Ah -- I think I might have misunderstood your goals. I thought you were going for a close colorimetric match from your camera. If, instead, you're looking to replicate the ``look'' of the camera's on-board RAW developing engine...I suppose your general approach is probably as good as any. I think you might be happier, though, manually tweaking the matrix until you get results you like. There's only nine numbers, after all, and it shouldn't take much experimentation to get a good feel for how they interact. A word of caution, though...be sure to do your experimentation on a wide variety of original images -- a true representative sample, including scenes with different types of illumination. It'd be very easy to come up with the perfect matrix for one single scene but to be completely unhappy with how it renders any other scene. Cheers, b& On Feb 11, 2013, at 5:09 PM, David Irisarri <zuiko3000@xxxxxxxxx> wrote: > So if you don't create any reference file with the colors you like, > how do you calculate the color matrix based on the uncorrected sensor > color and the color you like? > > I would appreciate your help. > > David > > On Mon, Feb 11, 2013 at 7:06 PM, Ben Goren <ben@xxxxxxxxxxxxxxxx> wrote: >> I take it what you really want is that matrix so that you can compile a >> custom version of dcraw? If so, I don't think you want to do that the way >> you're describing.... >> >> b& >> >> On Feb 11, 2013, at 4:49 PM, David Irisarri <zuiko3000@xxxxxxxxx> wrote: >> >>> Hi Graeme, >>> >>> Thanks for your reply. >>> >>> I want to do this. >>> >>> 1. Take a picture of the Color Checker chart with the camera and >>> export it in TIFF AdobeRGB (biggest color gamut of the camera, no >>> color patches out of gamut) >>> 2. Extract color values of these patches and create a CGATS file >>> according to these values. I want these values in L*a*b and XYZ. >>> 3. Process the same image with dcraw (no color management raw color) >>> and create a color matrix using the reference previously created. >>> >>> David >>> >>> >>> >>> On Mon, Feb 11, 2013 at 6:35 PM, Graeme Gill <graeme@xxxxxxxxxxxxx> wrote: >>>> David Irisarri wrote: >>>> >>>> Hi, >>>> >>>>> I ma David and I am trying to create a color profile. I would like to ask >>>>> you guys how to extract color values from an image in order to create a >>>>> CGATS file. I am using scanin but it only generates a val file with RGB >>>>> values. >>>> >>>> If you want to interpret the RGB image as CIE values, then you >>>> have to specify the device space using an ICC profile. You >>>> can use the "scanin -c" mode to do this, see >>>> <http://www.argyllcms.com/doc/scanin.html#c> >>>> >>>>> The image I have is in aRGB and at least I would need to extract >>>>> these values in CIE L*a*b. XYZ would also be very welcome. >>>>> I need to extract these values because I want to create a profile >>>>> according >>>>> to this TIFF image, I don't want to match the colors according to a chart >>>>> with >>>>> its own values. >>>> >>>> The RGB TIFF is being used as the reference values ? >>>> >>>> Graeme Gill. >>>> >>>> >>> >> >> >