(For SOTAI Squad) FINALLY! Consistent Characters in Wan Video! BindWeave ComfyUI Tutorial (Day 2)
Added 2025-12-15 17:40:38 +0000 UTCLink video: https://youtu.be/AWkkoO5yoCs
In this video, I’ll show you how to upgrade your BindWeave workflow in ComfyUI to achieve ultimate control over your characters in Wan video.
This is Day 2 of the series “Finally: Consistent Characters in Wan Video! BindWeave ComfyUI Tutorial”, so we’re moving beyond the basics to tackle the most requested advanced techniques:
✅ VRAM Optimization: A step-by-step guide to running this heavy workflow on mid-range GPUs (12GB VRAM)
✅ Outfit & Background Control: How to change costumes and environments completely while keeping the character's face perfectly consistent.
✅ Multi-Character Workflow: The secret to generating videos with 2 or more characters interacting naturally without any "character bleeding" artifacts.
Step 1 - Install the custom node
Install WanVideoWrapper node: https://github.com/kijai/ComfyUI-WanVideoWrapper
Search in ComfyUI Manager or instal via Git clone and pip install
Move to custom_node/WanVideoWrapper folder: git switch bindweave
Step 2 - Download the model and place it in the correct folder
https://huggingface.co/vantagewithai/BindWeave-GGUF/tree/main(GGUF)
Place the file in: "models/unet/Wan2.1-14B-BindWeave-Q3_K_M.gguf"
https://huggingface.co/Kijai/WanVideo_comfy_fp8_scaled/tree/main/Bindweave
Place the file in: "models/diffusion_models/Bindweave/Wan2_1-I2V-14B-Bindweave_fp8_scaled_KJ.safetensors"
Place the file in: "models/loras/Wan21_I2V_14B_lightx2v_cfg_step_distill_lora_rank64.safetensors"
https://huggingface.co/lllyasviel/misc/blob/main/clip_vision_vit_h.safetensors
Place the file in: "models/clip_vision/clip_vision_vit_h.safetensors"
Place the file in: "models/text_encoders/qwen_2.5_vl_7b_fp8_scaled.safetensors"
https://huggingface.co/Wan-AI/Wan2.1-T2V-14B/blob/main/Wan2.1_VAE.pth
Place the file in:
"models/vae/Wan2.1_VAE.pth"
https://huggingface.co/Kijai/WanVideo_comfy/blob/main/umt5-xxl-enc-fp8_e4m3fn.safetensors
Place the file in:
"models/text_encoders/umt5-xxl-enc-fp8_e4m3fn.safetensors"
Please check the workflow in the attachments.
Comments
1. You can choose any ComfyUI version you want, but it’s recommended to use the latest versions, as they support all the nodes in the workflow. 2. As I mentioned in the video, higher GGUF versions generally give better quality. You should try from higher to lower versions to find the highest GGUF your system can handle — Q6 is likely a good choice. 3. You should use SageAttention.
SOTAI
2025-12-29 03:38:38 +0000 UTCBased on my system (RTX 4080 with 16 GB VRAM), which ComfyUI version would you recommend for the best performance? Which GGUF models run faster while still providing good quality results? And which attention backend should I use for optimal performance—FlashAttention, SageAttention, or xFormers?
FintuBoi
2025-12-29 00:25:54 +0000 UTC
