Diffusion upscaling, export previews, and more. All in Video AI 7.
The power of Starlight, optimized for local processing on your machine. Only in Video AI 7.
Starlight Mini is the next generation of the groundbreaking Starlight model, optimized to run locally on machines. It’s the world’s first local video upscaling and restoration diffusion model.
Never before has it been possible to run a video diffusion model locally with this level of quality output. You can now transform low resolution, degraded, noisy, and compressed footage into 4K quality videos right on your machine.
Starlight Mini uses the same diffusion technology developed for Starlight to upscale, enhance, denoise, de-alias, and sharpen videos locally. All without any sliders or manual adjustments. It only takes one click to take almost any video to 4K.
Starlight Mini upscales and restores each frame by analyzing the surrounding frames to achieve temporal consistency.
Old videos with jumpy frames transform into smooth quality that’s never been possible before for a local video diffusion model. This breakthrough in innovative technology paves the way for video restoration to be more widely compatible and accessible across various machines.
Interested in Starlight and Starlight Mini for extended commercial use?
Until Starlight, all previous video enhancement models utilized GAN technology. Starlight Mini continues the breakthrough departure to diffusion-based video upscaling and restoration. Except now, it’'s local. In just three short months, our team of researchers and developers have taken a technology that was large, slow, and only available in cloud rendering, and optimized it for local use on machines. All while preserving the incredible quality results you’ve come to know from Starlight.
Cloud rendering available on all systems when machines don’t meet system requirements for local processing. Starlight Mini local rendering currently available on Windows with NVIDIA GPUs.
Operating System: Windows (cloud rendering with required Cloud Credits available for all)
GPU: Minimum 8GB NVIDIA GPU memory (recommend 16GB GPU memory for optimal performance and quality)
RAM: Minimum 12GB
Diffusion upscaling, export previews, and more. All in Video AI 7.