You're walking into a can of worms here. There are level mapping issues where the 709 spec maps the black point and white point in from the extremes of RGB to preserve headroom. That is what you are thinking about.
Then there is the sampled color of recorded "RGB" on tape which is 4:4:4 vs. components video at 4:2:2. Thus adding twice the chroma info.
Then there is RGB color space of which there are quite a few flavors. Some examples here:
<<http://en.wikipedia.org/wiki/RGB_color_space>>
When you create an RGB project in Avid, you are working with the 4:4:4 vs. 4:2:2 issue. So you have twice the color space. It doesn't mean that suddenly blacks and whites are remapped to RGB levels vs 709.
--- In Avid-L2@yahoogroups.com, "johnrobmoore" <bigfish@...> wrote:
>
> What does capturing Smpte bars look like on the scope. I haven't dealt with 444 but what is the spirit of RGB 709? I think of RGB as full 0 to 255 or 0 to 1023 and 709 as 16 to 235 or 64 to 940. Clearly I'm missing something about these projects. What does the AJA Log RGB codec to the footage. Does that act like a lut to make linear our of log or does the footage look washed out like log footage does sometimes?
>
> --- In Avid-L2@yahoogroups.com, Jay Mahavier <jay_mahavier@> wrote:
> >
> > For the past few projects we have off-lined in Avid Media Composer and then did our conform in FCP7. Our sources have been HDSR tape, 1080 23.98 444 RGB with a Log source. Panavision Genesis or Arri Alexa. When doing element capture or the conform we would use the FCP7 system with a Kona3 card. Capture using the AJA Log RGB codec, full range (0-1023). Bottom line of that was pixel in = pixel out. What I digitized was a match to what was on tape. I could digitize, insert edit back to the tape, and then digitize again and it was a pixel to pixel match on the resulting digitized footage.
> >
> > So I was interested in how MC6 would compare. I set up a test system running MCv6 on OSX 10.7.2 with a Kona3 card using AJA drivers v10.0.1.
> >
> > I digitized a few shots in the original FCP7 system as a base line. Then on the MC6 Test system I used the AJA VTR Xchange application to capture footage. Different CPU, different Kona card, and different source SR deck then I used for the FCP 7 system, but same source tape and same section of footage. The footage matched between the two systems.
> >
> > Then I launched MCv6. Made a new Avid project. 1080/23.98 RGB. I did capture test with 1:1, DNxHD, and ProRes4444. And none of them match the footage from either the FCP7 system or the test I did on the Avid system using AJA VTR Xchange. I think the problem is that the Color Space setting for the Avid project is RGB 709. The 709 part being the problem. To capture Log footage I need full range RGB capability. Rec 709 just won't do. It looked like what MC6 was doing was actually taking 64 (16) and moving it to 0 and taking 940 (235) and moving it to 1023. I think that's what's going on. At least when I start pixel probing the values and looking at histograms that's what appears to be happening. But maybe I have a setting wrong somewhere in MC6.
> >
> > I'm not sure if there are plans to fix this or not. But I sure hope they do. I would really like to be able to do my element capture and conform on Media Composer.
> >
> > If anyone has any observations about this they would like to share I would like to hear them.
> >
> > Thank you,
> > Jay
> >
>
No comments:
Post a Comment