Hacker News new | ask | show | jobs
by danShumway 1049 days ago
I still need to sit down and convert my personal Linux computer over to using JPEG XL for picture archival and figure out what tools need to change or be updated.

Using it on the web is one thing, but getting better compression for my family photos would also probably be a win, and I suspect it would be possible to build a pipeline for viewing/editing that would be fairly transparent.
1 comments
chungy 1049 days ago
I've been pretty happy with it, myself. All the base OS libraries support JPEG XL so it's nearly pain-free, excepting lack of web browser support.

Combined with GNU parallel, I did this:

    find -type f -iname \*.jpg -print0 | parallel -0 cjxl --lossless_jpeg=1 {} {.}.jxl 
    find -type f -iname \*.png -print0 | parallel -0 cjxl -d 0 {} {.}.jxl
    find -type f -iname \*.webp -print0 | parallel -0 dwebp -o {.}.png {} \&\& cjxl -d 0 {.}.png {.}.jxl
JPEGs get losslessly recompressed with JPEG XL. PNGs and (lossless) WebPs get converted to lossless JPEG XL.
link
danShumway 1049 days ago
Nice, I should give this a try and see what it does to my photo library size.

There are a few holdout programs that I think are missing support (Blender3D springs to mind), but I also think in the rare instances where that's a problem I could probably set up a quick shortcut or some hooks to on-the-fly run cjxl and convert back to jpeg/png temporarily for whatever operation I need to do.

link