Stable Diffusion XL on TPUv5e
Generate images from text prompts with various styles
Going from a text prompt to a nice 3D model
Generate images from text prompts with various styles
Note Start with this Space to generate some characters images (it's really fast thanks to TPUs). Try to use some keyword like "Bind pose" or "T-pose" to get a nice full-frame character . Don't forget to use the "Your past generations" tab to save everything you generate.
Note This is useful to remove all background and get a clean and sharp model. I also use Figma (free) to center my subject and remove extra things.
Note This is the magical part! Upload your clean character image and get a clean 3d model from It. If the Space is too busy, use "Duplicate this Space" to skip it (it should work well on a A10G small). Once that's done, you can download your model in glb format (textures included!) and use it wherever you want.