Grandmaster from Thor RagnarokTrained on 2 RTX 3090s locally in 2 hours, 42 minutes. Dataset:<ul><li>10 3 second clips from Thor: Ragnarok</li></ul> Training settings:<ul><li>[512] resolution bucket</li><li>[24] frame bucket</li><li>0.0001 LR</li><li>1000 Steps (100 epochs at 10 repeats)</li></ul>Trained using <a target="_blank" rel="ugc" href="https://github.com/alisson-anjos/diffusion-pipe-ui">https://github.com/alisson-anjos/diffusion-pipe-ui</a> (a fork of <a target="_blank" rel="ugc" href="https://github.com/tdrussell/diffusion-pipe">https://github.com/tdrussell/diffusion-pipe</a> but with a decent Gradio UI, TensorBoard, and Jupyter UI built in a Docker container for Runpod folks)TensorBoard Stats:<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/bd4169ba-318e-44d9-94cd-eab2e796de51/width=525/bd4169ba-318e-44d9-94cd-eab2e796de51.jpeg" />

Files with SHA256: d9c9d9fe247ee48b0654e9e7cca3bb15d09edaa28604d8af80bf93108534aac4