Innovate Futures @ Benji
patreon
Wan 2.1 Fine Tune Models going wild! CausVid t2v And MoviiGen1.1
🕑 Added 2025-05-16 13:58:56 +0000 UTC
Video : https://youtu.be/zO52xUBJlP0
In this video, we explore two powerful AI video models, CausVid and MovieGen, both fine-tuned from WAN 2.1. CausVid specializes in fast auto-regressive video generation, ideal for quick, short clips, while MovieGen focuses on cinematic-quality video with enhanced realism and higher resolutions like 720p and 1080p. We test these models in ComfyUI, comparing their performance, speed, and output quality, and provide practical tips for running them on consumer-grade hardware. Whether you're an AI enthusiast, content creator, or developer, this video offers valuable insights into leveraging these tools for AI-generated video production.
Who is this for?
AI researchers and developers
Content creators exploring video generation
Tech enthusiasts interested in diffusion models
Anyone working with WAN 2.1 or ComfyUI
Why it matters:
AI video generation is evolving rapidly, and fine-tuned models like CausVid and MovieGen push the boundaries of speed and quality. Understanding their capabilities helps creators choose the right tool for their projects, whether for fast prototyping or high-end cinematic visuals.
CausVid
https://causvid.github.io/
https://github.com/tianweiy/CausVid
https://huggingface.co/tianweiy/CausVid
MoviiGen
https://huggingface.co/ZuluVision/MoviiGen1.1
https://huggingface.co/Kijai/WanVideo_comfy/tree/main
Safetensors for ComfyUI
MoviiGen1.1
FP16: https://huggingface.co/Kijai/WanVideo_comfy/blob/main/Wan2_1-MoviiGen1_1_fp16.safetensors
FP8: https://huggingface.co/Kijai/WanVideo_comfy/blob/main/Wan2_1-MoviiGen1_1_fp8_e4m3fn.safetensors
CausVid
https://huggingface.co/Kijai/WanVideo_comfy/blob/main/Wan2_1-T2V-14B_CausVid_fp8_e4m3fn.safetensors
CausVid as Lora work with Wan2.1 T2V model
https://huggingface.co/Kijai/WanVideo_comfy/blob/main/Wan21_CausVid_14B_T2V_lora_rank32.safetensors
