AI-KSK/ksk-video-241-comfyui-workflows

Anima Base 1.0: Text, Image, Tiled Upscale and Inpaint Workflows With LLLite Control

Four Anima Base 1.0 ComfyUI graphs: text-to-image with LLLite control and face/hand repair, image-to-image, a four-tile upscale and masked repaint.

Contents

• README.md - this tutorial

What ships

• Four ComfyUI graphs: text-to-image with LLLite control and face/hand repair, image-to-image with the same tail, tiled upscale, masked repaint.
• Base model anima_baseV10.safetensors, text encoder qwen_3_06b_base.safetensors, VAE qwen_image_vae.safetensors.
• Optional anima-turbo-lora-v0.1 route (8-12 steps, CFG 1).

Deployment

• Run the graphs on RunningHub (links below), or drop the JSONs into ComfyUI/user/default/workflows.
• Model files must be present in models/unet, models/clip, models/vae; the JSONs reference them by the file names above.
• The prompt node is an API node (RH_LLMAPI_Pro_Node) - replace or delete it for offline use.

Links

• RunningHub - Anima Base text-to-image + CN + face/hand fix: https://www.runninghub.ai/post/2055109924421685249/?inviteCode=rh-v1111
• RunningHub - Anima Base image-to-image + CN + face/hand fix: https://www.runninghub.ai/post/2055116648939565057/?inviteCode=rh-v1111
• RunningHub - Anima Base tiled upscale: https://www.runninghub.ai/post/2055118475793838082/?inviteCode=rh-v1111
• RunningHub - Anima Base masked repaint: https://www.runninghub.ai/post/2055109942587215873/?inviteCode=rh-v1111
• YouTube: https://www.youtube.com/watch?v=FM5dGuTnz8o
• Bilibili: https://www.bilibili.com/video/BV1ci5q65EGQ

Limits

• Repaint is included for completeness and is the weakest graph.
• Turbo LoRA trades style fidelity for speed.
• Face/hand repair needs the Ultralytics detector + SAM models installed.