Created
September 21, 2023 19:15
-
-
Save pmarreck/3ac96786e6463d9a3a70f69d3cb5053d to your computer and use it in GitHub Desktop.
JPEG-XL tests
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
a few tests using `cjxl` (the reference `jpeg-xl` converter) I performed on a 120.7 MB PNG (the Carina Nebula, first photo from the James Webb Space Telescope): | |
there's 2 main parameters, "distance" (where 0 is lossless and 1 is the least detectable lossy; higher numbers are more lossy) and "effort" (compression effort, where 1 is quick and 9 is exhaustive/very long) | |
distance 0 effort 9 took 3 hours (obviously this is for a "write once, read many" sort of archival use-case) and got it down to about half of PNG, 67.2 MB; not bad for maintaining lossless. | |
distance 0 effort 6 took 1m43s, got it down to 75 MB, about 63% of PNG. | |
distance 1 effort 8 took 3 minutes and got it down to 7.7 MB which is about 1/15th of the PNG. | |
distance 1 effort 6 took 4s; got it to 7.1 MB... wait, _whut?_ That's... odd. Smaller than with effort 8. Anyway. | |
All encodings only used 1 of my 128 cores. Whole-image compression of this nature may not be "trivially parallellizable" ::shrug:: |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment