Monday, November 02, 2009

Color Grading Canon 7D

Colorist and Author Robbie Carman, has posted a couple of comments to the Cinema5D forum about grading DSLR footage, specifically the Canon 7D:

… 7D footage does appear to have an initial contrast ratio that is less than the 5D. […] on my scopes the 7D does have an elevated trace of about 10-15%/IRE given the exact same exposure settings between cameras.

[…] The wild thing is that the difference appears to be only in the blacks in and midtones. This observation leads me to believe that the 7D is in fact applying a luma adjustment to midtones/blacks in camera, almost like a reversed S-Curve

Both cameras suck when it comes to overall color latitude especially compared to RED native clips, but I guess that’s to be expected due to the 4:2:0 chroma subsampling. This has been especially frustrating for me getting good HSL keys for secondary corrections like skies, skin tone etc. But is no worse then HDV, XDCAM or HDCAM for that matter.

He also felt that the sharpness of the two cameras was virtually the same-at least visually-once the contract was adjusted, and speculated that people comparing the cameras and judging the 7D as being softer were perhaps being mislead by this contract difference.
After correcting the footage I cannot tell the difference between the two cameras from a contrast/color point of view the 5D of course does produce a shallower dof.

In a second post, he made some interesting comments about Network QC departments and whether they are willing to accept footage shot on DSLRS. His comment that most are still tape based was especially interesting:

I'm grading a show […] the production company has to "lie" to the network to get footage accepted because they have stuff that was shot HDCAM SR, DSLR, Betacam SP (yikes!) and high speed footage from the phantom. But as far as the DSLR footage.

1. H.264 is generally accepted as a deliver codec but not acquisition. So any format that uses it QC departments are scared of, for fear of it having big issues. Also network engineers are pretty strict to jitter and other technical issues they can see on waveforms and other display types. The weird part to me is HDCAM SR although at higher data rates also uses an MPEG 4 variant so why not DSLR variants?

2. I will say that rolling shutter is an issue for many QC departments. In fact some new QC specs I have read in the past couple weeks have basically said whip pans or fast movement with CMOS based sensors are not advised
Your best bet to have acceptance across the widest range of QC standards [is to] download the PBS standards and adhere to those...most networks follow those anyway

No comments: