Talk:Data compression

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

introduction needs rephrasing maybe...?[edit]

[...] imposes computational or other costs through decompression; this situation is far from being a free lunch. Data compression is subject to [...]

A free lunch, really? Sure it's in context but is it really contributing anything useful to the article?

Block compression[edit]

What is block compression?

I feel that readers who see the phrase "block compression" and look it up on Wikipedia should be directed to an article that at least mentions "block compression", in accordance with our WP:R#ASTONISH guideline.

Currently Wikipedia redirects that phrase to this data compression article, which never mentions the phrase "block compression". What is the best way to comply with that guideline: (a) Mention "block compression" in this "data compression" article? (b) Re-point that redirect to some other article that mentions "block compression"? Or (c) Change that redirect into an article about block compression?

In my experience with embedded systems and GPU-accelerated graphics processing, "block compression" usually refers to a kind of image compression that divides up a raster image into fixed-sized square blocks of pixels (typically 4x4 pixels), and compresses each block into a fixed-size block of coding bits. For example, "Block Compression 2" (BC2) is apparently part of the S3 Texture Compression de-facto standard for GPU texture compression. Block Truncation Coding (BTC) has been used in spacecraft image transmission.

I see that "block compression" previously redirected to the discrete cosine transform (DCT) article, which at least mentions "block compression". The people who wrote that article describe "block compression" as a slightly different kind of image compression that divides a raster image into fixed-size square blocks of pixels (typically 8x8 pixels), and compresses each block (using a discrete cosine transform) into a variable-sized block of coding bits.

--DavidCary (talk) 19:50, 24 March 2022 (UTC)[reply]

TYPING STYLE[edit]

TYPING STYLE 103.214.139.37 (talk) 11:55, 1 July 2022 (UTC)[reply]

AI[edit]

User:Vladimir Alexiev has added information about AI. I fear these sources are more about clickbait. I'm not sure how much actual AI is involved. Can go back these days and label any clever algorithm AI? ~Kvng (talk) 11:30, 12 April 2023 (UTC)[reply]

Hi!
I tried to check some papers by two of the cited authors: Gilad David Maayan and Gary Adcock.
  • Can't find reliable papers by Gilad, though some of his work on IoT is cited by others.
  • Can't find reliable papers by Gary, but he's cited in several relevant books or meetings, eg
    • "Color correction for video: using desktop tools to perfect your image" … For great contributions of video examples, I’m lucky to know Gary Adcock, an international expert on video and data cameras and workflow"
    • "The art and technique of digital color correction" … And to noted guru Gary Adcock, for helping with HD enlightenment and for the use of his Panasonic HVX-200.
    • "SMPTE Motion Imaging …, 2022 Section Meetings"… Presenters Gary Adcock, Studio37 Inc.,
    • "SMPTE Motion …, 2017 Section Meetings" … Gary Adcock, product ambassador for Vitec, the makers of the Vinten pedestals and robotic camera supports
I also browsed a bit MATLAB's Image Processing Toolbox, here's a blog/doc that's definitely about AI/ML: Deep Learning in Image Processing.
I should explain why I made this addition: we're working on a comprehensive taxonomy of AI (using Wikipedia categories as a key ingredient). I saw some references of application of AI to compression, so I did a bit of research, and also added cat "Applications of AI" to "video compression
I'm happy to be proven wrong! -- Vladimir Alexiev (talk) 10:30, 30 April 2023 (UTC)[reply]
@Vladimir Alexiev, thanks for the response. These new things you mention are more related to image processing than compression. I've read your original sources more carefully and they do support the text that you've added. That's good. I'm still believe what's being discussed here is more aspirational and research than current practice but things move quickly so let's leave it in there for now. Thanks for your contribution to Wikipedia. ~Kvng (talk) 14:26, 4 May 2023 (UTC)[reply]