Slowly getting somewhere with generating images (soon 3D scenes?) with Stable Diffusion directly inside Unity with local inference. no API, no subscription, no python #madewithunity #unitysentis

I don't know what you use for the depth but I recommend depth fm it's the best

Thanks! One more entry to the ever-growing list of models to look into, lol. Using Midas for now, but now that I have the base components of SD working that definitely paves the way to add support for models like marigold/depth-fm

Oh I was messing around with this last year, I believe i converted an outputted depth map from Midas into a displacement map i think, yours looks much cleaner xD.

Thanks for the kind words! Always feels weird to show off naive implementations like these when there's so much progress in AI with new stuff being released left and right. Gotta start somewhere though.

Very exciting! Following with interest

cool, let me know if there's any specific direction you'd find interesting when it comes to diffusion models in Unity!

You run it on GPU or CPU ?

on the GPU (using an RTX 3060)

Sounds amazing, will keep an eye on this.