Researchers at Adobe have created a new artificial intelligence system known as VideoGigaGAN, capable of enhancing the quality of blurry videos to a resolution up to eight times their original, while maintaining temporal coherence between frames.
Building upon the successes of the GigaGAN model in enhancing clarity, VideoGigaGAN aims to bring this level of rich detail and improvement to video recordings while preserving balance and temporal stability.
Adobe claims that the VideoGigaGAN technology stands out from traditional Video Super-Resolution (VSR) methods as it can add intricate details without introducing any unfamiliar artifacts to the scenes.
High-resolution video enhancement technologies face challenges in maintaining image consistency between consecutive frames while striving to create fine details with high frequencies.
High-definition video quality improvement techniques have shown stability in timing, despite a tendency to produce less sharp images compared to image enhancement techniques due to limitations in their generative capabilities.
Generative Adversarial Networks (GANs) excel in enhancing photographic image quality to higher resolutions, but encounter challenges when applied to videos, such as flickering and other unwanted effects.
VideoGigaGAN aims to achieve a high level of quality in diverse generative network models, while addressing issues related to flickering and distortions that may occur across different frames.
The new release of Adobe’s model immediately tackles this issue by enhancing the GigaGAN architecture to incorporate video clips.
VideoGigaGan stands out for its ability to revolutionize the video editing process, enabling users to enhance the quality of low-resolution videos to ultra-high resolution, while avoiding common issues like distortions or loss of image quality.
VideoGigaGan leverages advanced artificial intelligence algorithms to enhance facial details, increase clarity, enrich natural appearances, resulting in high-resolution videos with precise details akin to modern high-definition video recordings.
This method proves to be an extremely valuable tool for experts dealing with historical or vintage clips, including filmmakers, documentary directors, and video editors.
With the use of Adobe’s newly launched system, experts can now revive classic video clips and enhance their quality to meet contemporary standards.
The technology is still in the research experimental stages, with no confirmation that Adobe will make VideoGigaGAN available to the public through its Creative Cloud platform, like Premiere Pro.
During the MAX event held in October 2023, Adobe introduced a new technology called Project Res-Up to enhance clarity, focusing on improving the quality of lower-resolution GIFs and videos.
It is worth noting that Adobe is not the only company involved in this initiative, as Microsoft and NVIDIA have also developed advanced video quality enhancement technologies known as VSR.
