For anyone downloading the 1.0 version be warned you might encounter some issues, but There is an UPDATE in the works, the workflow is kind of broken because of UE links on the newest comfy update i think. So next version it will be more organized, but will have more LINES everywhere. And after some thorough testing, i've Achieved some nice quality improvements on output, Enough so you can use just the first render to only animate dances and movements, and i am currently working on what the second pass should be for, and that is, the hardcore NSFW tasks that need more guidance, it will take a while but this is getting good. Bellow are some results with the Q6 GGUF model, 2 seconds of video made on 16GB of VRAM. First image is the input, Last image is the last frame of the generated video with no color correction done, 2.5D images are not as blurry as before and i think it's good enough for even trying to extend the video. <img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/238ecd2d-ded2-491c-8bcb-a553dd833263/width=525/238ecd2d-ded2-491c-8bcb-a553dd833263.jpeg" /> to full control over motion, dancing etc, low vram with GGUF Q4 with very good quality. i don't even know if i am uploading this correctly, and i have little time. just throwing this on the internet hopefully someone crafts something better with this as base. GGUF of wan fun models can be found here <a target="_blank" rel="ugc" href="https://huggingface.co/city96/Wan2.1-Fun-14B-Control-gguf/tree/main">https://huggingface.co/city96/Wan2.1-Fun-14B-Control-gguf/tree/main</a> This is very node heavy, it has some instructions inside thrown around, please use some tool like comfyui manager to download whatever nodes you need. Some nodes that don't auto download a lot of times and can break the gguf loader <a target="_blank" rel="ugc" href="https://github.com/city96/ComfyUI-GGUF">https://github.com/city96/ComfyUI-GGUF</a><a target="_blank" rel="ugc" href="https://github.com/pollockjj/ComfyUI-MultiGPU">https://github.com/pollockjj/ComfyUI-MultiGPU</a> git clone these if the gguf loader is red. Just pointing out, finding yourself in this workflow at first is a PAIN! I'm not good at crafting workflows, i usually just slap whatever seems to work best but i did what i could to make the first version to be more "organized"... Here's some, maybe useful help.<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/b515c544-9a9f-4ba8-aad0-7f7be9b030c9/width=525/b515c544-9a9f-4ba8-aad0-7f7be9b030c9.jpeg" /><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/d306c11a-e5f6-492d-a267-b58de50596ab/width=525/d306c11a-e5f6-492d-a267-b58de50596ab.jpeg" />you can bypass splines, combine multiple, also neat little trick is if there's too much camera movement is to just create a 2 point spline in a corner where there's just static stuff, it will tell the model to not move "or stray too much" from that point, works with body parts, clothing etc.... combine as you wish. If you're interested in posting here whatever you've made or anything you've found to work better feel free to do so i will read it and test it too.

Files with SHA256: fa2bc0ab0955104662e55128caa1e2d4eb82ce0c5842923ce127aa49f6f26384