very similar to a human producing out an image at a distance, a CNN very first discerns really hard edges and straightforward shapes, then fills in details because it operates iterations of its predictions. This https://visiontech.uno/