rect">3.3.7. nvjpegEncoderParamsSetSamplingFactors()</a></h3>
<div class="body conbody">
<p class="p">Sets which chroma subsampling will be used for JPEG compression. </p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">nvjpegStatus_t nvjpegEncoderParamsSetSamplingFactors(
nvjpegEncoderParams_t encoder_params,
const nvjpegChromaSubsampling_t chroma_subsampling,
cudaStream_t stream);</pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-encoder-params-set-sampling-factors__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-encoder-params-set-sampling-factors__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">encoder_params</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">Encoder params structure handle.</p>
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">chroma_subsampling</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">Chroma subsampling that will be used for JPEG compression. If the input is in YUV
color model and <samp class="ph codeph">chroma_subsampling</samp> is different from the
subsampling factors of source image, then the NVJPEG library will convert
subsampling to the value of <samp class="ph codeph">chroma_subsampling</samp>. Default value is
4:4:4.
</p>
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">stream</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Inputt</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">CUDA stream where all the required device operations will be placed.</p>
</td>
</tr>
</tbody>
</table>
</div>
</div>
</div>
</div>
<div class="topic concept nested1" xml:lang="en-us" id="nvjpeg-encoder-api-reference"><a name="nvjpeg-encoder-api-reference" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-encoder-api-reference" name="nvjpeg-encoder-api-reference" shape="rect">3.4. nvJPEG Encoder API Reference</a></h3>
<div class="body conbody">
<p class="p">This section describes the nvJPEG Encoder API. </p>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-encoder-get-buffer-size"><a name="nvjpeg-encoder-get-buffer-size" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-encoder-get-buffer-size" name="nvjpeg-encoder-get-buffer-size" shape="rect">nvjpegEncodeGetBufferSize()</a></h3>
<div class="body conbody">
<p class="p">Returns the maximum possible buffer size that is needed to store the compressed JPEG stream,
for the given input parameters.
</p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">nvjpegStatus_t nvjpegEncodeGetBufferSize(
nvjpegHandle_t handle,
const nvjpegEncoderParams_t encoder_params,
int image_width,
int image_height,
size_t *max_stream_length);</pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-encoder-get-buffer-size__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-encoder-get-buffer-size__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">handle</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">Library handle</p>
</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">encoder_params</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">Encoder parameters structure handle.</p>
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">image_width</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">Input image width.</p>
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1">image_height</td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">Input image height.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">stream</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Inputt</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">CUDA stream where all the required device operations will be placed.</p>
</td>
</tr>
</tbody>
</table>
</div>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-encode-yuv"><a name="nvjpeg-encode-yuv" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-encode-yuv" name="nvjpeg-encode-yuv" shape="rect">3.4.2. nvjpegEncodeYUV()</a></h3>
<div class="body conbody">
<p class="p">Compresses the image in YUV colorspace to JPEG stream using the provided parameters, and
stores it in the state structure.
</p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">nvjpegStatus_t nvjpegEncodeYUV(
nvjpegHandle_t handle,
nvjpegEncoderState_t encoder_state,
const nvjpegEncoderParams_t encoder_params,
const nvjpegImage_t *source,
nvjpegChromaSubsampling_t chroma_subsampling,
int image_width,
int image_height,
cudaStream_t stream);</pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-encode-yuv__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-encode-yuv__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">handle</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">Library handle</p>
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">encoder_state</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">Internal structure that holds the temporary buffers required for the compression
and also stores the final compressed JPEG stream.
</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">encoder_params</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">Encoder parameters structure handle.</p>
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">source</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">Pointer to the <samp class="ph codeph">nvjpeg</samp> structure that holds the device pointers
to the <samp class="ph codeph">Y, U(Cb) and V(Cr)</samp> image planes and the respective
strides.
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">chroma_subsampling</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">Chroma subsampling of the input data.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">image_width</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">Input image width.</p>
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1">image_height</td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">Input image height.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">stream</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Inputt</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">CUDA stream where all the required device operations will be placed.</p>
</td>
</tr>
</tbody>
</table>
</div>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-encode-image"><a name="nvjpeg-encode-image" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-encode-image" name="nvjpeg-encode-image" shape="rect">3.4.3. nvjpegEncodeImage()</a></h3>
<div class="body conbody">
<p class="p">Compresses the image in the provided format to the JPEG stream using the provided parameters,
and stores it in the state structure.
</p>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">nvjpegStatus_t nvjpegEncodeImage(
nvjpegHandle_t handle,
nvjpegEncoderState_t encoder_state,
const nvjpegEncoderParams_t encoder_params,
const nvjpegImage_t *source,
nvjpegInputFormat_t input_format,
int image_width,
int image_height,
cudaStream_t stream);</pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-encode-image__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-encode-image__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">handle</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">Library handle</p>
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">encoder_state</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">Internal structure that holds the temporary buffers required for the compression
and also stores the final compressed JPEG stream.
</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">encoder_params</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">Encoder parameters structure handle.</p>
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">source</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">Pointer to the <samp class="ph codeph">nvjpeg</samp> structure that holds the device pointers
to the <samp class="ph codeph">Y, U(Cb) and V(Cr)</samp> image planes and the respective
strides.
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">input_format</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">Value of <samp class="ph codeph">nvjpegInputFormat_t</samp> type that describes the input
data.
</p>
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">image_width</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">Input image width.</p>
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1">image_height</td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">Input image height.</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">stream</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">CUDA stream where all the required device operations will be placed.</p>
</td>
</tr>
</tbody>
</table>
</div>
</div>
</div>
<div class="topic concept nested2" xml:lang="en-us" id="nvjpeg-encode-retrieve-bitstream"><a name="nvjpeg-encode-retrieve-bitstream" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#nvjpeg-encode-retrieve-bitstream" name="nvjpeg-encode-retrieve-bitstream" shape="rect">3.4.4. nvjpegEncodeRetrieveBitstream()</a></h3>
<div class="body conbody">
<p class="p">Retrieves the compressed stream from the encoder state that was previously used in one of the
encoder functions.
</p>
<div class="p"><a name="nvjpeg-encode-retrieve-bitstream__ul_dp5_2mf_kgb" shape="rect">
<!-- --></a><ul class="ul" id="nvjpeg-encode-retrieve-bitstream__ul_dp5_2mf_kgb">
<li class="li">If <samp class="ph codeph">data</samp> parameter is NULL then the encoder will return compressed
stream size in the <samp class="ph codeph">length</samp> parameter.
</li>
<li class="li">If <samp class="ph codeph">data</samp> is not NULL then the provided <samp class="ph codeph">length</samp> parameter
should contain the <samp class="ph codeph">data</samp> buffer size.
</li>
<li class="li">If the provided <samp class="ph codeph">length</samp> is less than compressed stream size, then an
error will be returned. Otherwise the compressed stream will be stored in the
<samp class="ph codeph">data</samp> buffer and the actual compressed buffer size will be stored in the
<samp class="ph codeph">length</samp> parameter.
</li>
</ul>
</div>
<p dir="ltr" class="p"><strong class="ph b">Signature:</strong></p><pre xml:space="preserve">nvjpegStatus_t nvjpegEncodeRetrieveBitstream(
nvjpegHandle_t handle,
nvjpegEncoderState_t encoder_state,
unsigned char *data,
size_t *length,
cudaStream_t stream);</pre><p dir="ltr" class="p"><strong class="ph b">Parameters:</strong></p>
<div class="tablenoborder"><a name="nvjpeg-encode-retrieve-bitstream__table_qc3_kqt_d2b" shape="rect">
<!-- --></a><table cellpadding="4" cellspacing="0" summary="" id="nvjpeg-encode-retrieve-bitstream__table_qc3_kqt_d2b" class="table" frame="border" border="1" rules="all">
<tbody class="tbody">
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><strong class="ph b">Parameter</strong></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1"><strong class="ph b">Input / Output</strong></td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1"><strong class="ph b">Memory</strong></td>
<td class="entry" dir="ltr" valign="top" width="34.75609756097561%" rowspan="1" colspan="1"><strong class="ph b">Description</strong></td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">handle</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">Library handle</p>
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">encoder_state</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">The <samp class="ph codeph">encoder_state</samp> that was previously used in one of the encoder
functions.
</td>
</tr>
<tr class="row">
<td class="entry" dir="ltr" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">data</samp></td>
<td class="entry" dir="ltr" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" dir="ltr" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">Pointer to the buffer in the host memory where the compressed stream will be stored.
Can be NULL (see description).
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">length</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Input/Output</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">Pointer to the input buffer size. On return the NVJPEG library will store the
actual compressed stream size in this parameter.
</td>
</tr>
<tr class="row">
<td class="entry" valign="top" width="32.774390243902445%" rowspan="1" colspan="1"><samp class="ph codeph">stream</samp></td>
<td class="entry" valign="top" width="17.225609756097562%" rowspan="1" colspan="1">Inputt</td>
<td class="entry" valign="top" width="15.243902439024392%" rowspan="1" colspan="1">Host</td>
<td class="entry" valign="top" width="34.75609756097561%" rowspan="1" colspan="1">
<p class="p">CUDA stream where all the required device operations will be placed.</p>
</td>
</tr>
</tbody>
</table>
</div>
</div>
</div>
</div>
</div>
<div class="topic concept nested0" id="notices-header"><a name="notices-header" shape="rect">
<!-- --></a><h2 class="title topictitle1"><a href="#notices-header" name="notices-header" shape="rect">Notices</a></h2>
<div class="topic reference nested1" id="notice"><a name="notice" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#notice" name="notice" shape="rect"></a></h3>
<div class="body refbody">
<div class="section">
<h3 class="title sectiontitle">Notice</h3>
<p class="p">ALL NVIDIA DESIGN SPECIFICATIONS, REFERENCE BOARDS, FILES, DRAWINGS, DIAGNOSTICS, LISTS, AND OTHER DOCUMENTS (TOGETHER AND
SEPARATELY, "MATERIALS") ARE BEING PROVIDED "AS IS." NVIDIA MAKES NO WARRANTIES, EXPRESSED, IMPLIED, STATUTORY, OR OTHERWISE
WITH RESPECT TO THE MATERIALS, AND EXPRESSLY DISCLAIMS ALL IMPLIED WARRANTIES OF NONINFRINGEMENT, MERCHANTABILITY, AND FITNESS
FOR A PARTICULAR PURPOSE.
</p>
<p class="p">Information furnished is believed to be accurate and reliable. However, NVIDIA Corporation assumes no responsibility for the
consequences of use of such information or for any infringement of patents or other rights of third parties that may result
from its use. No license is granted by implication of otherwise under any patent rights of NVIDIA Corporation. Specifications
mentioned in this publication are subject to change without notice. This publication supersedes and replaces all other information
previously supplied. NVIDIA Corporation products are not authorized as critical components in life support devices or systems
without express written approval of NVIDIA Corporation.
</p>
</div>
</div>
</div>
<div class="topic reference nested1" id="trademarks"><a name="trademarks" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#trademarks" name="trademarks" shape="rect"></a></h3>
<div class="body refbody">
<div class="section">
<h3 class="title sectiontitle">Trademarks</h3>
<p class="p">NVIDIA and the NVIDIA logo are trademarks or registered trademarks of NVIDIA Corporation
in the U.S. and other countries. Other company and product names may be trademarks of
the respective companies with which they are associated.
</p>
</div>
</div>
</div>
<div class="topic reference nested1" id="copyright-present"><a name="copyright-present" shape="rect">
<!-- --></a><h3 class="title topictitle2"><a href="#copyright-present" name="copyright-present" shape="rect"></a></h3>
<div class="body refbody">
<div class="section">
<h3 class="title sectiontitle">Copyright</h3>
<p class="p">© <span class="ph">2019</span> NVIDIA Corporation. All rights
reserved.
</p>
<p class="p">This product includes software developed by the Syncro Soft SRL (http://www.sync.ro/).</p>
</div>
</div>
</div>
</div>
<hr id="contents-end"></hr>
</article>
</div>
</div>
<script language="JavaScript" type="text/javascript" charset="utf-8" src="../common/formatting/common.min.js"></script>
<script language="JavaScript" type="text/javascript" charset="utf-8" src="../common/scripts/google-analytics/google-analytics-write.js"></script>
<script language="JavaScript" type="text/javascript" charset="utf-8" src="../common/scripts/google-analytics/google-analytics-tracker.js"></script>
<script type="text/javascript">var switchTo5x=true;</script><script type="text/javascript" src="http://w.sharethis.com/button/buttons.js"></script><script type="text/javascript">stLight.options({publisher: "998dc202-a267-4d8e-bce9-14debadb8d92", doNotHash: false, doNotCopy: false, hashAddressBar: false});</script><script type="text/javascript">_satellite.pageBottom();</script></body>
</html>
distrib
>
Mageia
>
7
>
x86_64
>
media
>
nonfree-updates
>
by-pkgid
>
b86a85131cc739c1c53d0b55840a4328
>
files
>
3626
