What ArtCraft's Vibe-Coded Apps Say About Adobe | Hacker News Reader

What ArtCraft's Vibe-Coded Apps Say About Adobe

tedium.co

30 pointsspeckx66 comments

> Is an open-source vibe-coded creative suite really worth cheering on?

No. They suck the oxygen out of the open-source ecosystem for other more credible software, such as Photopea or Gimp or FreeCAD.

Photopea is proprietary, but thats a nit.

I'd like to see how far echelon takes it. The current iteration is somewhat unusable, but there's room given weeks and enough tokens to get this into a usable state, and I'm interested in nuking the entire Adobe business model from orbit.

And haven't we had enough decades for GIMP to get air? I'm actually a big fan of it and it's my primary image editor, but the amount of hate it gets on this and other sites is incredible. Everybody just complains that they want it to work like Photoshop, and now a guy is making that real and we say it's taking energy away from GIMP. That doesn't track for me.

> I'd like to see how far echelon takes it.

Fascinating, I didn't know one of our resident "IDEs are obsolete and I never use them" commenters was demonstrating the complete uselessness of their mindset in production! Usually the vibe coders are cagey and never share their amazing 10x applications. Now we see why.

> but there's room given weeks and enough tokens to get this into a usable state,

There really isn't. I've certainly been down this road trying to understand the hype. The more complex the application gets the more the models will buckle under not having a solid foundation to work from, applying hacky patch after hacky patch. Simple changes will take 30 minutes, break three other things because the tests they write are trash, and introduce new edge case bugs. We've been hearing about how software development is over with every model release for the past two years, with the biggest upswing being since last Nov, and we're still hearing "just a couple of more weeks". When do we face reality?

I understand your skepticism. I'm actually pushing for this project to go as far as it can so that we can have the "facing reality" discussion, and see whether it completely collapses under its own complexity or if something usable comes out the other end.

>I'm interested in nuking the entire Adobe business model from orbit.

I really do wonder what will happen, is there any chance they manage to adapt instead of disappear? They can definitely afford more tokens than any random open source project, so maybe they'll have to expand their offering and reduce their prices, but that would probably only kick the can further down the road, and that's just Adobe, what about even more expensive software like Solidworks or Ansys?

<< I'd like to see how far echelon takes it

This will be the real test. I am an amateur, who needed some sprite editing, video edits and music changes. And I didn't use artcraft ( I might now though ) -- I vibe coded my own ( which brings me to another the other part upper echelon may be slowly realizing: they are not needed ).

> "Everybody just complains ... and ... we say it's taking energy away from GIMP."

It's almost like Hacker News is a giant group of people, and different subsets of people comment on different subsets of articles (aka, the Goomba Fallacy).

If only there were a commensurate number of people praising it, then I might not have framed it that way.

I would like to think that too, but the illusion was shattered in the past few years with AI comment spam and reddit-style injection of politics and astroturfing.

Why would they suck the oxygen out of the room? This sounds like claiming a new distro of Linux made by new rufugees from Desktop Windows for people accustomed to Windows desktops is harming Linux by ensuring they don't go to some existing flavour which defaults middle mouse paste and thinks everyone must go on a spiritual journey of becoming one with cli tools in order to set up their social media and documenting editing machine. New people are making something they want and are empowered to make something they want, not something we've all been grateful for but bemoaned as a comparise.

For example, I love GIMP but holy shit it was a downgrade from Adobe in most ways.GIMP should stay GIMP, though, I wouldn't want my tools being completely overhauled in terms of UI and workflow anyways. I'm sure the old heads who put TLC into it would be strongly against someone using AI to significantly rewrite it at base, but also changing the interface and how it works significantly enough to matter to users. It's a sad way how it's been gone about, but in a sense this new software is building on GIMP (among many, many others) shoulders by developing openly and allowing the lessons on how to do things be adopted and not re-learned.

One program or the other is better - and that’s the end of it, no? As always.

Also Krita

It's been 20 years, open source more usable for typical use case but gap/ecosystem expanding.

I don't know. As others mentioned, Gimp and others had plenty of decades of oxygen to breathe. Meanwhile, the software ecosystem keeps getting worse - that's a trajectory it was on before LLMs, so AI is not to blame here.

I find such vibe-coded efforts to resonate with me, because they reflect the same impatience I have. Impatience with enshittification of the commercial offerings, impatience with OSS following suit without realizing it (most modern UI/UX philosophy is driven by needs of enshittification), impatience with OSS being dominated by fakers and itself becoming a business model. Impatience with broken software.

Vibe-coding can be a way to vent the frustration, and in this case, it even turns into a spectator sport. Most of the vibe-coded projects I do, and that I see people I know doing, share the same, single, deep motivation - "I want a software that does this particular thing right now, and without all the bullshit". The vibe is that of awe, but it starts rooted in anger.

If those long existing open source tools were good everyone would be using them

Gimp is a joke. Anyone who has used Photoshop for decades turns to Gimp, uses it for about 10 minutes, and then realizes this is is a disaster and deletes it forever.

Gimp is terrible though and they've had ages to change their UI/UX but they won't.

I see plenty of posts from authors who didn't even bother trying any of these clones.

It seems it's buzzing because it make people who post about it, buzz.

We live a strange time.

People are buzzing about it because it is bold experiment whose outcome may tell us extremely interesting things abut vibe coding with current-generation tools. Most particularly, can you, or can you not build large applications with pure vibe coding? And lots of speculation over whether architectural rot will be better or worse than what humans do. Because human-generated architectures rot out too (Photoshop being an excellent example). I not aware of anyone, so far, who has explored that issue. So far, the grand experiment seems to support "can you not", but the experiment isn't over.

Fascinating stuff. I'm glad somebody else is doing it, and happy that I'm not doing it. Yet. But I will certainly be watching with interest.

I am a professional coder with decades of experience. I use AI regularly, but so far use it very tactically. The big challenge I struggle with at the moment is whether I can trust current-generation models that seem to be increasing in power exponentially at the moment with big tasks.

--- Just to qualify for your arbitrary impressionistic standards on who gets to say what about ArtCraft....

Photoshop: Used it for years. I have a perma-license CS6 copy that runs on Windows only. I sometimes (rarely) boot up my Windows partition when I have particularly complex edits to make. A UI that has rotted out over time. No way I'm paying the extortionate rental prices for current-gen Adobe tools. Their entry-level software seems to be toys, and are not usable for anything serious.

Corel Photopaint: I worked for Corel. I actually really like PhotoPaint, which at least follows a Windows UI convention. Much easier to use than Photoshop; but I have not used it for many years.

Gimp: tried it. Rotted out UI, and an absolutely incomprehensible UI convention (if any). I cannot use it and will not take the time to learn it. On Linux, I very occasionally use Gimp, but I loathe it.

Krita: So much easier to use than Gimp, excellent for consistent and modern UI conventions, but currently missing important features.

ArtCraft: I have not actually tried it, although it sounds very much like what you would expect from vibe coding: buggy as heck, feature deficient, and definitely not ready for serious consideration. Yet.

To be honest, I do most of my graphics work in Inkscape these days (lots of vector icon design). Brilliant UI, consistent UI conventions. Very usable. But mostly useless for photos.

I asked precisely that question on their discord. An account of token spends.

Noting that this project is examplary experiment that may demonstrate whether the state of genAI can produce complex software before an explosion of token consumption as the code base grows, or even collapse.

In my experiments, it collapsed, each time I tried. Every 6 months or so I take the challenge. Unless a human intervenes of course. Basically dictating what genAi should do, which becomes more painful than it's worth automating code generation.

I've actually been more impressed from a marketing standpoint then by the product offering. This whole thing has been a masterclass in promotion.

That being said, as somebody who views Adobe products as borderline spyware and STILL rocks Adobe Photoshop CS6 (the last non-SaaS version), I'm not exactly going to shed any tears of sympathy for them.

The audacity of this is what catches interest. I've first learned of it from a repost of a repost, framed as "this guy is trying to speed-run getting sued into oblivion". Then I saw who the suing party would be, and I can't help but cheer for the vibecoder.

To actually test them properly you need an understanding of Adobe's products. I only know the very basics of Photoshop - I'm guessing most are in the same position

I'm comfortable with several tools of the suite. So I went to the repos, the README felt odd. I checked the issues, and decided it wasn't worth installing.

I do want some open source Adobe suite clone to work. I just have far more fate in tools like Krita that don't give me a feeling of being fooled and sold a toy even if it's free.

In fact the open source project is there for profit as the easiest way to install it is to go to their website, where they also offer AI credits to use with those tools from $8 per month to subscriptions far more expensive if you need more credits.

Is it just me or is it really hard to get a complete/working product out of an llm?

I find that unless I have very detailed specs with rigorous completion verification/criteria (manually atm) or tests. It’s like 60/40 at best.

And adding all the tricks only gets me to like 75-80.

Tbh I am using cheaper Chinese models (ds4.1f) and only pay for the $20 plans from the us labs

Rigorous specification doesn't even help, the rigority of a good spec requires too much context for LLMs to adhere to and they will flagrantly violate clear specifications once you have too much detail. The only thing that does work is having specification encoded as tests, because LLMs can fix compile errors in a loop and that deterministic guidance is what they need to keep them on track, but... writing tests is coding. Writing good tests still requires understanding the code and architecture in fine detail.

This is why you see all of these headlines about porting things that already have years of human work as deterministic guidelines and yet absolutely nothing worth using in production is being vibe coded from scratch. Don't let the hype/FOMO get to you, you are not alone, that is just what the people selling you the hype want you to believe.

I wonder. My older boss was of the type that even if something wasn't 100%, his position was that 80% was better than nothing ( which offended most of our team, who had to deal with the inevitable support and fallout from his decision ).

lol. I used to work for that guy too! (Can there be two software dev managers on the planet who are that stupid? I pray that there are not).

I use only OpenCode free models and occasional 20$ sub from OpenAi or Anthropic. And get complete/working products out of them.

Examples? Advice?

Tell the model to make thing - model makes thing. If something dos not work ask to fix.

Adobe slept so incredibly hard on tools like Photoshop. They have been keeping that stack on life support for decades without any real innovation (Linux support?) and hostile dark patterns around subscriptions and cancellations. If we could peel the covers back and inspect the source I am sure it is as antiquated as a Pentium 3. They deserve to be supplanted by something else for languishing such a legendary tool. Even if it is vibe coded.

The crazy thing is, I’m not even opposed to paying a subscription for quality software that I get a lot of value out of, but increasingly Adobe CS is not that. For over a decade now they’ve been cramming in features for features’ sake with little regard to actual customer needs.

I would not hesitate to subscribe to a version of Adobe CS that is more or less feature-frozen at CS1/CS2, with all engineering effort and subscription money going towards bug fixes, optimization, adaptations for modern OSes and hardware, and code correctness. Can you imagine how insanely good Photoshop CS2 would be today if they’d spent all the time since its release working on those things instead of endlessly tacking on mostly useless gimmicks!? Its stability and performance would be a moat unto itself.

I still regularly use the Photoshop 7 I bought in Community College in the late 90s on my modern Windows 11 machine and in a VM on my Mac without issue. Its got 99% of what the newer version has for what I use it for.

I could fairly comfortably drop back to PS 6 without missing a whole lot. It’s been a complete product for a long time.

The first version I actually purchased was CS2! Over the years I have probably given Adobe like $3,000. For Photoshop. Joke is on me, honestly. I agree with you - I would gladly pay for good software. That is what surprises me so much about Adobe. They think their shit doesn't stink. They think they have it made. So they continue to drag ass on innovation, keep prices high, etc... If they released a Linux stack and kept their pricing sane, without hostile patterns around subscriptions -- they would continue their dominance for years to come. But that is certainly coming to an end when an enterprising individual can sit down with a stack of tokens and rebuild their software in a few days or weeks.

As of today, just the Adobe Reader is 3.5GB on disk. So they're doing something, but I don't know what it is. A similar pdf reader app is around 300MB, and Foxit Reader is 950MB with more features.

Last time I installed photoshop the core application was relatively small, it's just that they cobbled on a few gigabytes of locally-stored tutorial material for... I don't know why.

The author of this project is pushing and promoting it HARD, on HN and elsewhere. It does not appear to be even remotely at feature parity, but they write about it as if it were finished and working.

IMO it is vaporware until shown to be otherwise.

Parity can take time - and you probably don't want it because the apps like Adobe's are full of so much legacy anyway.

> and you probably don't want it because the apps like Adobe's are full of so much legacy anyway.

I'm not an artist, but isn't that what people are interested in sticking with the incumbent? See: google sheets vs microsoft excel.

Not only that, I tried the Adobe Reader counterpart and there sooo much things that look botched, eg. OCR.

That's what this post is about

> A quixotic new project takes an AI-first approach to battling Adobe, and while it’s unclear where it leads,

Did you try the downloads?

I tried the wasm version that simonw kindly hosted for them... https://static.simonwillison.net/static/2026/photocraft-web-...

The very first thing I tried to do - create a new document at a specific resolution and DPI - didn't work. It completely (and silently) ignored the values I entered.

It's still impressive in some ways to be sure - but I agree the marketing I've seen for it suggested to me that it is more done that it appears to be. It's the common theme with LLM developed that marketing is pushed far too early in the process.

So when I went through all of them last night and found a surprising amount of functionality, especially in the Photoshop clone, I was just imagining it?

Really?

I tried the Photoshop one when they promoted it a few days ago here. I loaded up one of the client projects I'm working on right now and I could barely move the document or do anything it was so laggy and every single button/feature I tried to use had some kind of bug.

Maybe they've fixed everything in 2 days but I kind of doubt it.

This is the kind of thing that is very easy to validate. Boring, but very easy.

Just revisit this discussion in 6 month and 24 months.

You're saying a complete reproduction of the Adobe suite made in days isn't at feature parity yet? Well I'll be.

>and promoting it HARD, on HN

From the other thread he said he wanted to hold off posting it here until it was more ready.

>IMO it is vaporware until shown to be otherwise.

You can download and run it today.

Honestly, I'd be surprised the Adobe or current workflow even survives in a few years with people who grew up on image prompting and sketch to image etc.

But it'd be nice to someone to throw enough tokens to make some parity Adobe software... like OpenAI wasting 10m on 1m math problem poor waste of resources... spend 100m and kill and charge $10 instead of adobe $50.

Pouring what would be adobe subscription $$$ into inference instead.

While we speak, these new apps only get better.

Clever.

I learned photoshop long before programming and the idea someone will vibe up not just photoshop but all the apps is absolute ballocks.

It isnt that there is no market for a cheaper version but people forget the amount of craft it takes to get something from 80% to 99%.

If all you can add is baby sitting agents, then you will never reach beyond 80%. Yes, spend 10m on tokens and 2 years improving and it may become useful but isnt that how it always was? Ela figma and canva?

Startup spending a few million dollars to decompile and replicate adobe suite in a few years would spring up so many competitors that it would basically kill adobe and creative software sub. That's at least a more useful allocation of tokens than logn -182. Porbably prosocial movement.

"Nothing. They're bad apps, coded by an amateur, offering no new ideas and only the most obvious subset of functionality."

Here's the thing, I suspect there are a lot of people who pay for Adobe tools who are in fact only using "the most obvious subset of functionality" and would love to have a replacement that suits their needs without the hefty subscription price.

So even if these tools never reach 100% parity with Adobe, if they offer the basics, and they're free and open source, it will be a win.

Joel Spolsky has a famous post about this:

=-=-=-

A lot of software developers are seduced by the old “80/20” rule. It seems to make a lot of sense: 80% of the people use 20% of the features. So you convince yourself that you only need to implement 20% of the features, and you can still sell 80% as many copies.

Unfortunately, it’s never the same 20%. Everybody uses a different set of features.

https://www.joelonsoftware.com/2001/03/23/strategy-letter-iv...

=-=-=-

Also, there are many image editors _already_ out there of varying levels of power and ease of use. Some open source, some commercial.

And yet Photoshop still rules.

As a power user of Photoshop, I will say yes, most people use about 3-10 features that adobe offers.

But there are thousands of free tools for those people already, the point is this guy is saying he cloned the entirety of the adobe creative suite with vibe coding in a week. That's not true. He made a bunch of simple and buggy "creative suite flavored" apps. That's a completely different thing than what he claims.

I'm sure that is true. On the other hand, the majority of their customers are likely using these tools for work and larger organizations are paying for the licenses. People in school for the fields that use these tools are probably being trained on Adobe products, with some discount for the school and the students. I don't think these vibe-coded apps are going to get much traction, there's a whole entrenched ecosystem around Adobe products.

My distaste for vibe coded apps aside, I wish them the best of luck.

It'll be interesting to see how it plays out, that's for sure! Adobe is down like 60% over the past 5 years (after a massive run up). So they're going to have to figure out something after the failed Figma acquisition + increasing competition from vibe-coded apps.

I think Adobe is in sort of a rock-hard-place situation, where they can't change their current apps too much because users are so invested in them, they can't build new apps without pulling resources away from their main apps and creating confusion, and they were denied the ability to just acquire their primary competitor.

Their best strategy, so far, has been to create new "mobile" versions of their apps where they can iterate faster.

I suspect the reason alternatives haven't taken away marketshare isn't because of the software itself - it's because of the industry around it treating it as the standard. That's not going to change with a vibe-coded alternative

You're free to use Gimp today. Almost zero professionals actually do.

And those free alternatives have existed for years

Here's the thing though, those "alternatives" don't always hit the mark. I've sure Inkscape and Affinity's vector programs do a fine job. I've been an Illustrator user since the 90's so... the shortcuts, the workflows are all burned into my brain. I truly enjoy it. If Adobe had a Linux version, I'd gladly pay the 20 bucks a month for it. But the best we've ever gotten is Affinity allowing some shortcut editing and it partially working under Wine - as long as you aren't using HiDPI.

I downloaded this on one of my Linux laptops this morning. Immediately I did some quick editing/drawing and was like, "well, this completely fits my needs for SVG editing now".

This is where I'm at. I'm not even a supporter of vibe coding and I have serious issues with overwhelming reliance on AI in software development. That said, the Illustrator alternative from this suite of apps is the best vector editor on Linux that I've ever used. I will only remain a user of this software if Affinity doesn't end up porting their app to Linux natively, like they are rumored to. Affinity's workflow is so much easier to use than the Adobe's, and I'd switch in a heartbeat from this vibe-coded suite.

Why would these users go for i.e., gimp? Following this logic gimp is a better choice.

Reply on news.ycombinator.com