[throwanem]

Okay, no, that's fair. You're absolutely right! I messed up. I made a mistake.

Specifically, I took the people who promote the Gimp at their word when they said it was, in every respect, a viable libre replacement for Photoshop. And I took the Gimp's ability to do trivial work acceptably, if without much comfort in the UI, as cause for confidence that it would do significantly complex work acceptably, as well. You're right, though. When quality of results really counted, I was wrong to rely on the Gimp.

Those claims of quality are still made on Gimp's behalf, maybe you know. RMS has been known to repeat them in public. It may interest you to hear that, when he and I had this same argument, I recall there being a great deal less swearing involved, and many fewer personal attacks. I have to admit, I don't really find those additions to be an improvement.

It's odd, though. By default, the claim is still that Gimp is a viable libre replacement for Photoshop. But as soon as someone happens to criticize some specific aspect of Gimp's functionality - in this case, its ongoing inability to render text at a level of quality comparable with Photoshop and with its commercial competitors more generally - suddenly "everybody knows" that that specific part of the Gimp isn't ready for prime time, never has been, and anyone would have to be a complete muppet to imagine it was intended for serious use.

I don't really know why that is. But, whatever the reason, it definitely doesn't incline me to feel differently about the software. I can't in good conscience recommend anyone use a tool that even its own strongest advocates so readily agree is so frequently unfit for purpose. That would be a worse mistake than to ever have thought the Gimp to be reliable in the first place.

[prokoudine]

> Specifically, I took the people who promote the Gimp at their word when they said it was, in every respect, a viable libre replacement for Photoshop.

You know what, this is where I completely agree with you. I'm very much against the idea of promoting GIMP as a Photoshop replacement because it will always lead to frustration. It cannot be otherwise. The software was never designed for that.

There are unavoidable similarities, some tools are specifically designed like Photoshop's ones simply because the way it's done in Photoshop makes sense for GIMP as well. But that is pretty much it.

> Those claims of quality are still made on Gimp's behalf, maybe you know. RMS has been known to repeat them in public.

I never heard RMS saying any such thing. But then again, how would I know? I don't follow him, I can't stand the guy.

> But as soon as someone happens to criticize some specific aspect of Gimp's functionality - in this case, its ongoing inability to render text at a level of quality comparable with Photoshop and with its commercial competitors more generally - suddenly "everybody knows" that that specific part of the Gimp isn't ready for prime time, never has been, and anyone would have to be a complete muppet to imagine it was intended for serious use.

I'm afraid you are conflating things here. So lemme unload a little.

Like every other software (incl. Photoshop), GIMP has loose ends, bad design decisions, etc. As a team member, I don't mind admitting it. Noone in the team minds publicly admitting it. There's no point arguing against obvious things.

So... Can you do serious work with GIMP? Yes, we've seen use cases of complex work. Commercial-grade work, shitload of layers etc.

Do you need workarounds? Depends on the project and the kind of manipulations involved.

Are there things impossible to do as compared to Photoshop? Yes, of course. All the 3D stuff, vector layers, smart objects are among the first things I can think of.

Are things getting better? Yes. Just two days ago I was talking to our guy who does a lot of work on performance. He has a test project file from a user. A real project, 500+ layers, over 1GB large. GIMP used to just crash on it. Now? No crashes, pretty much usable.

Would you be pleased if you tried again? I don't know, that is not up to me to decide.

[throwanem]

I doubt I would, not least because most of my use cases for image editing these days revolve around photography, and FOSS library support for late-model Nikon raws just isn't where I need it to be. No shade on the devs and maintainers, it's a closed format and they have no support. But I still can't get the quality out of Darktable that I can from Lightroom, so I use Lightroom. (I'd miss my Loupedeck a lot, too.)

I did pull down the current Gimp yesterday to do a quick test of text rendering cases similar to the ones that it failed so badly on back a decade ago. The results were the same as I remember them: aliasing everywhere, illegible at small sizes even in very high-DPI files. If I had to guess, I'd think Gimp just always rasterizes text at 72dpi and then nearest-neighbors it up to match the file resolution, but that's just a guess based on the behavior I'm seeing; I haven't been into the code.

Granted, high-DPI displays were quite rare back in 2009-2010. Print workflows weren't, though. In 2020, displays >72dpi are rapidly becoming the default. Print still exists, too. Both of those are things you'd expect to see well supported in a tool whose homepage advertises it as suitable for photographers and graphic designers, among others.

[enriquto]

Can't you just point to a concrete, specific problem with the Gimp (apart from its name, or the fact that it has a couple of easter eggs)? This is getting too abstract to be useful.