compress.cafe

Despite being proprietary, Discord is a great place to talk to like-minded people. Here are a few Discord communities – they will be focused on AV1, but you can have great conversations about anything multimedia there.

• AV1 for dummies. The official server for the codec wiki. Small and new. I’m active here.
• AV1 Community. The largest AV1 Discord community, but I find the atmosphere unpleasant.
• The AV1 EX dev hub. Split from AV1 Community, probably the second largest one.
• AV1 weeb edition. This is the community of the AV1 “unofficial content distribution” scene.
• JPEG XL is the inofficial Discord community for the JPEG XL image codec.
• Jaded Encoding Thaumaturgy (JET). Formerly known as Irrational Encoding Wizardry, JET focuses on appealing anime encoding.
• VapourSynth is the official Discord community for the VapourSynth programmatic video editor. It’s often used for filtering to remove artifacts.
• VVC is the Discord community for Versatile Video Coding.
• Av1an is the official Discord community for the Av1an encoding toolkit.
• AWCY is the Discord community for Are We Compressed Yet.

• A digital media primer for geeks, watch this to learn what digital media is!
• Digital show and tell (more in-depth on sampling)
• Introduction to video coding

• Into the depths: The technical details behind AV1 by Nathan Egge
• Spec
• SVT-AV1 docs
• AV2 video codec – early performance evaluation of the research

• Falsehoods programmers believe about [video stuff]​
• comp.compression FAQ [1]​ [2]​ [3]​

• How to write a FFmpeg demuxer
• How to write a FFmpeg decoder
• How to write a FFmpeg filter

• Soichiro’s encoding guide describes a ton of stuff around encoding in useful ways.
• JET encoding guide
  • AV1 encoding for dummies
• The fansubbing guide gives you some background knowledge for encoding anime.

• ImageWorsenser has many pages on resampling. They’re great!
• on the theory of the Spline16/36/64 resizers - Doom9’s Forum
• The Magic Kernel (no consensus on accuracy, but seems good)

• How to perform fast motion search
• Hierarchical Estimation of the motion vector field

• The hitchhiker’s guide to digital color gives a good overview of color.
• RGB/XYZ matrices tells you how to convert between any color space and CIE XYZ.
• Completely painless programmer’s guide to XYZ, RGB, ICC, xyY, and TRCs
• The luminance of an sRGB color, among other things, answers the question of “Why is there a Y channel in the xyY components of some primaries, but not others?”
• History of the very odd sRGB color space
• Cone fundamentals & the LMS color space

• upscale.wiki model database

• Atkinson Dithering
• The constrained directional enhancement filter
• Linearly constrained generalized Lloyd algorithm for reduced codebook vector quantization describes an algorithm to do what S3TC does.
• Theora spec
• Notes on the implementation of trellis quantization in H.264
• Applying perceptual vector quantization outside of Daala
• Technical articles on the PNG file format and the related compression technologies

There’s a kind of chicken-and-egg problem here, where you need to write code that uses Core to know what needs to be in it. The source code is still unreleased.

• 2024-04-14, colorspace improvements

A single-producer, single-consumer atomic queue in preparation for filtergraphs in Core. Source code is on Codeberg. It’s usable, but basic.

A Zig library to detect x86 CPU features in an ergonomic way. Source code is on Codeberg.

Some temporary inline asm replacements for intrinsics, or lack thereof.

You can upload to files.compress.cafe. To do this, ask me for a key, then use the following script:

#! /bin/sh

key='insert the key I gave you here'

url="https://files.compress.cafe/upload?${key}"

if test -z "${1}"
then
    curl -X POST --data-binary '@-' "${url}"
else
    if test -z "${1##*.}"
    then
        curl -X POST --data-binary "@${1}" "${url}"
    else
        curl -X POST -H "X-Extension: ${1##*.}" --data-binary "@${1}" "${url}"
    fi
fi

(Is this cursed? Would love to hear some feedback.)