CameraRGB and REDSPACE color spaces definition

[Jan Zahradníček]

Is it possible to obtain definiton of color spaces used by RED ONE (CameraRGB and REDSPACE)? Purpose is to create a truelight color profile for better dealing with RED footage in Baselight. Thank you.

[JonathanF]

I'll second that.

[John Tissavary]

Probably not going to happen... For better or worse, Red doesn't publish their sensor specs & color matrices the way some other mfgs do.

I think it would be of great help to the post industry if they did, better Red processing would be the net result.

[JonathanF]

Probably not going to happen... For better or worse, Red doesn't publish their sensor specs & color matrices the way some other mfgs do.

I think it would be of great help to the post industry if they did, better Red processing would be the net result.

We did this with film forever, you do it by recording known colors to neg, run it through the whole IN/IP/Print process and then run the print through a spectral densitometer.

You can easily profile stocks this way and profiling sensors is not much different in practice, in fact, it's probably simpler.

If the app you're reading your R3D data into has a floating point pbuffer then you can essentially do this yourself. You would want to have a way to generate specific colors and project them onto a transparency in front of the sensor to get your starting footage. Form there it's just a laborious slog to make the luts.

This isn't rocket science, well, maybe just a little, but if you've been around the block a few times you can figure this out.

[Stephane Labrie]

In our workflow whitepaper, we have described what we observed while comparing diffferent color spaces. Have a look at the Appendix A.
That could be handy.
http://images.autodesk.com/adsk/file...w_to_guide.pdf

[M Most]

We did this with film forever, you do it by recording known colors to neg, run it through the whole IN/IP/Print process and then run the print through a spectral densitometer....
This isn't rocket science, well, maybe just a little, but if you've been around the block a few times you can figure this out.

That would be true if you could get to the RAW data directly and thus directly measure the sensor response. Unfortunately, nobody outside of Red can currently do that.

[Kyle Mallory]

No, but you should be able to get a result that is "close enough" if you build your LUT (as Jonathan describes) from some baseline decode parameters, such as ISO 320, WB 5000K, 0 Tint, linear/flat curve, 1.0 RGB gain, and the given color space you are attempting to profile.

I'm not convinced that having the RAW/bayer-pattern data would be terribly meaningful because your color densities per pixel will vary depending on the debayer algorythm and which nearby bayer pixels are used to determine the current RGB triple.