In a paper published back in April, Adobe researchers have shared the news that they’ve been hard at work developing a new generative AI model which they call VideoGigaGAN which is aimed towards detail-rich video super resolution.This could end up being one of the best examples of artificial intelligence’s ability to upscale low-quality video (like smartphone videos or other AI-generated videos) to a much higher quality, in effect making any of these low-scale versions more commercially viable in the future.Let’s take a look at this new technology being developed by Adobe and see how it might be able to eventually work for your video production and editing workflows.Adobe’s VideoGigaGAN AI ModelIf you really want to nerd out here, there's a piece published exploring VideoGigaGAN on Github that goes into the high-level and abstract about what this video super-resolution project is trying to accomplish. All you really need to know is that these GANs (Generative Adversarial Networks) have traditionally been great for upscaling video, but have almost always had issues with distortion and flickering.With this new VideoGagaGAN project though this new upscaling method promises to help remove these distortions and flickers, plus other issues that we’ve seen with other AI upscalers for video. VideoGagaGAN promises to produce videos with high-frequency details and temporal consistency and has reportedly shown that it can produce video results with 8x super-resolution.Will This AI Tech Come to Premiere Pro?The big question here for most of us in the film and video industry might simply be when—or...
Published By: NoFilmSchool - Wednesday, 8 May