- cross-posted to:
- hackernews@derp.foo
- cross-posted to:
- hackernews@derp.foo
JPEG 's conspicuous issues like lack of transparency, animation, lossless compression and high bit depth support makes it little tough to justify it as leading format for time to come. Various attempts like WebP and AVIF are made (and are supported on modern browsers) to overcome problems of JPEG but they all also suffer some shortcomings of their own. Rather than discussing it, let’s see how is the web finally going to change with this new format - JPEG-XL.
JPEG-XL is what we believe is next 20 year format. Why?
It’s not one but multiple reasons.
60-70% smaller size compared to similar quality JPEG 😮 Dimension of upto billion pixels allowed (JPEG-XL exclusive) 4099 channels support (JPEG-XL exclusive) Animation support Tile support for large images Progressive decoding (see low resolution image first before full image loads - JPEG-XL and JPEG exclusive) Lossless encoding support Wide-color gamut support Extremely fast for both encoding and decoding (WebP and AVIF are much slower) Royalty free 🤘 This basically means anything that a photographer, developer or animation creator can think of, it’s supported by this format. Everyone stays happy and gets to use one format for everything.
No layer support? Or can that be faked by using 4 channel-chunks?
Probably easier to fake with frames of animation, but JXL is not a replacement for .PSD or .XCF files and I do not know what usecase layers would be for besides authoring
In my company’s product, we allow users to place images into a composition and alter the colors (like tinting/colorize in Photoshop, so only globally). When doing that, the outline of the drawn image should not change color, and also there might be multiple color areas (think an image of a house where the walls have a separate color from the chimneys).
To do separate full-image manipulation, we need multiple layers. Right now they’re just a tilemap with the layers next to each other, but that doesn’t work for repeating patterns.