Dear friends:
Regarding:
1- The randomic nature of the video in the time domain;
2- The video sinal with inverting non-inverting lines is clamped at the same DC level,
I'm pretty sure to be virtualy impossible, in time domain, recognize if the video signal itself was or not inverted.
To build a circuit based on APL changing of the video is a total waste of time, once this APL's changing will be senssible (to triger a voltage comparator) after several video lines had gone.
Also try to do something in the color channel wouldn't carry us anywhere, once this channel uses AMDSB-SC.
As the burst signal in this kind of encoder is kept unchanged, it wouldn't be useful.
Listen to me, the only way to have the video with the right polarity is to find out the key signal anywhere inside the baseband.
Sorry, but regarding a lot of experience I have had in this kind of anolog encoder, I'm pretty sure about what I'm saying.
NandoPG