Negative values on the histogram - what do they mean?

[magnusl]

Hi!

I'm getting some negative values at times in the histogram below the image when doing data processing with the DP module, see image



What do these negative values mean, really? Should I take them as an indication that my calibration process is problematic (for instance, that dark has been subtracted twice), despite the images looking good on visual inspection?

Magnus

[karenacollins]

It is possible to have legitimate negative pixel values after calibration, but I think it would be wise to make sure your bias and dark values are not being subtracted twice, or that a pedestal value that is inserted by some camera control software is not somehow getting subtracted twice. If you are running all of your raw calibration and science images through AIJ's DP module to produce the calibrated images, incorrect subtraction shouldn't be happening. However, if you are using a combination of calibration from other tools and AIJ, it can happen. In particular, if you are producing master calibration files from other tools and using then in DP, then one must exercise care that bias, dark, and pedestal subtraction only happen once.

Karen

[magnusl]

Hi!

Thanks.

OK, then there is something wrong with my process - fully within AIJ.

I have master bias, produced by the DP module, median checked.

I have master dark, produced by the DP module. deBias checked when producing the Master. Should it then be deselected when reusing the master Dark? I don't see much difference.

I have flats, also produced by the DP module. Median checked and Remove Gradient.

And with these settings and files, I get -7.739 for some images right now.

The Bias I use has a mean value of 370, and the dark varies between 3.5 and 191,7, if I read correctly.

The flat, when opened in AIJ, gives me values between 0.8510 (min) and 1.4992 (max). Which feels odd. When measuring in PixInsight I get a minimum of 221 and a maximum of 56983 (16-bits), with a mean of 15064.


What am I doing wrong here? And how come the values for the flat is so strange - or are they not? I'd be happy to share the files with you in a Dropbox folder?

Best,

Magnus

[karenacollins]

Hi Magnus,

If deBias was enabled when producing the dark, it should be checked when using the dark at a later time too. If deBias was on when you produced the master dark, but you turned deBias off when using the dark, then you should see an average difference in your calibrated images of 370 (i.e. the mean value of your Bias file). As long as you have deBias on or off when you create and use the master dark, then you should get the same result, although if your dark exposure time is different than your science exposure time, you should use the deBias option.

AIJ normalizes the flat to have a mean value of 1.0, so the range 0.851 to 1.4992 sounds reasonable. According to your information below, it seems that PixInsight is not normalizing the flat, which would explain the difference.

I'm not clear what you mean by "I get -7.739 for some images right now". Is that the lowest pixel value in the image? If so, that's not surprising, and I'm not sure there is a problem with your calibration process.

Karen

[magnusl]

Hi!

Thanks. OK, so I understand the flat, that sounds OK. And you are correct (sorry for being unclear): I get - 7.738" meant this value as the lowest. But then why do I have a negative value as the lowest? Sometimes I even get a far lower value, a few thousands even, but the "black" value is still slightly above 0.

So: It seems my calibration is OK then, just not clear on how I can get a negative value at all there.

Magnus

[karenacollins]

It is possible that after bias and dark subtraction, bad/dark pixels can go negative, and in the case of low sky background, random bias or other variations can cause small negative values. In general, the mean of your calibrated science image should be close to the mean of your raw science image minus the mean of your master bias minus the average value of your master dark. If your camera software adds a pedestal value, you'd need to subtract that amount too. If you find this to be the case, then I think your calibration is probably okay.

Karen