From: Bart Declercq Sent: Thursday, April 18, 2002 10:54 AM To: ccd-astro-imaging@yahoogroups.com Subject: Re: [ccd-astro-imaging] M81 & Colour Wheel (Part III) - Calibrating RGB Part 2 What if you took different length exposures for R, G, B ? -------------------------------------------------------------------- If you took images through different colour filters, but not in the calibrated way described in my previous message, then you can still rescue the result, but you need to take care of the steps you use: 1st Step : Normalise the images --------------------------------------- Before doing any other steps (other than dark-frame & flat-fielding), we need to "simulate" identical exposure times: Say the Red image was exposed for 15 minutes and the blue image for only 5 minutes. To normalize the two images, you should either multiply all pixels in the Red image by 0.333x or the Blue image by 3x (in my previous message, I mentioned removing the background, don't do this yet!!!) 2nd Step : Normalise the background ---------------------------------------------- Only after you normalised the images should you do this step (as described in my previous message) 3rd step : calibrate the images using the measured RGB ratios -------------------------------------------------------------------------- See previous message Voila, again a calibrated RGB image. Of course, if the exposure times are very different, one of the colours might be much noisier than the others, and this can result in worse image quality and strange colours. Also, if in one of the colours, a star is overexposed, and it isn't overexposed in the other colours, you can get some strange colour fringes, because the colour information is incomplete. Ideal situation : there's a known G-type star in the field of view ----------------------------------------------------------------------------- Because of atmospheric extinction, the calibration data is only accurate for the altitude in the sky the G-type star was when you imaged it, so you could get surprises if you shot Capella in the Zenith and use that data to "calibrate" for a Sagittarius object 15° above the horizon. The most complete solution would be to take calibration measurements at different heights and use the correction depending on the height, but this isn't always practical. Life is a lot easier when there is a star in the field of view that is not overexposed and is of spectral type G (F or K will also work, but is slightly less accurate). Many CCD-manipulation programs will allow you to measure a star's magnitude in the image. A G-type star should have the same magnitude through all three colour filters. First make sure you normalise the background to near 0. Then measure the magnitude of a G-star in all three images. You'll get a measured magnitude (it doesn't have to be the correct magnitude, just the value as measured by the program). You might get mag. 10.00 for the Red image, mag. 10.08 for the Green image and mag. 10.34 for the Blue image. Now experiment a bit, multiply the Green image by a factor close to 1.1 and vary it until you get a measuremenet of 10.00 as the magnitude in the Green image. Do the same for the Blue image (you'll probably need a bigger factor). Now you have calibrated directly on the star and the image should be OK. There is one important warning here : be very careful with this on stars near Nebula complexes, sometimes a G-type star will be shining through a dust cloud, which will absorb lots of blue light, leaving the star with an unexpected lack of blue. This will then impact your calibration quite severly (this extinction can easily be more than a magnitude, meaning that your correction would be of several hundred percent!). For some examples (with CMY filters, but the concept is the same), check out the CMY calibration article on my website http://www.bart.declerq.easynet.be/