This is a merge of Anything V3 and Protogen 2.2
Great for any form of art
Can make anime and cartoon or realism.
IMPORTANT
Use Anything' vae(desaturated otherwise) and CLIP skip 2
Useful prompts,tokens,settings:
I keep low steps(about 20-30) and cfg(5-7) low. High cfg generates images with high contrast
"lowres, (bad anatomy, bad hands:1.1), text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, artist name, b&w, weird colors, (cartoon, 3d, bad art, poorly drawn, close up, blurry:1.5), (disfigured, deformed, extra limbs:1.5)" - negative prompt(insane boost to quality)
"stylized" - sometimes give better results
"realistic" - can be used with negative prompt or/and "stylized", gives interesting results
"aura" - use when creating portraits or humans in general, magnificent results
Pics with red square are made with img2img from txt2img outputs, to show you possibilities, other are made with txt2img
Description
FAQ
Comments (4)
Saw your images on Reddit and was blown away, excited to give this a go. Thanks for sharing along with the generation params.
probably one of my favorite models; think we may be able to get a safetensors version in the future for those with real low vram?
Just a heads up. This model includes Protogen 2.2, which includes the Dreamlike license, so not creative ml licensing for this model
pls pony XL ver