Custom Simple Container

This repository contains a demo implentation of a strict minimal file structure.

Objectives

The RIFF container from WebP was not fully used. Some overhead can be reduced:

• No need for hierarchy, a flat structure works.
• Only signal what is needed to decode or skip the next chunk. Encoding the total file size is not needed.
• No progressive decoding but a solid color signaled in the header and/or a preview chunk of a few hundred bytes.
• Animation frame count is not stored in the header. Each frame should signal whether there is another one afterwards (left to codec implementation).
• Drop niche loop count. An animation either plays once or loops infinitely.
• Only the ICC profile is necessary before pixel decoding. XMP and EXIF metadata is located at the end of the bitstream.
• Alpha data should be interleaved within pixel chunks instead of having a separate cumbersome alpha chunk (left to codec implementation).
• Spatially slice the image into independent tiles, to limit memory usage and for easier encoding / decoding multithreading (left to codec implementation).

Specification

Coding

There are two ways of signaling a container value:

• Bit-packing: append booleans as single bits and integers as bit sequences.
• Variable-length integers: Each byte has one bit reserved, which is set to 1 if another byte is needed to store the whole value. Up to 4 bytes.

Chunks

An image has the following chunks in order:

Header

Being 10-byte long, the main header contains the most important information about the image, such as the format tag, width, height, orientation, has alpha, is animated, etc.
It uses bit-packing for easy encoding and decoding on top of providing a fixed header size.

Regular chunk

All remaining data is stored sequentially in regular chunks, whether it is a preview, pixels, XMP etc. A chunk is preceded by its size encoded as a variable-length integer, limiting it to ~500 megabytes.

Library API

Work in progress. See container.h.

WrapContainer() and UnwrapContainer() functions can be used to encapsulate encoded pixels in any format into a minimal image container. EncodeRawPixels() and DecodeRawPixels() are just a convenient demo for a generic “codec”.

Build and test

mkdir ../container_build && \
cd ../container_build && \
cmake -DCMAKE_BUILD_TYPE=Debug ../container && \
make && \
ctest .

Conversion from/to AVIF

AVIF is a MIAF/ISOBMFF wrapper around an OBU bitstream, which is itself usually made of a sequence header, a frame header and an AV1 tile group.

Stripping everything but the AV1 tile group and the necessary fields in the sequence and frame headers would produce lighter web image files:

• Easy to parse and write,
• No redundant field that could be contradictory and/or invalid (image dimensions stored in MIAF, sequence header and frame header),
• Fewer features means low vulnerability,
• Smaller files, especially for tiny images or icons.

Convert tool

Work in progress.

1. Start with an AVIF file (avifenc img.png -o img.avif -c aom).

2. Convert from AVIF:

   ./convert img.avif -o img.pic
   

3. Convert back to AVIF:

   ./convert img.pic -o img.avif
   

OBU layout

Taken from https://aomediacodec.github.io/av1-spec.

Comparison between a minimal header (left) and a Sequence Header OBU (right):

Frame header OBU: