Reddit user shares AI video made with Krea H3 and ComfyUI

Reddit user underlogic0 shared an AI-generated video made on a 3090 Ti (64GB RAM) using Krea's H3 image-to-video model in ComfyUI. Clips take 3-6 minutes each at 20 steps, with RTX Super Resolution, Apply Spectrum, and SageAttention wired into the default workflow.
1 source
Daily brief
Get tomorrow's AI brief in your inbox
More stories today
- Testing the Motion Context node for Stable Diffusion
- Krea2 Turbo BBOX fine-tune uploaded to HuggingFace
- Reddit user shares Minimax H3 character/object V2V swapping template
- ChatGPT accidentally makes photorealistic image mistaken for real photo
- Angelo: Video Edition coming soon for ComfyUI