What is the back-end for the Text-to-Image Generator?
I've noticed that the linked page gives me much better quality results than I get on my own locally run ComfyUI or StableDiffusion distros.
Does anyone know what they use for that page?
https://perchance.org/ai-text-to-image-generatorOpen linkView original on lemmy.world
https://perchance.org/t2i-framework-plugin-v2 as mack said its flux 1 schnell as far as the few others some run through pollinations.ai that i know of as im am sure there are others, i've got some that run through huggingface.
The majority of Perchance t2i generators use a Flux.1-Schnell model now, as far as I know. A few may be using something from outside of perchance.
i cant imagine it's flux-schnell! nowhere i've seen a flux1 instance being THAT NSFWy and i think it's almost impossible. it would take much more than just plain finetuning by 1000s of nsfw images rather than a massive redesign and token ingestion. does anyone really know what the model behind txt2img here?