iSIZE’s BitClear provides a scalable solution to denoise and upscale severely distorted low-resolution video assets. The current focus of BitClear is on removal of encoding artifacts (blocking, blurring, ringing, aliasing, etc.). However, the underlined neural network technology can also be trained to remove any other types of artifacts if indicative training data is available. Our solution is 2 to 10 times faster than the state-of-the-art in video denoising and upscaling, while offering superior quality. Via our patented perceptual quality optimization technology, BitClear is able to optimize both standard perceptual quality metrics like VMAF, SSIM and similar, but also visual quality as assessed by human viewers of controlled testing conditions like ITU-T Rec. It allows for single-pass processing with singleframe latency, without needing any side information on the exact encoding or processing that has already taken place, and it can (optionally) upscale the input video by up to 4x with quality/complexity that is tunable according to the compute capability of the deployment hardware. Unlike competing approaches for video denoising that can generate artifacts, BitClear is designed to operate at scale without the need for human inspection, as it is designed to disentangle the source and noise manifolds, and recover video details without changing the aesthetics, perceptual intent or the structure of the decoded video.