R/G/B gain -- makes no sense at DVinfo.net

Bill Edmunds · July 21st, 2006, 11:25 AM

Where I come from, Red and Cyan are opposite colors. Green and Magenta are opposite colors. Blue and Yellow are ooposite colors. But in the HD100, G gain increases BOTH green and megenta. B gain increases BOTH blue and yellow, etc. HUH?!?!?! This seems completely counter-intuitive and seems to defeat the whole notion of color adjustment.

Put another way... if I want to increase yellow, I DON'T want to increase blue.

Nick Jushchyshyn · July 21st, 2006, 11:54 AM

Yellow = R + G

The primary colors of emitted light are Red, Green & Blue
Cyan = Blue + Green
Magenta = Blue + Red

Try it with any computer program that lets you define a color with RGB values.
To get "more" yellow, you can either increase Red and Green .... or decrease blue.

Stephan Ahonen · July 21st, 2006, 06:22 PM

It makes more sense if you look at it on a vectorscope. The camera's not doing these color transformations in RGB color space, but rather in YUV color space. Watch the vectorscope, when you twiddle the R gain control, it's actually scaling the entire vectorscope plot along the axis defined by the color red, so colors on the opposite side of the vectorscope from red are affected as well.

Counterintuitive? You bet.

Stephen L. Noe · July 21st, 2006, 06:43 PM

Find some information on floating white point. It seems counterintuitive however it keeps the entire color space inside the gammut of ITU709. The opposite would be absolute white point which would affect a color individually without affecting the other colors (when adjusted).

These are two different ways to adjust color in relationship to white.

Paolo Ciccone · July 21st, 2006, 10:41 PM

Bill, as Stephen has suggested you have to look at the gain using a Vscope. In that way you'll see that not only the gain of Red, for example, affects blue, but also that by increasing the gain you cause a shift in the rotation. Take a look at my articles about the TrueColor configuration, there are some snapshots of a Vscope. The easiest, IMHO, way of making sense of it is to use a DSC chart and a Vscope and tweak the gain *and* rotation.

July 21st, 2006, 11:25 AM	#1
Bill Edmunds Major Player Join Date: Aug 2003 Location: Conway, NH Posts: 574	R/G/B gain -- makes no sense Where I come from, Red and Cyan are opposite colors. Green and Magenta are opposite colors. Blue and Yellow are ooposite colors. But in the HD100, G gain increases BOTH green and megenta. B gain increases BOTH blue and yellow, etc. HUH?!?!?! This seems completely counter-intuitive and seems to defeat the whole notion of color adjustment. Put another way... if I want to increase yellow, I DON'T want to increase blue.

July 21st, 2006, 11:54 AM	#2
Nick Jushchyshyn Major Player Join Date: Feb 2004 Location: Philadelphia, PA, USA Posts: 548	Yellow = R + G The primary colors of emitted light are Red, Green & Blue Cyan = Blue + Green Magenta = Blue + Red Try it with any computer program that lets you define a color with RGB values. To get "more" yellow, you can either increase Red and Green .... or decrease blue. __________________ Nick Jushchyshyn Matchmoving, Compositing, TD imdb

July 21st, 2006, 06:43 PM	#4
Stephen L. Noe Trustee Join Date: Apr 2005 Location: Chicago Posts: 1,214	Find some information on floating white point. It seems counterintuitive however it keeps the entire color space inside the gammut of ITU709. The opposite would be absolute white point which would affect a color individually without affecting the other colors (when adjusted). These are two different ways to adjust color in relationship to white. __________________ Advanced Avid Liquid Training found Here

July 21st, 2006, 10:41 PM	#5
Paolo Ciccone Trustee Join Date: Apr 2005 Location: Santa Cruz, CA Posts: 1,116	Bill, as Stephen has suggested you have to look at the gain using a Vscope. In that way you'll see that not only the gain of Red, for example, affects blue, but also that by increasing the gain you cause a shift in the rotation. Take a look at my articles about the TrueColor configuration, there are some snapshots of a Vscope. The easiest, IMHO, way of making sense of it is to use a DSC chart and a Vscope and tweak the gain and rotation. __________________ Paolo http://www.paolociccone.com Demo Reel

July 21st, 2006, 06:22 PM	#3
Stephan Ahonen Major Player Join Date: Apr 2005 Posts: 512	It makes more sense if you look at it on a vectorscope. The camera's not doing these color transformations in RGB color space, but rather in YUV color space. Watch the vectorscope, when you twiddle the R gain control, it's actually scaling the entire vectorscope plot along the axis defined by the color red, so colors on the opposite side of the vectorscope from red are affected as well. Counterintuitive? You bet.