Pre-Publication V2. A process that increases the apparent resolution of an image or video by using learned visual patterns from training data to infer plausible detail beyond what is directly readable in the source. This can include diffusion models, GANs, and in some cases other neural network approaches, but the key point is that the model is not revealing hidden information from the original. When detail is absent, damaged, or too limited to resolve, the system synthesizes a more legible interpretation guided by the source image and those learned patterns, rather than directly recovering the original image.
Deliberation Summary:
Consider edits as term “apparent” carries substantial weight in the definition and is only explained later in the text.
Pre-Publication V1. A process that increases the apparent resolution of an image or video by using learned visual patterns from training data to infer plausible detail beyond what is directly readable in the source. This can include diffusion models, GANs, and in some cases other neural network approaches, but the key point is that the model is not revealing hidden information from the original. When detail is absent, damaged, or too limited to resolve, the system synthesizes a more legible interpretation guided by the source image and those learned patterns, rather than directly recovering the original image.