I created this post to talk about Gimp's meta and how the software can be improved. I will be citing a another open source GPL licensed image editor (Pixelitor) to make many points on things it does right that Gimp doesn't do at all. My first controversial opinion is that new filters are NOT that important. (with exceptions)


How to improve Gimp

In my opinion, outside of things like N Point Deformation, Paint Select Wand (up and coming) and an intense GEGL based anisotropic filter . We don't really need new filters or tools for Gimp. My third party GEGL filters, other people's python and script fu stuff, GMIC, Pixelitor and other background design software I use does everything. Thousands of filter combinations exist.

If Gimp's meta is going to advance we direly need things like speciality layer types. Adjustment layers, re-editable smart filter layers, shared layers with external apps, clone layers and path layers. As well as a way for GEGL to control the creation and movement of layers and for GEGL filters to spit out their underlying graph info to aid in automation. That is what I think Gimp needs to hit the next level. I definitely brought Gimp into the future by making all these GEGL filters but now I don't think the meta has much to do with filters and it has more to do with needing speciality layer types and easier non syntax and script fu based automation.

Re-editable GEGL filters with special layers, shared layers with external apps, and a Curves filter that dictates to all layers below it is mandatory for Gimp's success and we are failing because we lack this. Pixelitor and Krita have all of this but gimp has nothing like it. Pixelitor's non-destructive editing is something that the Gimp team promised for twenty years and failed to deliver. The main problem being Gimp's limitations and not the difficulty of programming adjustment layers or smart filters. Gimp and GEGL were never compatible to begin with. That is why it took about twenty years to integrate the two. Gimp devs are waiting till the Gimp 3 port is complete before they begin work on non-destructive editing and other fun things. They have no hints on how this will work other then it having something to do with a hidden GEGL graph. Will they make smart filters, adjustment layers or something totally unique like nodes only? We don't know. We can only wait and see and hope they don't screw it up. I have pessimism in the official Gimp team as they only seem to care about photography and not graphic design and thumbnail card and banner creation. They don't view Gimp as text styling software; They view it as photography touch up software. Second problem, There are people I like to call "denialist" as they assume Gimp doesn't need non-destructive speciality layer types for success as they find shame in copying Adobe. They assume the lack of adjustment layers makes Gimp different in a good way. (Like a Blue Ocean approach) They are wrong. Gimp needs to follow Adobe on things that are tried and true. One of the core reasons Gimp has such a bad reputation is because it lacks non-destructive filters.

In the end there are only four things I can recommend Gimp users to help improve open source graphics.

1. Use other free and open source software to collab with Gimp. I strongly recommend Pixelitor, Pattern Monster and open source AI's like REMBG for background removal and GFP/ESR Gan for photo restoration.

2. Donate to Gimp/GNOME foundation and request this goes towards Gimp's future of non-destructive editing.

3. Learn Gimp's GEGL Syntax as that currently is the highest point of the current meta. Visit GEGL.org to study it. Maybe make a GEGL filter

4. If you made art, design or whatever using open source software let people know it was made using open source.

That way open source graphics can be as competitive as possible against industry SaaS stuff like Adobe and PhotoPea.


TLDR - We need non-destructive editing and automation - not so much new filters.

I'm always quick to prop up FOSS when I use it. In fact, that's all I ever use these days. I used to use Adobe products, but have no need for it anymore.

Gimp is pretty much the only image editor I use anymore and I think it is very good as it is already. I'm open to using new FOSS, but have yet to see anything better, including Pixelitor. Pixelitor is nice, kinda reminds me of the old Ifanview program in Windoze. It definitely has it's place, just not with me yet....

Pixelitor's experimental features have smart filters, vector layers, shared layers and in the latest beta build it has clone layers and adjustment layers.











This is the stuff Gimp needs to compete. Pixelitor can collab with GImp so the way I see it they go hand and hand.

GIMP 3 promises to be both a lossless (hate that term; I like using self documenting) editor as well as allow for nodel editing. Would really appreciate a node based GUI interface. Heard rumors a few years ago that some folk are trying to do just that but then no more news. That's what I would appreciate. No need to code for filters if you have a node editor since you literally cam make your own filters via nodal editing.

Exactly what i love about Filter Forge so much.And from FF (and Davids genius) came Rodilius.

Gimp 3 is is literally just the GTK3 port and nothing else, but it will break most python stuff. Non-destructive editing will come after Gimp 3 in the 3.2 series. GEGL filters in Gimp 3 (outside of paint select) can be ported to 2.10.

GEGL Graph is a direct access to GEGL's node based system. Gimp 3.2 likely won't use nodes but rather it will do what Adobe's smart objects and adjustment layers do.

Not "most". All. But you don't need such an extensive rewrite (just ported my first script to V3).

Pessimistic post incoming!

As we enter the early 2020s, Gimp is still stuck in the early 2000s due to not having any form of advance layers. GEGL is extremely powerful but Gimp is barely tapping into its potential. Gimp forum users are making tons of python stuff that will break in Gimp 3. Gimp 3.2 won't happen till the late 2020s and won't have decent non-destructive editing until 2030. Meanwhile all the commercial image editing is going to be non-free and monopolized by Adobe. Everything non open source is going to be cloud only and in the hands of the evil scary Adobe corporation.

So, ask yourself, what can I do to help? Ranting in all available Gimp forums that developers don't read anyway is useless. Gimp development is an ergocracy, your voice is only as important as your contribution. Gimp development doesn't need columnists, it needs programmers, testers, doc writers... Gimp developers don't need ideas, they need manpower.

And if you don't like the way Gimp goes, you can make a fork, that's the beauty of FOSS.

I did contribute. I pioneered GEGL's meta and I am eagerly waiting for more third party GEGL developers to show up. We can make GEGL into a GMIC 2.0 if we get lucky.

I do like GEGL's smooth by domain transform preset and have been using it instead of G'MIC's Bilateral filter for convolution purposes. GEGL is powerful for sure, but I tend to be just a user, contract_.

What we need is a way for GEGL inside Gimp to read GMIC strings like
inside GEGL Graph.

That is what will take the meta to the next level. So they can be non-destructive when 3.2 arrives.

Today is one of those days where I am upset just thinking about how Gimp doesn't have any non-destructive editing or special layer types in the year 2022. That alone is holding us back and no amount of fancy new filters will matter. We already have virtually everything (outside of like N Point Deform) so like hopefully someone new joins the Gimp team and adds smart filters and adjustment layers as I am losing my cool over this. lol

GEGL already has adjustment layer and reeditable filter properties in the graph. For example the color-overlay of a filter is just like an adjustment layer and chaining filters in a graph is like a smart object. The problem is Gimp does not save any information on what GEGL is doing and just rasterizes the entire GEGL Graph.

And what prevents you from doing so?

This said, you aren't responsible for Gimp's future. If it sinks, tough luck, but nobody will hold it against you, and there will be alternatives. If you want Gimp to go somewhere, provide code that helps going in that direction, or fork it and see how popular your fork becomes. The Gimp developers don't need people whining in forums (they are aware of Gimp shortcomings), they need help.