VideoGigaGAN beats other Video Super Resolution methods because it avoids the usual artifacts and flickering introduced by GAN , according to Adobe. At the same time, it adds sharpness and detail — where most other systems fail to do do both of those things at once.Of course, the system is making up detail that doesn't exist out of whole cloth, so this wouldn't be suitable for things like forensic video enhancement,CSI-style crime shows.
But the detail it does add looks impressively real, like skin textures, fine hairs,The model builds on a large-scale image upsampler called GigaGAN, according to to Adobe's researchers. Previous VSR models have had difficulty generating rich details in results, so Adobe married "temporal attention" , feature propagation , anti-aliasing and something called "HF shuttle" to create the final result.