Template talk:Compression methods

Separating algorithms from the actual compressors?
I see there is Brotli and LZ4 in "Dictionary type" - these are specific compressors, while the rest are more abstract algorithms. Maybe there should be created a separate category in "Lossless": "Compressors" with the most notable ones. If so, "compressors" should be also added to other categories, for example: Lossless: gzip, Brotli, Zopfli, LZ4, PAQ, ZPAQ, ZStandard, bzip2, WinRar, LZFSE ... FreeArc, 7zip Image: jpeg, LZ-JPEG, gif, png, WebP, jpeg2000 ... Video: h.264, h.265, VP9, Dirac, Daala, AV1 ...

Also, I believe there should be clearly distinguished transformations (LZ, BWT, MTF, RLE, DCT) from statistical prediction methods (PPM, Context Mixing) ... and there is missing article e.g. "Statistical modeling for data compression" - which describes and compares: - order, for Markov modeling, - static: count frequencies and write used probabilities in headers, - adaptive: updates probability distribution - with example of realization, like here: https://fgiesen.wordpress.com/2015/05/26/models-for-adaptive-arithmetic-coding/ - PPM - context mixing Jarek Duda (talk) 10:28, 30 October 2016 (UTC)