lostintranslation
Brick Duck@lost753541 I've been seeing your stuff on my timeline, and I'd like to ask: are you prompting for such unusual images? Or do you have to inpaint? Like, this image seems like it would have been hard to do with just a text-to-image prompt.
@woltiv these are functionally non-promptable - it's actually the core idea I had, to make distinct original "photomanips" with AI tools to make previously less than feasible concepts happen in photorealistic style.
No existing models support these, some have lora support that can get a rather awful first draft, but then it's 1-8h of hand painting, traditional photomanipulation, and (most heavily) inpainting of the former two, to get to this level of quality.
No existing models support these, some have lora support that can get a rather awful first draft, but then it's 1-8h of hand painting, traditional photomanipulation, and (most heavily) inpainting of the former two, to get to this level of quality.