"J. P. Gilliver (John)" wrote

| a format for storing photos. Similarly with GIF: It's
| handy for creating small files and it's cross-platform,
| but it's lossy insofar as it reduces an image to 8-bit
| color.
| []
| Not quite: it reduces it to 8-bit _storage_, but it does that by using a
| palette. I think the palette entries are at least 16-bit. Basically, it
| reduce an image to 256 _colours_, but they're not the _same_ 256 for any
| given image: a picture of a sunset, for example, will have a lot of
| oranges and reds.

I wouldn't argue with that. But it's still reduced to a
max of 256 colors. It's best for logos, cartoons,
simple images. A sunset will dither. (Remember the
old days on Windows monitors? If you used a sunset
desktop photo you would have had stripes.)

| And once the reduction has been done, there's no
| _further_ compression

As I understand it there is, but it's not lossy. It's
a formulaic system that will record things like "43
pixels of color #18" as a data record, rather than
using 43 * 3 bytes to record 43 pixels. It's very
efficient in that context because repeating pixels are
the norm.

(though some image editors - like, unfortunately,
| IrfanView, which I think is great in most respects - tend to operate in
| maximum-colours mode, so edit actions in them _do_ cause degradation
| when resaved in GIF. But that's not the format's "fault"; if the editors
| could be constrained to work in 256-colour mode, there'd be no loss).

That's just not true. Few 24-bit images use only
256 colors. Try this one:
https://www.jsware.net/Files2/sunsetMV.jpg

IrfanView says there are 100,627 unique colors there.
If I reduce to 256 colors in PSP I get something
like a comic book image, where Superman is in 3
colors.
There are 3 reduction routines down to 256 colors
and the effect varies with each, but all drop out a
tremendous amount of data. If I save as GIF from
PSP I get a pointilistic image. PSP16 does a slightly
smoother job of it than PSP5, but both end up
looking like a print from an old printer. And that
image started as a low quality JPG that had already
been resaved at least twice, so it wasn't a great
picture to begin with. It had already dumped a
lot of the richness. The degradation from the original
would have been heartbreaking to see.

To me that's a great example of the role of JPG
and GIF: Great for onscreen images that need to
be small and that need to be accessible across
platforms. I use GIFs a lot for diagrams. But they're
not good for much else. It would be crazy to store
photos as GIF in order to save space.

I find it kind of ironic when this topic comes up.
I don't think I've ever heard you say this, but whenever
I talk about conserving space on disk, many people
will respond with, "Ah, that's not worth the trouble.
Disks are so cheap these days!" Yet when it comes
to saving large images for a good reason, those
same people think it's crazy: "Takes up WAY too much
space!"

I suspect most people who feel that way are taking
loads of pictures with their phone. They just want
2,000 vacation photos to fit on disk. They have no
intention of doing any involved editing or printing of
those images, or even going through to dump the
bad ones, so they're happy with downgraded JPGs.