3 ms·

There’s a Twitter thread by a JPEG XL dev who tries to explain some of this in terms of the XYB color space giving more bits to dark tones: https://mobile.twit

by jl6 4y ago

There’s a Twitter thread by a JPEG XL dev who tries to explain some of this in terms of the XYB color space giving more bits to dark tones: https://mobile.twitter.com/jonsneyers/status/1550215859305586691 https://mobile.twitter.com/jonsneyers/status/155021585930558... But I’m not clear if this was a design goal of JPEG XL, or a theoretical side-benefit, or something you would need an HDR screen to appreciate (I couldn’t tell the difference in any of the example images, but those may have been processed by Twitter).

Semaphor 4y ago
Also not sure if this is a Twitter issue (wait, another reason not to use twitter? So why do people keep using it for something it’s bad at? /rant), but I got a different result. I can clearly see the differences between the encodings (I shuffled the images and sorted them by perceived quality), and I disagree that JXL looks better / is less racist. For both sides, the AVIF version looks significantly better, but especially for the black woman. The white woman already has fewer visible details that can be lost, the black woman has more, and JXL just smoothes those over, making it look like someone used the smudge tool (in general, but it’s especially bad here [0]). So going from this example, I’d say AVIF wins by quite some margin. [0]: https://i.imgur.com/2WU6iKe.png https://i.imgur.com/2WU6iKe.png
- jonsneyers 4y ago
  It was probably a bad idea to put such examples on twitter, since twitter recompression is quite aggressive and makes a big difference. I don't have the original decoded images nearby but it is best to do these kinds of comparisons without any recompression.
  - Semaphor 4y ago
    I’d happily look at them again in their actual versions
vanderZwan 4y ago
In this case I would say that regardless of intent (which may still be important) we should go for the better image format if we have the option. Like, to take the historical example mentioned in the NYT article linked above, this is obviously bad: > Concordia University professor Lorna Roth’s research has shown that it took complaints from corporate furniture and chocolate manufacturers in the 1960s and 1970s for Kodak to start to fix color photography’s bias. .. but without excusing any of that, black people were still better off with the newer films. Similarly, regardless of the intent of the JXL design, if it is the more "inclusive" tech then that speaks in favor of it. Also, the difference in the example images requires you to look actively look for JPG artifacts. That's something that usually only codec developers and people who work with images for their day job do. Almost everyone else subconsciously does the opposite: they learn how to "look past" the compression artifacts. But speaking as someone who did spend a lot of time trying to optimize JPG sizes for a short while in his life there is definitely a difference. The difference is hard to see when compressing an image once, but keep in mind that we live in a world where images go through lossy recompression dozens of times, causing the "needs more JPEG" meme[0][1] to live on even when it should be a thing of the past. If your image is, say, 5% more visually lossy for black people that adds up. [0] https://knowyourmeme.com/memes/needs-more-jpeg https://knowyourmeme.com/memes/needs-more-jpeg [1] http://needsmorejpeg.com/ http://needsmorejpeg.com/
worldofmatthew 4y ago
Besides that jokes you can make at that tweet in how it claims a image codec can be racist. If you know how to limit the image quality range in AVIF (and how to reduce smoothing which can be done), you can get images that have dark areas that are near to that of JPEG-XL. I have been able to (by both reducing the quality range and turning smoothing off) to get much better results on a high contrast image with most of its details in darker, having far-more details in those dark area with AVIF than JPEG-XL (not small differences either as you could see brick work that was destroyed by compression artifacts in JPEG-XL). libavif default settings suck.
- pbowyer 4y ago
  Can you share the settings that have worked well for high contrast, darker images? I agree the default settings are not great.

by jl6 4y ago

Semaphor 4y ago

Also not sure if this is a Twitter issue (wait, another reason not to use twitter? So why do people keep using it for something it’s bad at? /rant), but I got a different result. I can clearly see the differences between the encodings (I shuffled the images and sorted them by perceived quality), and I disagree that JXL looks better / is less racist. For both sides, the AVIF version looks significantly better, but especially for the black woman. The white woman already has fewer visible details that can be lost, the black woman has more, and JXL just smoothes those over, making it look like someone used the smudge tool (in general, but it’s especially bad here [0]). So going from this example, I’d say AVIF wins by quite some margin. [0]: https://i.imgur.com/2WU6iKe.png https://i.imgur.com/2WU6iKe.png

jonsneyers 4y ago
It was probably a bad idea to put such examples on twitter, since twitter recompression is quite aggressive and makes a big difference. I don't have the original decoded images nearby but it is best to do these kinds of comparisons without any recompression.
- Semaphor 4y ago
  I’d happily look at them again in their actual versions

vanderZwan 4y ago

In this case I would say that regardless of intent (which may still be important) we should go for the better image format if we have the option. Like, to take the historical example mentioned in the NYT article linked above, this is obviously bad: > Concordia University professor Lorna Roth’s research has shown that it took complaints from corporate furniture and chocolate manufacturers in the 1960s and 1970s for Kodak to start to fix color photography’s bias. .. but without excusing any of that, black people were still better off with the newer films. Similarly, regardless of the intent of the JXL design, if it is the more "inclusive" tech then that speaks in favor of it. Also, the difference in the example images requires you to look actively look for JPG artifacts. That's something that usually only codec developers and people who work with images for their day job do. Almost everyone else subconsciously does the opposite: they learn how to "look past" the compression artifacts. But speaking as someone who did spend a lot of time trying to optimize JPG sizes for a short while in his life there is definitely a difference. The difference is hard to see when compressing an image once, but keep in mind that we live in a world where images go through lossy recompression dozens of times, causing the "needs more JPEG" meme[0][1] to live on even when it should be a thing of the past. If your image is, say, 5% more visually lossy for black people that adds up. [0] https://knowyourmeme.com/memes/needs-more-jpeg https://knowyourmeme.com/memes/needs-more-jpeg [1] http://needsmorejpeg.com/ http://needsmorejpeg.com/

worldofmatthew 4y ago

Besides that jokes you can make at that tweet in how it claims a image codec can be racist. If you know how to limit the image quality range in AVIF (and how to reduce smoothing which can be done), you can get images that have dark areas that are near to that of JPEG-XL. I have been able to (by both reducing the quality range and turning smoothing off) to get much better results on a high contrast image with most of its details in darker, having far-more details in those dark area with AVIF than JPEG-XL (not small differences either as you could see brick work that was destroyed by compression artifacts in JPEG-XL). libavif default settings suck.

pbowyer 4y ago
Can you share the settings that have worked well for high contrast, darker images? I agree the default settings are not great.