I am shooting some interviews in the next week or two and just as a technology experiment will shoot them RGB 444 to Pix in ProRes 4444 (Talk about overkill... this is really a DSLR job, but I'm curious).
I don't think an interview will really capture the difference, or potential difference but I guess it would depend on the lighting and what kind of highlights are in the shot. IMO highlights are the dead give-away to the "video look" so if there's anyway you can swing it and get a window with a wide outside view in the shot that might be great to see. Maybe during a run-through or something? And also do three takes each time stopping down 1 stop to add extra highlight protection to see how the differences turn out in the grade restoring mid-tones.
In the meantime I'm going to contact Tata here from DVX who lives near me and see if we can't do a shoot out with "cinematic lighting", a model, and the F3 in the following modes/formats:
I even asked the folks over on the Canon threads if they have tested the PIX in 10-bit RGB but apparently you need two recorders to do it?!? Not sure what that's all about. I left quickly and didn't ask any questions. :huh:
For the F3 if anyone has any tests for consideration let me know what they'd be interested in seeing.
just what you're describing, i think - for a given frame that pushes the camera's latitude, just what can be achieved in each bit depth and color space.
I shot a series of videos for Fairmont Hotels last year. This one was shot with the F3, RGB to Gemini, DPX converted to Pro Res 4444, production company edited in FCP, and was graded in a Davinci Resolve suite. I was not able to sit in on this grade. My other Mazda shoot that we did the same way has not been released yet.
We have seen the difference grading Pro Res 4444 from converted Gemini DPX files, compared to Nanoflash 422 XDcam in the Davinci suite on our Fairmont spots and Mazda videos. There was a lot more room for the colourist where there was not so much noise introduced, as well as ability to isolate colours, and digital artifacting. The S log when recorded in RGB also looked flatter to my eye.