Overview of JPEG XT

JPEG XT (ISO/IEC 18477) specifies a series of backwards compatible extensions to the legacy JPEG standard (ITU Recommendation T.81 | ISO/IEC 10918-1).

While JPEG is still the dominant technology for storing digital images, it fails to address several requirements that have became important in recent years, such as compression of images with higher bit depths (9 to 16 bits), high-dynamic-range imaging, lossless compression, and representation of alpha channels.

JPEG XT extends the JPEG specification in a completely backwards compatible way. Existing tools and software will continue to work with the new code streams, while new features will help move JPEG into the 21st century.

JPEG XT, like most other JPEG standards, is a multi part specification. The following parts are currently in the process of standardization:

Part 1, Core coding system

JPEG XT Part 1 specifies the base technology, and specifies as such the core JPEG as it is used nowadays, namely as a selection of features from ISO/IEC 10918-1, 10918-5 and 10918-6. Part 1 defines as what is commonly understood as JPEG today.

Part 2, Coding of high dynamic range images

JPEG XT Part 2 is a backwards compatible extension of JPEG towards high-dynamic range photography using a legacy text-based encoding technology for its metadata.

Part 3, Box file format

JPEG XT Part 3 specifies an extensible boxed-based file format all following and future extensions of JPEG will be based on. The format specified in Part 3 is itself compatible to JFIF, ISO/IEC 10918-5, and thus can be read by all existing implementations.

Part 4, Conformance testing

JPEG XT Part 4 defines conformance testing of JPEG XT.

Part 5, Reference software

JPEG XT Part 5 provides the JPEG XT reference software.

Part 6, IDR Integer coding

JPEG XT Part 6 defines extensions of JPEG for backwards compatible coding of integer samples between 9 and 16 bit precision. It uses the file format specified in Part 3.

Part 7, HDR floating-point coding

JPEG XT Part 7 uses the mechanism of Part 3 to extend JPEG for coding of HDR images, i.e. images consisting of floating point samples. It is a super-set of both Part 2 and Part 3 and offers additional coding tools addressing needs of low-complexity or hardware implementations.

Part 8, Lossless and near-lossless coding

JPEG XT Part 8 defines lossless coding mechanisms for integer and floating point samples. It is an extension of Part 6 and Part 7, allowing for scalable lossy to lossless compression.

Part 9, Alpha channel coding

JPEG XT Part 9 allows the lossy and lossless representation of alpha channels, thus enabling the coding of transparency information and conding of arbitrarily shaped images.