I think it is more accurate to say: Estimate in manual mode checks the values needed on the frame you are on when pressing the estimate button and then applies the same values of that frame throughout the entire video ignoring the fact that other parts of the video (frames) might react better to different values.
Not exactly. in Manual when you press “estimate”, it only estimates the values needed from the frame you are on when pressing the estimate button.
It doesn’t comb through your entire video to estimate it and provide you some kind of an average value(s) for the video, no, it doesn’t. Only one “representative” frame for the entire video.
Auto mode is a different story altogether. In Auto mode, the estimation happens every 8 frames throughout the entire video.
e.g. If your video has 8000 frames in total (2m 22s clip at 60 fps), there would be 1000 estimations happening behind the scene during the upscale / enhancement process.
i.e. each “estimation” frame would stick its values for itself and to the upcoming 7 frames and use those values to all 8 frames ( itself + 7 frames e.g. frame 1 - 8). then a new estimation would happen (on frame 9) and the values would be used for itself and the upcoming 7 frames (frame 9 - 16). then a new estimation would happen (on frame 17) and the values would be used for itself and the upcoming 7 frames (frame 17 - 24), and so on until the video is completed.
