Krea 2 Photo to Anime: Vision-Reference Restyle

Turn a photo, or any picture, into a painterly cel-shaded anime illustration, with the composition kept. About 15 seconds on our card.

Required models

• Diffusion model: RedCraft 3.0 Krea 2, the INT8 convrot build, redcraft23_30Krea2_int8_convrot.safetensors, in models/diffusion_models. Plain Krea 2 and Krea 2 Turbo load too; RedCraft is what we tested.
• Text encoder: the Qwen3-VL 4B text encoder from the Krea 2 downloads (see the Krea 2 guide on comfyui-wiki.com); we run a community int8 ConvRot build of it (qwen3vl_4b_int8_convrot), and the bf16 or fp8 file works the same. It goes in models/text_encoders, loaded with CLIPLoader type krea2.
• VAE: the Qwen-Image VAE, qwen_image_vae.safetensors, in models/vae, from Comfy-Org on Hugging Face.

Settings

• Steps 8. It is a distilled model, more steps do not help.
• CFG 1.0, sampler er_sde, scheduler simple, denoise 0.75
• Size: 1 megapixel buckets, 896x1152, 1152x896 or 1024x1024
• Negative: a ConditioningZeroOut. At CFG 1.0 a negative prompt does nothing, so the graph does not pretend.
• The picture is scaled to 1 megapixel first.

How it works. The picture goes in twice: as the img2img start, and as a Qwen3-VL vision reference (Picture 1) in TextEncodeQwenImageEditPlus. So the model sees what it is redrawing, and only the medium changes.

• Start at denoise 0.75. Below 0.7 it reads as a filtered photo, not a drawing. Go higher for a stronger drawn look.
• Landscapes and wide scenes: raise denoise to 0.80. At 0.75 they can come back looking like the photo with a light paint pass.
• Do not wire a VAE into the text-encode node. We tested it without.
• This is a restyle. It keeps composition, pose and background, and it is not a likeness tool.
• Use your own photos, or pictures you have the rights to.

Tips we learned the hard way

• Density beats brevity. Write a full paragraph: who or what, where, the light, the camera. A short prompt gets you a short picture.
• Say what you want, never what you do not. No negative prompt, and no SDXL habits: weight brackets and the old separator keyword do nothing here. Write "alone in the frame, no other people" instead of a negative list.
• Name the medium. "Drawn in clean cel-shaded anime line art, drawn rather than photographed" holds a drawing; "semi-realistic" alone drifts to photo.
• Stay at 1 MP. The buckets above are what the model was trained around.

Tested on ROCm (a 7900 XTX). Core ComfyUI nodes only, so CUDA builds should run it as is.

If this worked for you, a thumbs-up here helps the next person find it.