On Friday 19 October 2007 08:04:07 [EMAIL PROTECTED] wrote: > On 10/18/07, Michael Shneier <[EMAIL PROTECTED]> wrote: > > I am a new lcms user. I want to calibrate the color of a camera for use > > in image processing. > > I have a 2c to add as well, for what it's worth: > > * In my opinion the IT8 is not very suitable for camera calibration as > the patches are formed from combinations of only three dyes. I prefer > the Macbeth chart, where (I think) they use mixes of 18 dyes to make > patches with spectral curves similar to those you find in real > objects.
This can be an issue in light that does not have a smooth distribution of light across the spectrum. So for profiling in florescent lighting in particularly it might be advantageous to use one of the GretagMacbeth targets since florecent light has significant spikes in its spectrum. But the difference even under lighting conditions that are less than optimal for using an IT8 target are fairly small although visible in a side by side comparison. > > Gretag sell a postcard-sized 'mini-macbeth' (the standard one is about > A4) which is small enough that you can include the target in your > scene without obscuring too much other stuff. The 24 patch targets are not well suited to profiling and should be avoided if possible. These targets simply do no have enough patches to do anything other than a matrix/shaper profile. If you are going to use a GretagMacbeth target it should be either the ColorChecker DC or the ColorChecker SG. These targets are considerably more expensive than a high quality IT8 target at about 5 times the cost for the same size target. You can get an A4 sized IT8 target from Wolf Faust for about $40 including shipping to the US and you will pay around $200 for one of the (more than 24 patch) GretagMacbeth targets. You also need to keep in mind that targets need to be replaced every two to three years unless you have the ability to remeasure the target (IE. you have a reflective spectrophotometer). So the cost difference over the long term is considerable. In short IT8 targets do have some issues when used in some types of lighting and users who work in lighting with spikes in the spectrum should consider using other targets with the understanding that these targets will be considerably more costly. For most users a good IT8 target will do the job and be considerably less costly. By the way LProf CVS now supports all of these profiling targets including the ColorChecker 24 patch targets, as well as the EyeOne Scanner 1.4 and Hutch Color (HCT) targets. The HCT, ColorChecker SG and ColorChecker DC work was sponsored by Google Summer of Code 2007(tm). > > * Camera calibration is actually very simple. All you need to do is > remove the camera gamma (if any), fix the black point, and use a 3x3 > matrix to go to XYZ. You can find the matrix using least-mean-squares > from the Macbeth colours, or if you are feeling brave, do a non-linear > optimisation to get a matrix that is minimal with respect to errors in > CIELAB. You can do this in just a few lines of matlab (or equivalent). Again this is not a calibration this is a characterization that is being created. > > * Getting nice, repeatable results is hard though :-( as other posters > have said, things like lighting are really difficult. You'll find > cameras are rather sensitive to temperature, illumination uniformity, > lens vignetting, aperture, the list goes on and on. > > A particularly awful gotcha is scatter in the optics. You'll find a > white patch next to a black patch 'bleeds' a lot of light into the > black, significantly altering its colour. This is a very serious > problem with trying to use any lens to do absolute colorimetry. This is particularly true when you try to use a small target image (IE. the target is only a small part of the whole image) and for cameras with lower quality lenses. > > I have a HOWTO on my website: > > http://www.vips.ecs.soton.ac.uk/index.php?title=Colour_calibration_with_nip >2 > > nip2 is quite fun for playing with this stuff. It has an LCMS menu, so > you can also import and export with ICC profiles. > > John > > ------------------------------------------------------------------------- > This SF.net email is sponsored by: Splunk Inc. > Still grepping through log files to find problems? Stop. > Now Search log events and configuration files using AJAX and a browser. > Download your FREE copy of Splunk now >> http://get.splunk.com/ > _______________________________________________ > Lcms-user mailing list > Lcms-user@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/lcms-user ------------------------------------------------------------------------- This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now >> http://get.splunk.com/ _______________________________________________ Lcms-user mailing list Lcms-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/lcms-user
