The AnimeRAster project aims to create a lossless image format with at least comparable compression ratios to the best lossless image methods, but optimized for drawings instead of photographs.
The following scenarios and features are considered:
Screen-caps of video sources
- YUV support with chroma-subsampling. Early tests suggest a 25-50% size reduction compared to converting it to RGB.
- High bit depths, for saving Hi10p without any loss at all
- Subtitles as a separate layer, both as a rendered overlay, and storing the ASS data as meta-data.
- More meta-data, such as file source and frame-number
Visual Novel CGs
Continuation of the cgCompress project.
- Multi-frame format to store all related images in one group.
- Exploit correlation between images. cgCompress already is 20-25% better than the next best alternative.
- Do it in the filtering phrase, and use the concepts from normal image compression to further improve compression.
- Consider how memory usage can be reduced when there is a lot of frames.
Status
Format far from finalized
- Basic single-frame RGB compression. Currently about 5 % worse than webp, and a bit more compared to BCIF.
- Basic single-frame YUV compression. Needs more experimentation with color decorrelation.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent