bar comparison?

[bturkus]

this may not be possible, but I think it'd be cool if qct-parse (or signalserver) could perform a color bar comparison, not just detection and disregard. I'm not sure exactly how detection is currently implemented (looks like it's based solely on Y values), and I'm sure if/how chroma-related signalstats values (U,V, Hue, Sat) translate to specific vectors, but I always perform a quick color bar check on any file that has bars at the head (if the bars are way off, I'll review technician notes for a justification). Finding a way to automate this process, and flag the worst offenders, would be really awesome.

[brnco]

This is a pretty good idea; just to make sure I have it: you have video1.mov and video2.mov, or video1.mov and reference-bars.mov, you'd like to pass both of their QCTools reports to QCTparse, run bar detection, then tell you if the bars are close to one another in measurement.

You're right that the bar detection is primitive at this point, it kind of relies on having good bars in the first place. That's something @iamdamosuzuki figured out, so maybe we can make that a bit more robust, if we're going to be finding bars that look a bit wacky.

I think we can do it though, we'd have to make a cache of like, an average frame from the bar-detected segment for each input, then compare them..... we'll get to it