Some people are having trouble recreating the zit samples, so I’ll share what I use to generate them.
Latest Comfyui
z-image-turbo bf16 model
euler/beta or beta57, 8 steps
lora strength 0.8-0.9
Generating images at 1.5MP (1088x1440) or 2MP (1248x1664) can help fix issues like missing fingers or awkward body angles.
I use the Ultimate SD Upscale node to enhance my images, not SeedVR.
Chroma:
Model used to generate samples:
https://huggingface.co/dawncreates/UnCanny-Photorealism-Chroma-GGUF/blob/main/UnCanny_v1.3-flash-Q8_0.gguf
Text encoder:
https://huggingface.co/city96/t5-v1_1-xxl-encoder-gguf/blob/main/t5-v1_1-xxl-encoder-Q8_0.gguf
Use res_2m/beta, 17-22 steps CFG 1
Realistic(fake) american women.
Description
Play with the lora strength, try 0.7-1.0.
Text encoder I use:
Triangle104/Josiefied-Qwen3-4B-abliterated-v2-Q8_0-GGUF