ATTENTION! This is my third Wan 2.2 LoRA and is mainly for experiment and sharing ideas about training a Wan2.2 LoRA.ComicBook-Wan2.2-T2V-A14B. You can use <a target="_blank" rel="ugc" href="https://mediaarea.net/en/MediaInfo">MediaInfo</a> to check the generation info in the video. The workflow I used is the official one but adding sage attention and LoRAs after loading the ckpt.If you love my work, you can buy me a coffee : <a target="_blank" rel="ugc" href="https://ko-fi.com/ghostshell">https://ko-fi.com/ghostshell</a><h1 id="update-version-2.0">UPDATE VERSION-2.0</h1>The effect I tested is crazy. Sample videos are generated in [720,960].wan22-style7-comic-sel-high-2-000040.safetensors is for high noisewan22-style7-comic-sel-low-4.safetensors is for low noiseThe Image below contains the Workflow.<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/a232ce2a-c4ca-4ba2-95de-159c8697b405/width=525/a232ce2a-c4ca-4ba2-95de-159c8697b405.jpeg" /><h1 id="version-1.0">VERSION-1.0</h1>ComicBook-Wan2.1-T2V-14B. You can use <a target="_blank" rel="ugc" href="https://mediaarea.net/en/MediaInfo">MediaInfo</a> to check the generation info in the video. The workflow I used<a target="_blank" rel="ugc" href="https://github.com/kijai/ComfyUI-WanVideoWrapper"> ComfyUI-WanVideoWrapper</a>.LoRA strength suggestion(to balance the movement and the style) : 0.85. It is trained by images, most of the time the style works ,but sometimes it doesn't.I uploaded an article of my experience of training Wan2.1 LoRA (Both I2V &amp; T2V): <a target="_blank" rel="ugc" href="https://civarchive.com/articles/16936/my-personally-training-experience-of-wanstarting-with-data">My Personally Training Experience of WAN:Starting with Data</a>.

ComicBook-Wan2.2/2.1-T2V-A14B

Wan2.2 Version,unzip the file and put high lora for high noise, low lora for low noise

Files with SHA256: 8344c13a0d1b1b1e4c71d9f401983622e3916621fddb4e0f098ae49954fe33e3