I think Deep Dreamer is a good start, but I'm really feeling like the video processing is lacking in options.
There is no way to view or control things like framer rates or codecs. If I drop a video into Deep Dreamer, I keep getting it outputting in different frame rates (drop frame, I think) which results in a jumpy unusable output video with a mismatched number of frames from the source video. Since there are no video options to be found, there is no real work around other than hand dragging each frame and hitting process individually (which isn't really an option when you have 20,000 frames in a video to process).
It would be great if there were the option to load an image sequence (perhaps dragging a folder of an image sequence) and treat it like a video. It gets problematic when you have to encode and re-encode video over and over, plus doing this would help to solve the first problem, because Deep Dreamer would be focused on processing images individually, and then someone can use an external encoder that is purpose built for video. Basically, it could just be exactly how it is now, only instead of forcing the user into using a video as an input and then altering the framerate in the output, it would just take a folder of an image sequence, automatically batch process all of the images and then spit out the exact files, only processed on the other side.
To be honest, I think #2 would be the easiest and best option (I would MUCH prefer image sequence batch processing as it would make this app infinitely more useful for video), giving the user the option to merely use Deep Dreamer as a way to batch process all the video frames without destructively altering things like the codec/compression/framerate. Even when I output each frame as images and then encode it externally, it has dropped quite a few frames which ruins the video.
Are there instructions working with the video anywhere?
Here's an example page of video people have made with the python app:
But mine definitely aren't there yet! I need a bit of guidance. Thanks!