Did a bit more trial and error testing with the latest nightly, Six Thirty TL2.0 and a newly compiled bin from 1%'s repo.
Settings using SanDisk 45Mb card
1st test
960X540
1080 @ 24fps
GD off
Nightly build = 500 frames w/34 pink frames
TL2.0 SixThirty = 500 frames w/3 pink frames
Newly compiled from1% repo = 500 frames w/29 pink frames
2nd test
960X540
640 @ 15fps (fps override)
GD off
Nightly build = 500 frames w/44 pink frames
TL2.0 SixThirty = 500 frames w/22 pink frames
Newly compiled from1% repo = 500 frames w/39 pink frames
I've read that changing pic quality to 640 & lower fps will yield better results, but not so in my test.
The SixThirty build seems to be the most efficient of the three. I don't understand why the newly compiled build from 1% repo is almost as bad as the nightly, unless some changes in the unified code was pushed to 1%.
I guess the best option is to try and improve the h.264 with higher bitrates, as raw just won't cut it for commercial use. Either that or just get a better camera
