ffrm

The best path forward for serif is to open up an api. It side steps legal issues like rights (since they won't provide their own generative ai) and it allows the community to iterate quickly and it helps explore various workflows without punishing their busy devs. For example I can already think of various types of generative plugins I would write if I had the chance.

A subscription is not required to generate AI content. For example, https://stablehorde.net/ provides a cluster of servers that anyone can use, for free. They have REST APIs described on that page. A lot of the more technically minded folks would probably prefer directly invoking their local stable diffusion instances, which have plugins to expose the services as APIs. Basically, these services are available for easy integration at the HTTP level. They are practically begging to be used. Beyond generative AI, also worth noting are tools like Facebooks Segment Anything which can - select anything. https://segment-anything.com/ The digital art landscape shift is absolutely tectonic, and its as easy to integrate as an HTTP call. Again, let the community do the work! We just need the door opened.

Last I heard you needed a Photoshop license to be able to develop for it, which is exactly what I'm trying to avoid. If I'm wrong I'm happy to start on that right away.

I'd like to reiterate my preference for an sdk over direct integration. I understand smaller teams are strapped for time. Give us a way to access a network resource and your community will do the work. As a software eng myself I'd have already made an automatic1111 plugin by now if it had been possible. I also believe the generative tools are changing too quickly for small teams to really respond well. Affinity has a loyal base that should be tapped! I'm not a full time artist but I'm happily enjoying a new relationship with the visual arts thanks to stable diffusion.

Just an update, here's a great example of a modern AI based workflow

Focusing a little more on the actual mechanics of integration, I'm realizing more and more that AI assistance is really more of a painter tool. I've been generating incredible cityscapes with vehicles and noticing I would love to add elements. For example, 1. add a new visual feature (e.g. generate a city background via prompt, or maybe select from an initial list?) 2. draw a crude car on a new layer, or paste an example one in 3. run the AI to upgrade the car on that layer 4. refine via AI and back to step 1 You can argue you can skip all that with the right prompt, but I believe this is where the modern digital artist will want to spend their time. Not texting your photo editor. The workflow will shake the foundations of all digital photo editors. Even Photoshop will have to do some big moves to accommodate a proper AI integration imo.

Signed up just to comment on this thread. Stable Diffusion is EXPLODING across the internet and it will absolutely revolutionize digital art. To soak some of it in, visit us over at reddit.com/r/StableDiffusion or just jump on Twitter. Combining this with photo bashing and there's no way you'll be able to tell that a portion of the piece was "ai generated". It's nonsense that it will "destroy jobs". If you're job was to "paint me an vanilla ice cream cone with sprinkles on top that looks like a photograph", well then yes it's over for you. But if you're using the digital medium to produce graphics, then AI assistance is just one more "Effect" that should be in the tool options.