each point is a frame
there's a slight bend in it near the middle where it used less frames but this is almost linear
hmmmmmm that probably means this was a bad metric for how different the two images are?
i'd rather not do a full p-hash but that might be unavoidable?
tiny resolution because i've not multithreaded this renderer and it take aaaaages
this is the corresponding render
weirdly it decided to use less frames right in the middle where it's most complex so i think it might be doing the exact opposite of what i want