Confused with Fits Files

Hi all,

I’m finding a strange situation when I use fits files produced using a DSLR. I use SGP to aquire images, and have found that SGP is happiest working with fits files and not RAW. However, when I use PixInsight to debayer the fits images I have to use a BGGR bayer pattern, and not the RGGB pattern that is correct for the camera and works with the RAW images. Is SGP mixing up the channels? It has been suggested that perhaps there is an offset row, however PI does not offer the option to use that.

Not sure if this is a bug or user error. If the latter some help would be appreciated :wink:


Did you try the Up-Bottom FITS option in PI in the preprocessing script? That might be the reason why the bayer pattern is ‘reversed.’ One thing I’ve been doing: saving both CR2 and FITS. I don’t remember exactly why but I agree that SGP seems happiest with FITS. I typically delete the FITS later, and use the CR2 in PI (or elsewhere).

I agree with Bhwolf. If you save both Raw and Fits files you will find most processing programs do better with true Raw files while SGP prefers Fits.

I have had a problem with the FITS files created by SGP. PI does not seem to like them, especially when it did the conversion of my other raw CR2 files (darks, etc…). I leave SGP set to save the CR2 files only but one night I made a mistake and have an entire run of SGP created FITS files. When I try to use these in PI with my darks, flats, bias, the stack image comes out terrible and blotchy. I have yet to find a way to convert that night’s files into something I can use. Does anyone know of a way to make them compatible with my other files. The second night on that target is done in CR2 and converted to FITS by PI.


SGP doesn’t really do anything “weird” with the FITs data off of DSLRs. We read the RAW data and place it in a FITs scaling it to 16bit.

It’s possible that the row data is different from other applications. There is no standard for FITs being top-down or bottom-up. Having said that we did go through a fairly lengthy testing process to make sure that the FITs coming out of SGP from DSLRs were similar to other capture applications where the row order was concerned.



If you have a moment, would you mind posting a FITS file from night 1 and
then a CR2 file from night 2? I would like to take a look first hand to
see if we can spot anything odd. Dropbox or similar would be fine.


Sure. No problem.

FITS CR2 samples

I renamed them slightly to indicate which is which. The FITS with SGP in the name was created by SGP on night 1. The CR2 is from the second night. I also included a FITS of the CR2 file which was converted by PixInsight.

If you look at the individual image itself, there is not a lot there because they are Ha images of M33. It sure takes a lot to show with this target.

Thanks for the help,

Thanks. Don’t worry about the image content… I will be having a look at
the pixel level so the only thing that matters is that they are at the same
ISO and exposure length (btw that is an assumption… is that true for
these images?).

Yes, they are the same iso and exposure duration. Nearly the same temp too.


Thanks all, seems like the best answer will be to leave the fits files for SGP use and the raw for processing. I will do what has been suggested and remove fits files later as both would eat disk space.


To @cmassa: I have found that for either FITS or RAW to work with PI, all the images/darks/flats/bias must be created the same way. I can’t mix FITS images with RAW darks/flats/bias. I think if you create a set of FITS using SGP for all the darks/flats/bias for that run, you will get a good result with PI.

That’s what I have found too. I just have one night’s worth of lights converted to FITS by SGP. I don’t have any calibration files done in that way. I was trying to find if there is a way to convert them to something compatible with my existing files. I usually do everything in PI and have SGP save the CR2 files but I messed up that one night.