So y ou're saying 12 bit fixes moiré patterns on the 5D2? very excited to watch the video and test it out...
This means you have somehow changed the binning as a side effect?
No I don't think so , all I change was the
write channel , normally I have very little to no problems with moiré patterns even with h264 .
I'll try to explain my logic : I believe the main issue (be on line skipping) is motion blur , sharpness & contrast .
For is reason I only use technicolor cinestyle picture style profile for Liveview preview (flat) , I correct my w/b & expose
to that profile . I try to be within f4 to f8 , one thing I do is to expose for shadows instead of exposing for hi-lights .
I usually find most moiré patterns are a issue in the hi-lights (sometime not always)
I know I can always recover hi-lights (with 12+ stops) 99% of the times , drag the shutter a bit (1/43th-ish) , I make sure I have at least a
3 point light setup with more light power (like the sun
) that I need (in the case of the green screen , third light was over top of me instead of behind me)
so I can run lower ISO's (100 , 200 but never over) . I like to run a little soft on the focus (instead of turn down sharpness in "cinestyle profile)
I use canon "L" lens so not sure if that help too , but what I thing is really helping is the 12bit black level range
So I have to wonder , does 12bit still come from the 14bit silent image 1880x1250 ? or is there another pipe off
the A/D convertor to 12bit & 10bit or a recalculation of the 14bit ? but is really 14bit ?
I know in some of my very early tests with the bit reduction code , I implemented it in the old ml core .
I was abled to get 16bit to save test files , but couldn't decode it .
If you compare 14 to 12 for black levels 1792 to 448 , It seem to be the color sweet spot , can't explain it why just is !
I least to my eye's , You tell me for yourself @Ilia3101 , I'm I wrong ?
I need to do a long 10bit test to see if it's as clean & clear of issue like moiré patterns
