The photoshop relight feature - what it means for designers

[JulianZI]

But in the long-term, with AI getting better at coding, we'll get to a place where the whole design ecosystem is simply too poisoned by AI to salvage. In the end, Adobe snake will bite its own tail when it has nothing left to feed on. For that reason alone, I wish the jerks at getartcraft well.

That guys who created a business with generative ai which is based on the work of artists who were not asked if they want to contribute are now the robin hood against big bad A? What they are doing is not praiseworthy at all, it is just cheeply done AI content which burries everything human written, because it is just a big thick layer on top of everything.

I find it sad that here at PL nothing seems to happen anymore. Now would be the time and the chance to get a 90€ Claude-Code subscription and use it to optimize the old PL code, restructure it and create a new, innovative product with native Linux support and an UI which feels to be made for the users.

[shijan]

Well, i asked Martin Huber few weeks ago and he answered:
"We're mostly focused on fundamental tasks such as documentation, restructuring, and investigating what needs to be done to ensure that PhotoLine continues to run properly on Windows and macOS in the future."

[Vivi_Ram]

I'll confess that 2 years ago, when I first began using PL, after Affinity went to the dark side, I was very frustrated at the UI. I found it very different and very hard to use. The lack of artboards was crippling. And to this day, I find the text bounds in PL very odd. But after 2 years, I have made friends with the UI, and come to terms with certain limitations. So I have stopped complaining about UI. Others will remember my frequent suggestions to change things around a bit. However, I'd love for the devs to bring PL up to speed in terms of bit depth support for masks and layer styles. They're unfortunately non-negotiable. I do also miss calculations panel from PS. Most people might not be aware. Calculations existed before Layers! They're that old. Herbert123 already showed how they can be manually done. But Calculations are still mightily convenient and very quick. But I dream!

To your point about using Claude to optimise PL. I am extremely opposed to the idea. In the extreme. Why? If you refer back to my original post, you'll see that https://getartcraft.com/ - this project was essentially reverse engineered from Adobe apps. I recently saw a video of a guy running these apps. They're actually very good for something built entirely by AI. Adobe has already taken cognizance of them. I am sure a lawsuit is on the way. Or a cease and desist to start with.

Allowing Claude any type of access to the codebase is akin to handing the keys to your house on a velvet pillow. It is certain to leak IP. IP that the Hubers have built over decades.

[shijan]

If believe to Antropic info, as for today Claude AI writes more than 80% of the code merged into its own codebase
But using AI coding may not a easy problem fix, especially for classic era coders. AI code could be written in strange and not logically optimized ways and so became very hart to read by humans. So coder became babysitter person who checks everything and fix AI mistakes instead of real coding. But i am not a coder and don't use AI, so it is not my personal experience, i just hear this somewhere.

P.S. Currently just for fun i look at the NegPy film negative editing application development on GitHub which actively use Claude, and to be honest it feels like some endless never finished bugfixing soap opera. It was started few months ago i guess. App is rather complex and multifunctional so every few days they release new beta version with huge amount of bugs and fixes and changes that hardly possible to understand from description in every release. But same time, until i report to them, all that time they didn't see very simple but fundamental problems, like totally broken color management which makes app totally unusable.

[bruce1951]

My 2 cents worth on the PS v PL debate.

The way I see it PS is going down the "you pay to let Ai edit your work". That's not a future I look forward to.
My wish for PL is to give me a solid platform that is capable and reliable. On top of that I want the PL platform to allow me to use plugins and external programs seamlessly. If I want an Ai edit then I want PL to allow me to skip to my chosen software and bounce back to PL.

If PL started to go down the Ai rabbit hole then I think I would stay with the current version and stop upgrading. But that's just me.

[Hoogo]

AI coding is developing very fast. A year ago it was "vibe coding" vs "write requirements", and it seems today it's more "be teamleader of a bunch of agentic coders, reviewers, testers, give guidelines and distribute work".
The given examples here may be buggy, unfinished and full of problems, but the time pend might be measured in man-days or -weeks, not hundreds or thousand man-years Adobe needed for PS.

[Vivi_Ram]

There's a problem with thinking that we could simply Ai code the bejesus out of a program as complex as Adobe or PL. Adobe, when it first began was solidly backed by academic research. Adobe holes 7000 patents approximately. It is not as simple as prompting to claude to "give me a photoshop". It is not just about making individual features work, but every feature work with every other feature properly. My favorite example is vector morphing. At the outset, it is fairly simple. But then, morphing has to work with objects with styles, feathering, etc. It has to work with transparencies. It has to work with complex fills. It has to work with text. It becomes exponentially complex as you add elements to it. This is why Corel bought Xara originally. They simply couldn't compete at the time. And so they bought the marketing rights to Xara, a far superior program, and they simply sat on it. Hoping it would die off. CDraw was the best damn vector engine at the time, written to work on 4 MB BBC Micro RAM. The original code was written by Charles Moir and team in extremely lean x86 Assembly Level code. Can Claude do this feat? I doubt it. I'm sure PL has equally optimised code that allows them to ship the whole software in a 30 mb package. Ai simply cannot do that as it currently exists.

Edit: Charles Moir wrote Xara first as ArtWorks for ARM Acorn Archimedes which ran on Risc architecture. the design world has come a full circle with Apple going to ARM.

[bruce1951]

Just an off topic Ai comment.

The media, and politicians, are setting thier hair on fire about Ai and its so called ability to "hack". "The world is doomed and Hal will take over". Yeah my logic says if Ai is smart enough to "hack" then it's smart enough to prevent "hacking". Yes/no?

[JulianZI]

There's a problem with thinking that we could simply Ai code the bejesus out of a program as complex as Adobe or PL.

I am sorry, that has changed.
The only moat now are AI modules and patents, and of course a community.

[Hoogo]

Just an off topic Ai comment.

The media, and politicians, are setting thier hair on fire about Ai and its so called ability to "hack". "The world is doomed and Hal will take over". Yeah my logic says if Ai is smart enough to "hack" then it's smart enough to prevent "hacking". Yes/no?

Difficult...
Yes, AI could also work as "white hat" hacker, hacking to report and fix loopholes and bugs. The public release of Fable 5 was delayed, so citical software could be checked for errors and patched.
But: The masses of reported bugs was overwhelming, it takes a lot of work and time to verify and fix. Were new bugs introduced? Were ALL problems found? What about other AIs, would they also be used for white hat hacking first? Surely a lot of problems were fixed that way, but also surely not 100%, and AI is still learning and improving.
And what about all the webcams, old phones without support that won't be fixed anymore?
Finally, there is also the bad side: Developers flooded with AI slop bug reports that waste their time.

After all, I see the hackers having an advantage: Using a found exploit is fast, but fixing, testing and deploying it takes time.

[bruce1951]

Hmmmm. I'm an Ai sceptic. Sure it's real and the future. BUT I'm of the belief that the current "hype" is the result of ignorant politicians and money chasing investors. Remember the Tech crash?

As for PS and other editors using Ai. At what point does Ai end? What is Ai "perfection"? I just hope, selfishly, that PL doesn't go down the Ai rabbit hole.