Remember that this topic is specifically about Krita plugins. The software you describe would never be on this forum as far as I can imagine since it’s a general human-computer interface. It wouldn’t be affected by this and there this is not a ban on what you can do with Krita. You still can do whatever you want with it.
Oh, so what it is is a plugin for krita. That allows a commandline interface (CLI) to interact with a lot of krita’s tooling. I haven’t expanded krita or assessed how much can be exposed, but as homage to Simon Willison’s Pelican on a Bike test of LLMs ( Pelicans on a bicycle ), the only drawing I’ve used it for is to draw a Pelican on a Bike. If people are comfortable with me posting that picture on here as an exception, I can demonstrate. Rather than it being a raster image generated from a diffusion or large language model, the LLM uses tools to control the krita API. So it moves the cursor, changes the paint colour, etc. As you would see from the pelican, it’s not particularly advanced (that’s quite possibly because of the operator, i.e. me), and unexpectedly I guess it demonstrates an intellectual question that I had, which is whether an LLM (gAI) could actually draw art, not just generate it.
I feel this is going into an entirely different direction. If you want to discuss if AI issues can be ignored for accessibility’s sake, I think it’s worth creating a new topic in General Forums > off-topic talks or join the AI related meta thread
I worked with disabled people for two years and I’m always interested in this topic and would like to see what others think.
On the site you linked you wrote yourself that the whole image was created by one single prompt. It wasn’t made by using Krita’s individual tools (by you through individual commands). This isn’t much different than any other generative AI except that it’s not generating the output directly. Basically just GenAI with extra steps.
I’m not sure if you couldn’t do the same with a normal eye tracking device.
Also stop posting links to the AI plugin. The rule is already in effect.
Yep, that was sort of the point, as a single composite test of a somewhat generic object, like with Willison. Interesting point though→suggests a perceived ethical threshold between individual actions rather than composite instructions.
As I wrote, this belongs in another topic. But ask yourself. If you ordered a Pizza online for delivery to your home, is it the same as if you baked the pizza just because you chose the toppings? The individual steps are important for having a pizza made yourself, as well as an artwork. An artwork is a sum of many tiny decisions. That doesn’t even touch the ethics of stolen labor. The opening post already gives an example of when an “AI” is allowed.
That’s enough off topic talk.
@doughnut As you are aware, this forum does not allow the posting of AI-generated art in any form or the promotion of any plugins or other resources to create AI-generated art.
However, a plugin for krita has been developed that enables a krita user to ‘make’ AI-generated images, if they have an internet connection. The developers of that plugin have their own website where you can download it and there are YouTube channels where its use is explained and demonstrated.
Those developers are free to do that but what they and its users can’t do is advertise or promote its use in any way here on this forum, or ask for help with it, because of the rules of this particular forum.
Similarly, you and any colleagues/collaborators would be free to develop any kind of krita plugin you like for any purpose you wish, using AI-generation techniques and to make it available to anyone via a suitable website and to promote it in any way.
What you wouldn’t be able to do is to make it available or promote it, or ask for help/advice with it, in any way, on this particular forum.
If you want to make the kind of plugin that you are describing then you and others can do that but you can’t expect the support of this forum for any aspect of it.
Thanks for your response. I’m not seeking to promote it, rather on reading your views I was wondering whether your preference is that I get rid of it.
I was explaining the situation, as I understand it, by way of comparative analogy rather than expressing a personal view.
As I’ve said, you are free to develop any such plugin. Krita is FOSS and it’s API documentation is available to the public and Python is a well known language. I like that freedom for people.
I have no preferences about what you do outside of this forum because that is none of my business; it’s your freedom.
I checked the website earlier. Is this what I think it is . write prompt and AI will open krita and do steps to produce artwork in krita. If yes then I think I have seen job offers for actual artists to scren record themselves using krita and send these screen recording with detailed descriptions and time stamped annotations which will in turn train the AI to use krita. such a sad state of affairs, I do not know who uses this. if they want someone to use krita for them then hire an artist and tell them to paint live for you ![]()
The Krita project now has a moratorium on GenAI use and discussion on GenAI policies[0] until October, so maybe we could do the same thing here? We disallow all GenAI use in coding plugins and revisit the policy in October.
Of course, I feel it is better to just keep the ban, but we could just follow with the core project.
I feel the policy where we can file reports for obvious GenAI use is a sane one that won’t put undue burden on the reviewers. The reviewers can do the same they have been doing, and others could file reports of GenAI use. There could be a note saying the report has to be a detailed one, or it will be thrown away.
as a solo gamedev, artist and a programmer (who studied LLMs and GenAI from inside just to know the enemy) i fully agree with you. but at the same time i agree with Takiro’s reply that it is just not very possible unfortunately.
I agree that it can be a challenge, but I also believe it is important to make a stand. We can still have a policy against it, and let the community submit reports if they see anything. Just make sure the reports have to be detailed enough, so the burden on the reviewers is as low as possible.
It’s still horrible technology, we still should not stand for it.
Also, this is a good read:
I think banning it is the right approach.
This technology is very unethical, GenAI as a whole is wreaking havoc in so many professions, including programming. Its drawbacks are massive and its benefits are very much questionable.
There are repetitive tasks, yes, but they can be minimized in other ways. Usually they are not the biggest challenge, though. Just something to plow through sometimes.
It’s not a matter of showcasing skills, not at all. It’s a matter of what we, as a community, should stand for. And I do not think we should be accepting of this technology that devalues humanity, props up the worst people on the planet, burns the planet, and many other things.
It’s a wild speculation that in 2-4 years generative models would be able to generate whole programs like Krita. I do not think that’s true. But even if that were the case, it doesn’t matter what we do right now, anyways. We better just stick to our morals.
I can see that this technology can help you with writing code, and that you really would not want to code without it. I wish I could give you an ethical alternative for this, but sadly I don’t think that a direct alternative exists.
We are a community, we can work together on projects, helping one-another out. If there are things you want, you could request them, and other people might pick it up if they’re also interested. This is, to me, one of the biggest benefits of a free software community. Together we can achieve great things, and we don’t need to take shortcuts or use unethical means for it. And doing this will only make our community stronger.
Be aware that I am arguing against generative models specifically. I am not against machine learning in general.
If it was trained on inputs where all the authors have given explicit consent, and the goal is not to replicate human creativity, then I take no issue with the model.
The internet was not created to train “AI”, I don’t know where you would get this idea.
I’m talking about ethics, not legality. Morality and legality are two very different things.
Because Krita being proprietary software would be unethical.
I don’t know what point you’re trying to make there.
Yeah, we can agree to disagree on this. No hard feelings, wish you a wonderful day.
This explains the borderline nonsensical questions in the threads you posted. People like you compound to a huge waste of time of free software developers that think they may actually be helping a potential contributor learn instead of dealing with a dribbling prompt engineer that got stuck in the vibes.
The legality of Generative AI is being argued in the courts, especially by authors and artists who make money from their work and object to it being copied for Generative AI training.
Those are the people who can, via their puiblishers, afford to take matters to court.
Other, less wealthy and less supported, people can’t afford to fight against having their original creative works copied and used in this manner for the benefit of wealthy corporations with no payment or even recognition to themselves and no say in the matter.
It’s ironic that you used the word ‘hallucinated’. I laughed because it feels better than crying.