WAN 2.1 AI in ComfyUI: Native Integration for Stunning Video Generation!
Added 2025-02-27 17:31:36 +0000 UTC
Video : https://youtu.be/ttlS0I-eClA
This tutorial covers text-to-video and image-to-video workflows, upscaling, and sound effects using MM Audio. Learn to optimize performance with FP8/BF16 models, TCache patches, and dynamic camera motions for cinematic results like alien invasions and angelic scenes.
Who is this content suitable for?
AI developers, video creators, and ComfyUI users with mid-to-high-end GPUs. Perfect for those exploring advanced AI video generation, cinematic storytelling, or optimizing workflows for faster, high-quality outputs.
Why does it matter?
WAN 2.1 brings professional-grade video generation to ComfyUI, offering smoother motions, richer details, and dynamic camera angles. Whether creating sci-fi scenes, character animations, or experimental clips, this model empowers creators to push the boundaries of AI-driven storytelling—locally and without subscriptions!
Resources:
https://comfyanonymous.github.io/ComfyUI_examples/wan/
https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/tree/main/split_files
https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/tree/main/split_files/vae
https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/tree/main/split_files/text_encoders
https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/tree/main/split_files/diffusion_models
https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/tree/main/split_files/clip_vision
Workflow for txt2vid img2vid, mmaudio, upscaler attached below:
Comments
Hey Derrick, have you try the math expression part, for width, height, connect with tile , then try it with this: width*(upscale size)*0.25
Benjamin Law
2025-05-13 08:33:19 +0000 UTCHi, Ben. I have been using this work flow and the upsacle module you produced. Maybe an update of comfyui broke the upscaling part. I tried to rebuild the upscaling part using ultimate SD upscale node with SDXL models with denoise at 0.1. But the video come out flickering. Any suggestion. Thanks.
Derrick Pang
2025-05-13 04:10:32 +0000 UTCHello , which version of wan model you are using? Thanks
Benjamin Law
2025-03-21 06:19:30 +0000 UTCHello Benji! Thank you for your tutorials and workflows! I have a question - whenever i try to make a generation a get a noisy mess (example https://s3.ru1.storage.beget.cloud/88095fdffa8e-tidy-trish/gen/img2vid/vid_img2vid_output_00005.mp4) The issue is only with this workflow, and if i generate in original comfyui wan workflow it goes well. Where can i start looking for an issue? I tried different models and differnt weight types in a node, but still no luck
Oleg Parshentsev
2025-03-20 21:43:56 +0000 UTC
