Researchers at Adobe have unveiled a new upscaling technology they are developing, called VideoGigaGAN. Technically a generative AI model, Adobe claims VideoGigaGAN can upscale low-resolution video by up to eight times while avoiding the typical drawbacks associated with upscaling, namely artifacts and flickering.
In the examples shared by Adobe, the quality of the upscaling is notable, with textures and fine details becoming visible in videos compared to their low-res counterparts. How much of these details are derived from the original video, or simple “imagined” by AI, is unclear. As VideoGigaGAN was revealed in a research paper to share what researchers are currently working on, plans for the technology to be widely available have not been made public. Given the company’s recent moves to introduce AI features to a plethora of their products, including considering integrating third-party AI applications into Premiere Pro, its likely that VideoGigaGAN will be integrated into the Adobe suite in the future, at the very least.