I've occasionally looked around for a node-based image editor (á la Blender, but for 2D), and I've only found simple proofs-of-concept.
When discussing Photoshop alternatives, I often find the lack of smart layers and other non-destructive editing to be a painful gap; this is a bit of a paradigm shift towards the other extreme.
johanvts•May 23, 2026
How does it compare to graphite.rs ?
vessenes•May 23, 2026
I can’t tell if I love or hate the idea of this.
On the hate side, ComfyUI is just so, so difficult to use on a normal size screen with a trackpad. It’s designed for someone with a 34” gamer monitor and a mouse with like six buttons, and I haven’t seen a good working node based interface that would be comfortable on a Mac or iPad, so I feel frustrated just looking at the images and thinking about zooming in / out and arranging the nodes.
On the love side, everting the workflow into the main thing is really interesting and clearly a thing people who do graphics in production need. Photoshop has a history palette, but it just does not do (easily) what this lets you do, which is be process first, and automate the process.
Anyway, not for me I think, and I’d like to imagine there’s a better UI waiting to be developed to do some of this, but I think it’s cool and interesting to see new ideas in graphics production.
sorenjan•May 23, 2026
What do you consider to be a "normal size screen"? Since you mention trackpad I assume it's a laptop, so 13-15 inches? That was considered a normal size screen in the 90s, I don't think we should consider the compromise that laptops are to be the norm.
stavros•May 23, 2026
It all depends on your vision, doesn't it? 13 inches was a normal-size screen when we had five pieces of information on it at 640x480, but now that we have 4k screens jam packed with elements, it's no longer a good size.
I'm not sure saying something like "we used to send letters to each other a hundred years ago so we should be fine without mobiles" is such an ironclad argument.
Daub•May 23, 2026
If Mari (texture painting app) and Nuke (vfx compositor) had a baby together it would be the perfect node based photoshop alternative. The brushes of Mari are insanely good and color editing on nuke is a dream.
CyberDildonics•May 23, 2026
The vast majority of what people do in photoshop can be done with node based editors that already exist. Most people edit photos and that can be done with roto shapes and manipulation instead of painting.
archerx•May 23, 2026
Interesting. I’m in the process of making a node based image editor myself so I’ll see what this does right and what points of friction still remain. The main reason I want to do is to make automating tasks easier, batch processing in photoshop is ok, but it could be so much better.
viraptor•May 23, 2026
This is either an extremely weird timing coincidence... Or someone saw the announcement/devlog of Plasma Studio and decided to vibe-code-front-run it as a paid offering. This page appeared 3 weeks ago.
Original video a month ago for the plasma studio which is basically the same thing: https://youtu.be/WlgrCqgnk-M
The Arc homepage is clearly vibe slop but with 75 nodes and backend supposedly coded in C, it looks to me like it would have taken a couple of months to get here at least, certainly not a release within 1 week of Youtube-person's mention.
It doesn't seem to be open source unfortunately.
jayd16•May 23, 2026
Is there more of a smoking gun than its a node based editor?
Most of the game industry tooling is supporting that these days. Unreal, and Unity but also Houdini, Blender, Maya...
Free and open source and, as far as I can tell, does everything this is claiming to do and more. It's part of our workflow for the game my son and I are making.
edit: minus the AI stuff
thih9•May 23, 2026
I like the terms of service; I found this:
> All of our content is carefully written by hand, no AI was involved during the process.
Node editing is great for certain pro workflows, but it’s not as user-friendly as programmers assume. And for pros, there are times when text-based scripts are actually easier to parse.
Source: I launched a node based video compositing tool 20 years ago and watched people struggle compared to the layer-based workflows they were used to.
weitendorf•May 23, 2026
This looks really cool!
Unfortunately since it's not FOSS and there's no information about the company/individuals behind it, or even a way to pay for it/get licensing information from your UI, there is absolutely no way I would download it as a binary and run it on my computer as you suggest. That is, IMO, incredibly sketchy
xbar•May 23, 2026
I have never been in the target audience for such software and I am not intending to change that.
However, I love it. Not because I think it executes on its promise perfectly nor that it feels safe to use. I love it because it knows what it hates: Electron; subscriptions; AI-first; single-platform; interpreted software; big, fat, sloppy files. I hate those things in my end-user applications, too.
I hope they figure out how to create enough transparency and trustworthiness to make a sustainable business out of it, because I want to think that such businesses are still possible.
unD•May 23, 2026
No multiple colorspaces support, alas, only RGBA.
It reminds me of https://filterforge.com/ which has been around forever.
virtualritz•May 23, 2026
RGB is not a color space but a color model.
sRGB or ACEScg are color spaces.
CMYK or RGB etc. are color models.
This matters a lot. And yeah, it seems this supports neither and it is unclear if operations happen in linear color light at a bit depth of at more that 8bit/channel.
If they don't, most of them will produce subtly wrong results.
11 Comments
On the hate side, ComfyUI is just so, so difficult to use on a normal size screen with a trackpad. It’s designed for someone with a 34” gamer monitor and a mouse with like six buttons, and I haven’t seen a good working node based interface that would be comfortable on a Mac or iPad, so I feel frustrated just looking at the images and thinking about zooming in / out and arranging the nodes.
On the love side, everting the workflow into the main thing is really interesting and clearly a thing people who do graphics in production need. Photoshop has a history palette, but it just does not do (easily) what this lets you do, which is be process first, and automate the process.
Anyway, not for me I think, and I’d like to imagine there’s a better UI waiting to be developed to do some of this, but I think it’s cool and interesting to see new ideas in graphics production.
I'm not sure saying something like "we used to send letters to each other a hundred years ago so we should be fine without mobiles" is such an ironclad argument.
Original video a month ago for the plasma studio which is basically the same thing: https://youtu.be/WlgrCqgnk-M
Devlog #1 https://youtu.be/JDsoKhgNtHQ
More design / timelines https://youtu.be/L1O2ALT0A14
The Arc homepage is clearly vibe slop but with 75 nodes and backend supposedly coded in C, it looks to me like it would have taken a couple of months to get here at least, certainly not a release within 1 week of Youtube-person's mention.
It doesn't seem to be open source unfortunately.
Most of the game industry tooling is supporting that these days. Unreal, and Unity but also Houdini, Blender, Maya...
Free and open source and, as far as I can tell, does everything this is claiming to do and more. It's part of our workflow for the game my son and I are making.
edit: minus the AI stuff
> All of our content is carefully written by hand, no AI was involved during the process.
https://pixieditor.net/docs/terms-of-service/
Source: I launched a node based video compositing tool 20 years ago and watched people struggle compared to the layer-based workflows they were used to.
Unfortunately since it's not FOSS and there's no information about the company/individuals behind it, or even a way to pay for it/get licensing information from your UI, there is absolutely no way I would download it as a binary and run it on my computer as you suggest. That is, IMO, incredibly sketchy
However, I love it. Not because I think it executes on its promise perfectly nor that it feels safe to use. I love it because it knows what it hates: Electron; subscriptions; AI-first; single-platform; interpreted software; big, fat, sloppy files. I hate those things in my end-user applications, too.
I hope they figure out how to create enough transparency and trustworthiness to make a sustainable business out of it, because I want to think that such businesses are still possible.
sRGB or ACEScg are color spaces.
CMYK or RGB etc. are color models.
This matters a lot. And yeah, it seems this supports neither and it is unclear if operations happen in linear color light at a bit depth of at more that 8bit/channel.
If they don't, most of them will produce subtly wrong results.