Tight, tight, tight! Cool, cool, cool!I'm currently experimenting with using H3 as an image editing model, and it's pretty interesting.It turns out someone has already figured out how to generate images with H3. This isn't my original work—I was just so excited that I couldn't wait to share it with everyone. (<a target="_blank" rel="ugc" href="https://huggingface.co/Mamad8/MiniMax-H3-Image-VAE/tree/main">https://huggingface.co/Mamad8/MiniMax-H3-Image-VAE/</a>)Replace <code>ComfyUI\comfy_extras\nodes_minimax_h3.py</code> (Back up first!) with the new file, then restart ComfyUI. After that, simply import the workflow file and you're ready to go.Download the VAE file here: <a target="_blank" rel="ugc" href="https://huggingface.co/Mamad8/MiniMax-H3-Image-VAE/tree/main">https://huggingface.co/Mamad8/MiniMax-H3-Image-VAE/tree/main</a>Images decoded from a single frame using this VAE have noticeably higher quality.<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/e66b14b9-cec7-4766-9f15-4dcb2bffc1a1/original=true/e66b14b9-cec7-4766-9f15-4dcb2bffc1a1.jpeg" />

Files with SHA256: 1e3b18d2771ff6a5c3dbb0805ec60726ebf58fd099067240bbc22307fb33f8ef