CoolFace
Apppublic

adeloscope/sulphur-2-video-demo

sourceHugging Faceupdated 2mo agoView on Hugging Face
0likes
App README

Sulphur 2 Video Demo

Generate short text-to-video or image-to-video clips with SulphurAI/Sulphur-2-base.

This Space uses Sulphur 2's FP8 base checkpoint with its supplied distillation LoRA, the LTX 2.3 x2 spatial upscaler, and Gemma 3 as the text encoder. It uses ZeroGPU, so a request may enter a short queue while a GPU is assigned.

Tips

  • —Describe a single shot in chronological, visual detail. Camera movement, lighting, and action usually improve results.
  • —Add an optional first-frame image for image-to-video.
  • —Start with the 49-frame option; higher frame counts use more GPU time.
  • —Width and height are intentionally fixed to 768×512 for predictable ZeroGPU memory use.

The model and its outputs can be imperfect. Only upload media you have the right to use, and do not use the Space to create harmful or deceptive content.

Credits