Gimp 3 Beta Released
gimp.org
gimp.org
Here's a more recent beta from February 2022: https://www.gimp.org/news/2022/02/25/gimp-2-99-10-released/
If I create a selection, then copy and paste, why isn't the result a new layer? Why is it something floating? I don't get that.
And why do all tool settings suddenly change when I start using my Wacom instead of my mouse?
Of course there are probably good answers for the above but it is just not intuitive.
It's a little sad because so much time and energy is put into Gimp. But every time I use it I start looking for alternatives. Krita for example.
I have been a Blender user for over 15 years. While the UI had a learning curve it always made sense to me. With Gimp this is not the case. I sometimes wonder if the people who built it also use it.
Affinity tools are great-- unlike Adobe, they are a one time purchase, and really polished and easy to use.
The other day I was trying to apply a mask on a layer and in the tutorial they painted it with a black brush to hide some parts of the changes on the layer. I failed miserably because:
1. The toolbar has like 5 different tools with a brush icon, I had no idea which one to choose
2. I tried them all, and the result was always the same: I paint with the black paint, see the changes, release my click, and 1 second later the changes are cancelled for no apparent reason
I feel like having a global command palette would help keep tutorials relevant for a longer period of time, because if I don't find an icon I could just global search it by name...
Particularly working with masks/alpha channels is sub par. The developers are aware but a fix has not been release in years. Users request about this regularly but nothing happens.[1]
1 https://forum.affinity.serif.com/index.php?/topic/109734-fea...
> In the below example, I am storing a semantic name for several layers featuring sleeping marmots
My guess? Gimp is a design tool made by developers with some graphic design inclinations. That's why it's weird.
I don't think you'll ever hear <<ANY>> actual, professional designer say "semantic name".
>. For the geekiest of us: by default, this uses simple text search, but in Preferences, tab “Interface”, you can also set the search entry to use the Glob or Regular Expression syntaxes!
You don't find this kind of language in Photoshop documentation
So it's extra frustrating, as creating an understandable UI is evidently possible.
Great interview with Blender Guru: https://youtu.be/qJEWOTZnFeg
If I was to go back to making indie games or indie music videos I would rather pay autodesk $50 a month for a hobby/indie license than get grey hairs from using Blender. I would love to not give that money to autodesk and use blender instead but the UI pain threshold is still too high for me to bypass. I really hope progress continues and maybe in 5 years things will actually be better.
Blender was designed from day 1 by and for working 3D artists that had to deliver real commercial projects. Gimp started as a university student project meant as a platform for playing with different image processing algorithms. Those roots clearly shine through for both projects even today.
Gimp's interface is a bit janky but after the initial frustration it's always made sense to me... then again I started using it as a replacement for my own image editor which I made as a university student project, so maybe there's a reason for that. :P
I can't confirm it but I wouldn't be surprised if the floating selection thing was also Photoshop behavior at the time.
So the difference isn't really how gimp started, it's just that development was mostly dead for 20 years and nobody put the work into improving the UI.
Blender also used to have a bad UI but people worked really hard to change that.
Without having been exposed to other paint programs, this seems reasonable to me. You said it yourself, you create a selection, then copy / paste. So why would the result be a new layer? You didn't copy a layer, you copied a selection. And this way you can either anchor it into the same layer (which I often do) or decide that it should be a new layer and get the result you're expecting.
I just gave it a try and I could load the image, edit it in various ways, and save it just fine. If there are any features that require being online, I didn't find them.
We all assume there's a vast army of dedicated people behind these things. Most people do just a tiny bit and don't stick around.
The reality is the core ranks are thin and the troops are exhausted. If you've got the time, they could use a hand (this is for all open source projects)
They people behind them are also usually effectively invisible when compared to the effort they put in so if you're thin on time, hunting down an email to send a sincere personal thank you would really make their day.
Just another lame example of how GitHub both (a) insists upon itself and (b) prioritizes its social network over providing actually useful tooling for getting work done (including but not limited to trying to glean insights about a given project).
GitHub only shows people with an email address linked to a GitHub account in that chart; the last time I checked it was 4 or 5 people working on it regularly, which is still a very small team (none are working on it full-time) so your point still stands (it's a point I've made myself a few times before when people compare GIMP to Photoshop or the like).
The GitLab chart[1] also doesn't seem to be accurate; e.g. Mitch only has 986 commits on that one (and almost 14k on the GitHub chart).
One example to test this: create a new layer, select a rectangular area, fill it with the paint bucket tool. Then use free transform to rotate that rectangle 45 degrees. The edges will be a mess. Its even worse at other angles.
I don’t want to slander Photopea because as I said I use it a lot, but I’d suggest to spend a while with it before deciding that you’re ready to ditch Photoshop/GIMP/Affinity.
regardless, congrats for the release! I wish them all the luck and thanks for supporting the software community
https://www.gimp.org/ still describes it as the GNU Image Manipulation Program in the <title>, in giant all caps letters at the top, and a third time at the center of the first screenfull of the main page ("This is the official website of the GNU Image Manipulation Program (GIMP)."). https://www.gimp.org/about/ also links to the GNU Project under "Important GIMP Links."
Gimp is listed as a GNU package: https://www.gnu.org/software/software.en.html#allgnupkgs
Unfortunately I can neither edit not delete my inaccurate comment.
https://www.gimp.org/donating/
Most work on GNU software is either done unpaid by volunteers or paid for by companies like RedHat who spend engineer time on it.
I remember years ago it was a planned feature, but recently I tried to gather info about it and could not find anything. Has the team decided to remove it from scope?
Currently I know of Cyan[1] as an open source tool that I can use to prepare an asset for printing, happy to hear about any other good alternatives.
EDIT: Here's an issue in the Milestone 3 that is concerned with CMYK exporting: https://gitlab.gnome.org/GNOME/gimp/-/issues/356
It's a shame, but no wonder people move on.
[0] https://docs.krita.org/en/general_concepts/colors/color_mode...
Thanks, I was aware that Krita could do CMYK but how about importing assets created in other programs? Can it import any type of image files? I ask because I thought that Krita was more of a drawing/painting program and I'm not familiar with its import capabilites.
When I talk to designers about Gimp they say "Uh, I tried it once years ago. But back then it was so archaic that you could not even change the contrast settings after you did something else to the image".
Little do they know. It is still that archaic.
12 years ago: https://news.ycombinator.com/item?id=2091318
11 years ago: https://news.ycombinator.com/item?id=2890549
10 years ago: https://news.ycombinator.com/item?id=4814360
9 years ago: https://news.ycombinator.com/item?id=5912145
8 years ago: https://news.ycombinator.com/item?id=8969088
7 years ago: https://news.ycombinator.com/item?id=9932717
6 years ago: https://news.ycombinator.com/item?id=12092173
5 years ago: https://news.ycombinator.com/item?id=15101108
4 years ago: https://news.ycombinator.com/item?id=17926027
3 years ago: https://news.ycombinator.com/item?id=20422647
2 years ago: https://news.ycombinator.com/item?id=25012155
1 year ago: https://news.ycombinator.com/item?id=29005637
Today: https://news.ycombinator.com/item?id=31419514
Don't get me wrong. I use Gimp daily and love it. I am very grateful to the people who build it. But how this feature can be ignored after people have been begging for it for over a decade now is beyond me. As far as I can see, it is the number one reason why people do not consider Gimp as a Photoshop replacement.
If you can't get traction and it's really objectively that big of an issue, a separate post seems more appropriate.
I taught Photoshop at the college level prior to 2010, offered experienced PS students extra credit to try out and evaluate software like GIMP, and many of them thanked me for the experience and went on to use it in their professional life. Those who didn't want to take the PS route seemed more than happy to duplicate layers as backups and work destructively a lot of the time, too. We focused on pixel editing logic quite a bit so maybe that's why.
But "silent mouths don't get fed". And I think I am speaking for a large number of users. And it's been over ten years now. So the voice of users might become a bit louder? I dunno. I read through all the features in the announcement and think "I don't need any of that. Who asked for all these things? Are they really user driven?".
Edit: Probably not in the form of HN search results though? Lol come on, a bit of a quality issue there if we're going for objectivity.
GIMP devs have been hammered by highly subjective UX & feature critics for decades now. They are humans and know they'd be doing work on behalf of people who seem complainy, persistent, and lacking perspective compared to their own priority view. So, software Karens? :-) Where's the carrot? How is this issue supposed to turn out well for anyone involved?
If it's really objective and a big deal and not just subjective annoyance compounded by years of boundaries set in your path, show the proof and be convincing about it. And this may seem a stretch but I think it can be done gently if you will it.
get those users to give you objective
evidence that they exist
https://hn.algolia.com/?dateRange=all&page=0&prefix=false&qu...
Here are a few hundred thousand:
If you want Free Software to be different, change it. If you don't know how now, like with anything else, you can learn how, or you can pay someone who has already spent the time to learn. The gimp project has a page where you can easily reach out to developers working on Gimp and fund them directly. If you think this feature is important, you should reach out and find out how you can help make it happen.
Now I don't work on gimp, but I do make Free Software, and I think most of us who program Free Software aren't working for "exposure". We aren't excited to be "used" by "users", and so of course we do not do anything "user driven". We do stuff because we want to, and when we don't want to, we may want-to for money. Like everybody else.
Imagine someone building furniture in their free time because they enjoy doing so, and then offers to give it away for free. You see this offer and now have two polite options:
1) decline, because you do not like it
2) accept because you do like it
I have never understood the people who take the third option:
3) complain that the furniture is not to their liking, and then demand that the person builds it to his/hers specification and then give it to them. Free of charge, because that was the original offer right?
Every time there's a push for commercial apps on desktop Linux, you have a bunch of rabblerousers throwing dirt at said devs. "They should make it Open Source! Micro$oft paid them off"
You can see it in every Linux forum.
At some point Linux distributions should just admit defeat for this kind of software, recognize that the Gimp and co. are just hobbyist software, and accept the real world and endorse commercial software for these use cases.
Instead every time this kind of discussion happens, someone compares a kiddie toy truck (Gimp) to a 10 ton semi (Photoshop).
It's really a shame when true commercial quality Open Source software exists out there, such as Blender.
That might be because the Blender users banded together and bought Blender[1] (with the help of one of the authors IIRC)
[1]: https://web.archive.org/web/20021010045558/http://www.blende...
> At some point Linux distributions should just admit defeat for this kind of software, recognize that the Gimp and co. are just hobbyist software, and accept the real world and endorse commercial software for these use cases.
Aren't they already doing that? I think commercial software has been targeting Linux almost since it came on the scene, simply because it made eval easier. Ubuntu distributes lots of nonfree stuff. If you count the enterprise (and perhaps depending on how you count that), I think it's possible more commercial software runs on Linux than on any other operating system.
But I can certainly agree there just isn't very much good commercial Linux desktop software. Do you think it's possible perhaps too many Linux desktop users consider price to be the biggest reason they use Linux (or assume, reading comments like that online that most users anyway) and so there just isn't any money to be had?
As long as it is missing, you have to duplicate layers and do some mental version control, which becomes really memory intensive after you have descended the tree a few steps. That's why you cannot really recommend Gimp to anyone, who wants to work professionally with it.
It is also a reason why Gimp cannot get better at Photoshop compatibility, so the issue even blocks the second biggest reason hindering further Gimp adoption.
Not entirely convinced the archaic base is even worth building off at this point.
I’d be more inspired in building the thing that would destroy GIMP and Photoshop than help make GIMP better.
I just started looking into Krita for bitmap editing. It has adjustment layers (which I can't do without). Switching from Lightroom to darktable was quite dramatic, but I actually prefer darktable now. Took me two years of on and off fiddling to finally get to that point.
If you want to argue this from first principles, it's actually very easy: One of the first principles of UX is to respect the user's work. Destructive editing is pretty bad at that. qed.
Because begging for something which ordinarily is very expensive to be done for you for nothing is a meritless activity. An infinity of begging pays for not a single second of developer time.
It is likewise useless to suggest people would use it for free if only a few more million dollars worth of developer time was invested by other people. This isn't an offer of a favor. It's more like me saying that I would gladly come and stay in your guest room if you would only serve me coffee and donuts.
Just developer time alone invested doesn’t have inherent value if it doesn’t make the end product better.
Having fun developing whatever stuff they wanted, would be my guess.
And if they're getting paid for this, probably doing a good job convincing whoever's paying them that that stuff is important :-)
Just feels what this project is shifts into whatever it needs to be to ensure it doesn't ever get any better or progress in any meaningful way.
There is a difference between desiring to serve users needs and obligation to serve a particular need. No amount of need on your part constitutes an obligation on their part because simply you have paid them nothing and they offered you nothing other than the ability to use the current software as is or modify it. Nor indeed are they obliged to provide you a way to pay them for a particular feature although if you offered them a sufficiently large pile of money you might be able to reach a mutual understanding and certainly nothing prevents you from developing a fork or paying someone to do so right now.
A large problem with such an understanding forming is that the cost of such would be driven by the cost of expensive labor not your perception of the value to an individual end user. I have no idea what the actual cost of this is as I know nothing about their code but lets do a thought experiment.
Imagine if you will that making that happen in 2023 were to require 2 people who currently contribute to gimp part time to work on gimp full time for a year. This would in turn require replacing what they earn at their actual job including benefits. Suppose that for example required replacing $300,000 in wages. Unfortunately they may actually have long term employment that they like and not be enthused about working for one year and then going back on the job market. It might be much more feasible if it were possible to fund gimp development similarly for 5 years enabling not only your feature but others. This costs $1.5M dollars.
If someone responded to your feature request with sure thing will that be check charge or giant bag with a dollar sign I'm guessing you would laugh your ass off and move on.
https://gitlab.gnome.org/GNOME/gimp/-/milestones
Gimp 3.0 is still at 23%.
No adjustment layers stops every designer cold.
Makes every other effort seem like "the road to get to adjustment layers".
I've sent many an invoice for work done in those very apps...the city where I live (3D modeling illustrations for promoting a concert series, rendered at 12+ MP for print), a distributor of various different wine brands, and a grocery business are a few example clients.
That alone was a lot of work, hundreds of hours at least.
same for kdenlive/libreoffice. these two are there but people don't take them seriously. kdenlive i hear has stability issues and libreoffice has compatibility with MS office.
gimp is still a meme.
LibreOffice would be a "maybe", depending on some factors. The others though, seemd to be a ways off.
Every single junior who asks me what 3D software to learn I say Blender, I would have used to say C4D but Blender has improved so much. Never in a million years would I suggest GIMP or Libreoffice.
The support workload is not worth the pitiful money involved.
Do Adobe products supports that?
On Windows they support COM, so any language with access to COM (which is a lot of them) can do it.
GIMP always damages my indexed PNG's when it opens them, and I have to remember to convert them to RGBA before opening them, and then back to Index+Alpha.
The software itself just isn’t anywhere near the feature parity it needs to be. The UI and workflows are just not there for what users are expecting. I can’t suggest someone learn it because it just isn’t going to serve them well in the end.
But because it exists it means someone who would do it better has less incentive to do so.
Not that I consider Photoshop a good image editor anymore, over the past 8 years it’s degraded horribly. So now we just don’t have good image editors.
> But because it exists it means someone who would do it better has less incentive to do so.
That makes no sense. There are a ton of great image editors always striving to do better in the same domain - Affinity, Paint.NET, Photoscape, Canva, etc.
None of these fill the niche GIMP has claimed and I just feel in whatever alternate reality where the creators of GIMP never made it that someone else would have made an app that would manage to ship adjustment layers within a decade.
This just seems like such a poor take. Can you imagine being a gimp dev and reading this?
Non-destructive editing next please.
Non-destructive editing next please.
It's been the main thing holding gimp back for decades. Even if it's just the run of the mill "layer styles" and "adjustment layers".
Always good to see a basic UI pass but we've had PhotoGIMP forever.
Thank you.
https://sourceforge.net/projects/pixelitor/
It didn't run here in OpenJDK due to class versions, maybe Oracle Java is best.
Good alternative or not depends on what your needs are though, good luck.
Just read some of their artist interviews (https://krita.org/en/features/artist-interviews/). Also, being part of KDE, they have a very strong UI/UX team that does lots of a/b testing and other usability testing, which I sometimes feel the Gtk devs are lacking in (however, I don't follow Gtk development too closely so I might be wrong here)
The GIMP just doesn't manage to hit the productivity nexus for anything constructive. Simple things are annoying, complex things are infuriating, vector support sucks, mixed vector/raster mode with layer styles doesn't really exist, layer management is barely reaching parity with where Photoshop was 20 years ago, and overall it has consistently left me frustrated and unable to get into the flow - you have to work around its deficits rather than being able to compose its features in novel ways.
Being able to maintain several 'versions' of the same image just by showing/hiding layer groups while still being able to edit & adjust things fluidly is a crucial feature in Photoshop, with GIMP you end up making copies of copies of copies and if anything needs adjusting or re-composing in the source layers you'll have to go back and perform the same sequence of steps and discard most if not all of the layers you made.
Yes, you have 'make selection from curve' and other tricks in GIMP to maintain your masks as vector data, but it's an awful lot of trouble and is a laborious N step manual process after every change.
Layer styles were basically CSS, for complex raster data, and it works beautifully.
Literally every modern graphics workflow.
The copy/paste flow in particular still trips me up.
Pasting creates a "floating layer". A "floating layer" usually needs to either be "anchored" (merged down into the layer below) or converted to a "new layer". In the meantime, for example:
* If the move tool is selected and the "floating layer" is clicked, the "floating layer" is draggable on that click. * If the move tool is selected and the canvas is clicked outside of the "floating layer", the "floating layer" is anchored and undraggable on that click. * If the unified transform tool is selected and the canvas is clicked anywhere, the transform frame is activated for the "floating layer" and it is rotatable on that click. * If the erase tool is selected and the canvas is clicked anywhere, nothing happens, except that the UI flashes. * If the paintbrush tool is selected and the "floating layer" is clicked anywhere, it "paints" – though pixels are only affected within the bounds of the "floating layer", as that's just how things work in GIMP.
I'm sure there are perfectly sane reasons that things ended up like this, but the result feels weird to me. I don't really want to have to think about "anchoring" a "floating layer" after pasting it, much less think about what the UI is capable/incapable of doing while there's a "floating layer".
I've since moved on to Krita. It feels far less weird.
- multi-select in some (admittedly non-standard and treeview-like) listbox... How did such a "feature" require 7 years to implement?
- switching between dark and light themes with a single button... Doesn't every UI library implement this as default by now?
Also, refactoring file I/O completely so that it abstracts away path separators, which, apparently are a real headache? I mean, yeah, other toolkits have had these abstractions for years. Or they could just convert everything to forward slashes which also works on every OS...
When reading changelogs like these, I keep wondering how Gimp (and to some extent Gtk) has been able to work as well as it does for so many years
How often do you need to multi-select items on a list box? I don’t I’ve ever found myself needing to do that in the many years I’ve been a GIMP user.
> switching between dark and light themes with a single button... Doesn't every UI library implement this as default by now?
Which UI libraries do that?
> Or they could just convert everything to forward slashes which also works on every OS...
...except Windows.
In other picture-editing software I use it all the time to select multiple layers to move them at the same time, or to show/hide multiple layers all at once
>> Or they could just convert everything to forward slashes which also works on every OS...
> ...except Windows.
Wrong. I have no idea where this comes from, other than Linux-fanboys re-iterating the same old story from Windows 95 over and over again. Windows has supported forward slashes as path separators since Windows 2000 (at least).
Fair enough. When you said "listbox", I assumed you were referring to generic listbox widgets as you might find in a filter/settings window.
> Wrong. I have no idea where this comes from, other than Linux-fanboys re-iterating the same old story from Windows 95 over and over again. Windows has supported forward slashes as path separators since Windows 2000 (at least).
It comes from the fact that a lot of native win32 APIs do not support forward slashes, and Microsoft's own documentation says that you need to use back slashes, not forward slashes (example: https://docs.microsoft.com/en-us/windows/win32/api/libloader...).
Most of the filesystem APIs have a preprocessing step that, among other things, can convert forward slashes automatically. However, this is not always available or desirable, and sometimes it can be disabled by an API without you realizing it. Not using the native path separator will cause your path handling code to be fragile, especially if you need to handle arbitrary user-provided paths (as GIMP does).
Also, many (all?) Windows software with a command line interface use the forward slash for arguments. For example, on a unix you may invoke a command like `ls -a` whereas the equivalent on Windows would need to be `ls /a`. Having forward slashes in your paths can lead to problems.