Recently I discovered a new AI-powered addon for Blender, so obviously I needed to play with it. I liked it so much that I recorded a video tutorial for it:
was also generated by AI (with my help on top of it) :D. And the title of the YT video. And script was spellchecked by AI. And voice over is AI-generated
The installation is a bit lengthy, but overall it’s IMO well done addon. Very easy to use after installing it.
For sure, and some of those textures are really, really good IMO.
Without good setup images can have shadows too.
What we see here is just first, and rough generation of those tools. I’m sure that after couple of iterations they will get better. Right now we are going through AI hype with very overinflated expectations about it. This usually results in disappointment and a kind of anti-hype. But the hype will die down, that tech will be improving and at some point we will have it in all our favorite tools (like AI Denoising now).
I have created a short video showing some of the textures I made:
I also experimented with DALL-E to generate some textures:
My first impressions is that DALL-E is way better for this tasks (but also I might have not used the right prompts for Stable Diffusion).
I think I will also experiment with Substance 3D Stager - it is better at making PBR materials from images than Materialize. And also I think I’ll try some other AI techniques to upscale those textures.
Seems a lot of it being worked on. This one still experimental?
I was wondering with the first one mentioned, I think I have seen a couple of videos on it and just got the impression it throws out the same or very similar results? Rocks with moss. I guess it would same program same sorts of words used as starting reference point. It may just be as users and demonstrators gain experience they will add in extra words and will get more varied results.