Skip to content

Ultra HDR is a true HDR image format, and is backcompatible. libultrahdr is the reference codec for the Ultra HDR format. The codecs that support the format can render the HDR intent of the image on HDR displays; other codecs can still decode and display the SDR intent of the image.

License

Notifications You must be signed in to change notification settings

DichenZhang1/libultrahdr_dev

 
 

Repository files navigation

Background

libultrahdr is an image compression library that uses gain map technology to store and distribute HDR images. Conceptually on the encoding side, the library accepts SDR and HDR rendition of an image and from these a Gain Map (quotient between the two renditions) is computed. The library then uses backward compatible means to store the base image (SDR), gain map image and some associated metadata. Legacy readers that do not support parsing the gain map image and/or metadata, will display the base image. Readers that support the format combine the base image with the gain map and render a high dynamic range image on compatible displays.

More information about libultrahdr can be found at https://developer.android.com/guide/topics/media/platform/hdr-image-format.

Building libultrahdr

libultrahdr compresses base image and gain map image in to jpeg format. For this libjpeg-turbo is used. This is cloned from https://github.com/libjpeg-turbo/libjpeg-turbo.git and included in the build process.

Requirements

  • CMake v3.13 or later

  • NASM or Yasm (If libjpeg-turbo needs to be built with SIMD extensions)

    • If using NASM, 2.13 or later is required.
    • If using Yasm, 1.2.0 or later is required.
    • If building on macOS, NASM or Yasm can be obtained from MacPorts or Homebrew.
  • Compilers with support for C++17

Should work with GCC v7 (or later) and Clang 5 (or later) on Linux and Mac Platforms.

Should work with Microsoft Visual C++ 2019 (or later) on Windows Platforms.

Build Procedure

To build libultrahdr, examples, unit tests:

Un*x (including Linux, Mac)

mkdir {build_directory}
cd {build_directory}
cmake -G "Unix Makefiles"  -DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++ -DUHDR_BUILD_TESTS=1 ../
make
ctest

This will generate the following files under {build_directory}:

libultrahdr.a
Static link library for the ultrahdr API

ultrahdr_app
Sample application demonstrating ultrahdr API

ultrahdr_unit_test
Unit tests

Visual C++ (IDE)

mkdir {build_directory}
cd {build_directory}
cmake -G "Visual Studio 16 2019" -DUHDR_BUILD_TESTS=1 ../
cmake --build ./ --config=Release
ctest -C Release

This will generate the following files under {build_directory/Release}:

ultrahdr.lib
Static link library for the ultrahdr API

ultrahdr_app.exe
Sample application demonstrating ultrahdr API

ultrahdr_unit_test.exe
Unit tests

Visual C++ (Command line)

mkdir {build_directory}
cd {build_directory}
cmake -G "NMake Makefiles" -DUHDR_BUILD_TESTS=1 ../
cmake --build ./
ctest

This will generate the following files under {build_directory}:

ultrahdr.lib
Static link library for the ultrahdr API

ultrahdr_app.exe
Sample application demonstrating ultrahdr API

ultrahdr_unit_test.exe
Unit tests

MinGW

NOTE: This assumes that you are building on a Windows machine using the MSYS environment.

mkdir {build_directory}
cd {build_directory}
cmake -G "MSYS Makefiles" -DUHDR_BUILD_TESTS=1 ../
cmake --build ./
ctest

mkdir {build_directory}
cd {build_directory}
cmake -G "MinGW Makefiles" -DUHDR_BUILD_TESTS=1 ../
cmake --build ./
ctest

This will generate the following files under {build_directory}:

libultrahdr.a
Static link library for the ultrahdr API

ultrahdr_app.exe
Sample application demonstrating ultrahdr API

ultrahdr_unit_test.exe
Unit tests

NOTE: To not build unit tests, skip passing -DUHDR_BUILD_TESTS=1

Building Benchmark

To build benchmarks, pass -DUHDR_BUILD_BENCHMARK=1 to cmake configure command and build.

This will additionally generate,

ultrahdr_bm
Benchmark tests

Building Fuzzers

Refer to README.md for complete instructions.

Using libultrahdr

libultrahdr includes two classes of APIs, one to compress and the other to decompress HDR images:

List of encode APIs:

Input HDR YUV SDR YUV JPEG Encoded gainmap Quality (0 ~ 100) EXIF Use case
API-0 P010 No No No Required Optional Experimental only.
API-1 P010 YUV_420 No No Required Optional Raw SDR input. Primary image will be encoded from the raw SDR input in the library.
API-2 P010 YUV_420 Yes No No No Both JPEG and raw SDR inputs. Gainmap will be calculated from raw HDR and raw SDR inputs, the JPEG input will be preserved (including metadata) as the primary image.
API-3 P010 No Yes No No No SDR JPEG input. Gainmap will be calculated from raw HDR and the decoding result of the JPEG input, the JPEG input will be preserved (including metadata) as the primary image.
API-4 No No Yes Yes No No SDR JPEG and gainmap inputs. The library will only generate the Ultra HDR related metadata and write everything into the Ultra HDR format, all other metadata from the JPEG input will be preserved.

List of decode API:

Input Usage
compressed_jpegr_image The input data. Pointer to JPEG/R stream.
dest The output data. Destination that decoded data to be written.
max_display_boost (optional, >= 1.0) the maximum available boost supported by a display.
exif (optional, default to NULL) Destination that exif data to be written.
gain_map (optional, default to NULL) Destination that decoded gain map data to be written.
output_format
ValueColor format to be written
SDRRGBA_8888
HDR_LINEAR(default) RGBA_F16 linear
HDR_PQRGBA_1010102 PQ
HDR_HLGRGBA_1010102 HLG
metadata (optional, default to NULL) Destination of metadata (gain map version, min/max content boost).

For more info:

About

Ultra HDR is a true HDR image format, and is backcompatible. libultrahdr is the reference codec for the Ultra HDR format. The codecs that support the format can render the HDR intent of the image on HDR displays; other codecs can still decode and display the SDR intent of the image.

Resources

License

Code of conduct

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • C++ 84.1%
  • Yacc 9.9%
  • C 3.2%
  • CMake 2.7%
  • Shell 0.1%