Hey Andy, well no I haven’t figured it out entirely, but I’m getting closer. In my situation, I discovered a few potential problem sources.
1. I used studio tungsten lights mixed with ambient light from the window. Perhaps the two light sources made the camera’s sensors record askew. I also had the camera’s exposure on +6dB, which did introduce some unwanted noise. (I doubt that either of these are the case because my WB was manually set to a white board prior to every shot, and the strange flickering didn’t appear until AFTER compression to MPEG-2)
2. I recorded with my Canon GL2 in “Frame Mode” and captured as 30P. I’ve since discovered that you should still capture as 29.97. My field dominance was set to None when captured, yet the Item Properties in FCP insist they are “Even (Upper)” My sequence settings were also at Timecode 30, Field Dominance “none”, yet it won’t playback to my external TV monitor without having the Timecode set to 29.97. I believe this is because of the conflict with the file properties.
I just notice that MPEG2 in essence tries too hard to compress every pixel. Comparing Sorensen or H.264 at 320×240, I notice that if a part of the image is unmoving, the pixels are NOT refreshing. They are stagnant, as well they should be. The movement looks great, and the perceived quality is far better (even though it’s throwing away more information.) I just wish MPEG2 would do this. I’m not letting it stop me though. Starting this morning, our production team shot without the “Frame Mode” and leaving simply as 29.97 interlaced, and imported/edited the same. Also we made sure the exposure was +0dB. Though we haven’t yet finalized to MPEG2 (keeping our fingers crossed) I hope we’ll see better results.
If you find a solution that works for you, let me know. Till then, good luck.