Upscaling
Using AI to increase a video or image's resolution while inferring plausible new detail.
Upscaling increases a video or image's resolution using a model trained to infer plausible fine detail, rather than simply stretching the existing pixels. A naive resize just makes each pixel bigger and blurrier; an upscaler adds texture and edges that look native to the higher resolution.
It's commonly used to take a generated clip up to a deliverable resolution, or to rescue lower-resolution footage for a modern aspect ratio and delivery size.
See also: Latent Space, Inpainting.
Put these ideas into motion.
DownloadFrequently asked questions
AI upscaling increases a video or image's resolution by inferring plausible fine detail that wasn't in the original, rather than simply stretching existing pixels.
No — resizing stretches existing pixel data and gets blurry at larger sizes. Upscaling models add new, plausible detail, which is why results look sharper rather than just bigger.