Matching gain for different chips in mosaic images
Frank Winkler wrote on Jun 26, 2009
I'm processing data from a mosaic camera (IMACS on Magellan I) where the gain is significantly different from one chip to another. Ultimately I am trying to make a single image by putting the pieces together and stacking a set of several dithered exposures.
I'd like to scale all the 8 pieces from the individual chips of one exposure onto one common intensity scale--so that the same amount of light has the same number of counts no matter which chip. The usual flat fielding normalizes the flat for each chip by the "ccdmean" parameter and so preserves (approximately) the stats in an image once it has been flat-fielded. So if the gain of the different chips is different, those differences (in sky level, say) from chip to chip will be present in the raw images, and also in the flat-fielded ones. [Note that I am NOT talking here about the problem of variations in intensity scales because the effective pixel size is different at different distances off-axis.]
I can think of one or more kluge techniques to deal with this problem, but it must be a common one and I suspect there is some more elegant way to deal with it than I'm likely to come up with on my own. I just haven't found it yet and would appreciate it if someone else has.
Thanks,
Frank Winkler
I'd like to scale all the 8 pieces from the individual chips of one exposure onto one common intensity scale--so that the same amount of light has the same number of counts no matter which chip. The usual flat fielding normalizes the flat for each chip by the "ccdmean" parameter and so preserves (approximately) the stats in an image once it has been flat-fielded. So if the gain of the different chips is different, those differences (in sky level, say) from chip to chip will be present in the raw images, and also in the flat-fielded ones. [Note that I am NOT talking here about the problem of variations in intensity scales because the effective pixel size is different at different distances off-axis.]
I can think of one or more kluge techniques to deal with this problem, but it must be a common one and I suspect there is some more elegant way to deal with it than I'm likely to come up with on my own. I just haven't found it yet and would appreciate it if someone else has.
Thanks,
Frank Winkler
Francisco Valdes wrote on Jun 26, 2009
Hi,
When I first started working with mosaic data I had to learn that any normalization applied to a flat field had to be the same for all detectors. This then evolved to special features in tasks like mscred.ccdproc that handle flat fields appropriately.
So what I think you are asking about is that to match detectors in a mosaic with different gains (which is always the case), this is done by the dome flat fielding process but it is important that all elements of the dome flat use the same normalization and not, say, the average of the individual image.
I hope this addresses your question.
Yours,
Frank Valdes
When I first started working with mosaic data I had to learn that any normalization applied to a flat field had to be the same for all detectors. This then evolved to special features in tasks like mscred.ccdproc that handle flat fields appropriately.
So what I think you are asking about is that to match detectors in a mosaic with different gains (which is always the case), this is done by the dome flat fielding process but it is important that all elements of the dome flat use the same normalization and not, say, the average of the individual image.
I hope this addresses your question.
Yours,
Frank Valdes
Frank Winkler wrote on Jun 26, 2009
My original data from IMACS on Magellan has 8 separate files (one per chip) rather than a single multi-extension fits file. For reasons that may be spurious I processed the data through flat fielding as individual files, so each had its own normalization, and combined them into a single MEF file later. To get all normalized to a common level I ended up doing a dumb little script to rescale each chip to a common level (determined from the flats)--so I effectively did what you are suggesting, but perhaps with more work than had combined into MEF files at the start and used mscred.ccdproc from the outset.
thanks,
Frank/Paco
thanks,
Frank/Paco
Last post on Jun 26, 2009