Spaces:

---

Kunbyte

/

Lumen

like 4

Running on Zero

Dear Hugging Face Team,

I am writing to request a high-capacity GPU (>20GB VRAM) for our Hugging Face Space to host Lumen, our state-of-the-art video relighting framework achieving consistent video relighting and harmonious background replacement.

Lumen represents a significant advancement in AI-powered video editing, enabling:

Consistent lighting adjustments across video frames
Seamless background replacement with proper light adaptation
Preservation of foreground object properties during transformation

The computational demands of our model necessitate substantial VRAM to process video data efficiently while maintaining high-quality output. Our implementation leverages large-scale video generative models that require significant memory resources for optimal performance.

By hosting Lumen on Hugging Face Spaces with appropriate GPU resources, we aim to:

Provide an accessible demonstration platform for the research community
Showcase practical applications of generative AI in professional video editing
Enable broader experimentation with advanced video manipulation techniques

This deployment would serve researchers, content creators, and developers interested in exploring cutting-edge video transformation capabilities.

We greatly appreciate your consideration of our request and would be delighted to provide any additional information you might need.

Thank you for your support of the AI research community.

Sincerely, Jianshu Zeng