<h1 id="heading-408">AnimateLCM-I2V</h1><p></p><h2 id="heading-2153">LoRAs and Motion weights for fast Image-to-Video generation.</h2><p></p><h2 id="heading-2154">We support high-quality image conditioned video generation in 4~8 steps.</h2><p></p><p>Important Notes:</p><ol><li><p>Use LCMScheduler for sampling.</p></li><li><p>CFG should be kept between 1 and 2. CFG is set to 1 by default. </p></li><li><p>LoRA weights is set to 0.8 by default.</p></li><li><p>Motion scale factor is set to 0.8 by default.</p></li><li><p>16 frames is preferred. Too long or too short videos tend to cause generation failure.</p></li><li><p>Image size greatly influences the generation quality.</p></li><li><p>Generation failure might occur due to the limited model size and training resources.</p></li></ol><p></p><h3 id="heading-2155">For more details, please refer to <a target="_blank" rel="ugc" href="https://github.com/G-U-N/AnimateLCM">https://github.com/G-U-N/AnimateLCM</a></h3><p></p><h2 id="heading-2156">Links</h2><p>Contact: <a target="_blank" rel="ugc" href="mailto:fywang@link.cuhk.edu.hk">fywang@link.cuhk.edu.hk</a></p><p><a target="_blank" rel="ugc" href="https://arxiv.org/abs/2402.00769">[Paper]</a></p><p><a target="_blank" rel="ugc" href="https://github.com/G-U-N/AnimateLCM">[Code]</a></p><p><a target="_blank" rel="ugc" href="https://animatelcm.github.io/">[Demos]</a></p><p></p>

Files with SHA256: d083e6a5dedc9140695b8cb9a2f32ce6be32ef6da346ebca55340873b4139475