Calibration of DSLR images in photometry

American Association of Variable Star Observers (AAVSO)
Mon, 05/09/2022 - 18:58

Hi, I am new to photometry and I am trying to read up on it. To this end I have read this AAVSO DSLR photometry manual:

In this manual, I have read about image calibration, and the section below makes me confused:

"Calibration is required to correct vignetting and dust shadows, uneven pixel sensitivity and various
sources of noise. The master calibration frames must be applied in the following order to ensure
systematic effects are properly removed:
1. Create master bias, dark and flat frames
2. Subtract master bias from master dark and master flat, and all science frames.
3. Subtract master dark from all science frames, (but not from master flat since flat
exposures are usually only a few seconds therefore dark correction is not necessary).
4. Divide the normalized master flat into all science frames."

In particular, it is adviced in step 2 above that a master bias should be subtracted from both the master dark and the science (light) frames. In step 3 the master dark is then subtracted from the science (light)  frames. I feel that these instructions are different from common practice in general puropse astrophotography, and also they seem illogical to me. In the most commonly used software for astrophotoghraphy, PixInsight, the standard procedure for image calibration does not include bias subtraction from neither the master dark, nor the light frames. This seems more logical to me. The above procedure seems to imply that it is assumed that the bias signal is present as a part of both the master dark frame and the science images, since it suggested to subtract it from both. But if  this is the case, the bias signal will get subtracted from the science image also by just subtracting the uncorrected master dark from the uncorrected science image. The bias subtraction from both is an unnecessary step.

Also, in for example the PixInsight manuals, it is adviced against subtracting bias from the master dark, since this may cause negative values in the master dark, which may get clipped, and so cause data loss.

I would like to have your comments on what is a correct procedure for image calibration in DSLR photometry, and in particular how bias images should be used.




Variable Stars South (VSS)
Callibration of DSLR images in photometry

I will leave the authors of the AAVSO DSLR Observing Manual to answer your specific questions.

However, I would like to comment on subtraction of the master bias frame from the master dark. As I understand it, the reason for doing this is to allow the use of scalable darks.

It seems to me that strategy (using scalable darks) is possible when the dark current of the camera increases linearly with increasing exposure time. However, dark current of some DSLR cameras does not behave in this manner according to Craig Stark. The reference is:

"Profiling the Long-Exposure Performance of a Canon DSLR by Craig Stark. Note: Originally published on Cloudy Nights, 7/2/12. Minor edits in place"

The issue is captured in the following quote from the article:

"If you try to measure the dark current (of some Canon DSLR cameras), you end up at your wits end as increasing the exposure duration would by and large either do nothing to the average image intensity or it would decrease the average image intensity. Now, longer exposure times decreasing the image intensity sure is an odd thing and sure points to internal processing of the data before it hits the CR2 file."

I tested my own (somewhat old) Canon DSLR camera and found that it behaved just as Craig Stark describes. 

It therefore seems to me that scalable darks should not be used for images taken with such cameras. Other models of DSLR camera may behave differently. Craig Stark's article describes how to test any camera for dark current behaviour.


DSLR calibration

Hi Ulf,

Thanks for your email, I hadn’t seen your post. It's been quite a few years since I rewrote the DSLR Manual so my recollection may be a little hazy. Roy’s comment is important, DSLR dark current, at least for the Canon cameras I tested, does not behave the same as CCD dark current. CCD darks can be scaled for different exposure times, but DSLR darks cannot be scaled. If your science image exposure is 30 seconds, then your dark frames must also be 30 seconds.

In step 3 of the calibration steps, I said that the master dark should not be subtracted from the master flat because the flat exposures are usually only a few seconds. The implication is that dark current doesn’t have time to accumulate to a significant level. However, the bias is always present in the flat frames. So, the master bias must be subtracted from the master flat.

To avoid subtracting the master bias from master dark and master flat frames, as you suggest, you would need a second master dark made from darks with the same exposure time as the flat frames. But that requires an extra set of images to make the second master dark. However, these short exposure dark frames are essentially just bias frames, so why not just use the master bias?

I’m not familiar with PixInsight, but if its calibration procedure is as you stated, then two master darks would be necessary for DSLR calibration.

I use MaxIm DL. To avoid negative values in calibrated images and master frames it adds a fixed value (usually set to 100) to every pixel, which they call the “pedestal”. At each step in the calibration process this pedestal value is taken into account.

I hope this answers you concerns about the DSLR calibration process. Cheers,