In all discussion of the modern image formats JXL and WebP, it has been mentioned that the formats support both lossless and lossy encoding, as if this is some sort of breakthrough feature. I think, at the very least, lossless and lossy image formats ought to use a different container, or file extension, so that -I- can differentiate them. As long as there is this sort of ambiguity, and unless search engines widely add a "lossless" option to their advanced image search tools, I will keep searching for "filetype:png" when I want a lossless image.
Hey, maybe some Googler could convince their boss to convince their boss to add the "filetype:lossless" flag, to support these new formats?
MiroslavPokorny•3h ago
I can think of a few others, such as bit depth, resolution and of course resource is fuzzy, and thats the problem. You cant really stuff all attributes into the filename or file extensions.