I would advise you guys to work on weekly updates just until the current UI works properly, then start updating the software monthly. I am worried about the team’s mental health for the pressure of rushing to fix something every week. Just take your time! I think there are too many models right now, I’d be pretty glad if this happens:
- Make Proteus the only enhancement model (maybe with another name), and train it to do all the enhancements tasks, each one with a different slider, including stabilization, motion deblur and face enhancement with slider options. So we can control how much of these filters affect the footage.
- The frame interpolation model can be a separate one. Take into consideration the dupe frames that have been talked about in the forum.
- Separate another space for interlaced footage. It would be great if we could deinterlace the footage first and without any enhancements or upscaling, and use the second enhancement to apply the necessary filters at a higher resolution. Ex.: QTGMC is pretty enough for me to deinterlace and double the frame rate at 1x resolution, and then use Proteus’ sliders to control my filters and upscale the footage.
- Separate another space to add noise, recover detail and apply grain. I even think that the type of temporal noise that’s applied must go… Grain and recover original detail are enough for me, both can be applied to the final result.
- Make a big screen for previewing the results, even if the full window is not enabled… The current one is too small. Put every option on the side of the screen, and let a big space to the actual footage.
- I like the two windows scheme for comparsion, and the option to choose which footage I want to compare against another, that’s dope, especially at the 1 screen view, the differences are pretty clear. This was possible by being able to switch between view 1 and 2 pretty easily… Great work on that.
- Make the parameters analyze the full video in order to choose the best clip to apply the auto settings, so we don’t have to eyeball the best frames.
- Make the preview zoom according to the final resolution, not the original one.
- Try to spot the artifacts that commonly appear and get rid of them, or maybe there could be an option to mask a certain part of the video to blur the artifacts carried thoughout an x amount of frames, that would be awesome.
- Bring back the option to preview 1 frame per preview, that speeds up some testing.
- Disable the software filling the whole screen from top to bottom like a locked game, when we re-enter the UI coming from another program.
That’s it for now