Low gain darks not read by PI

[joelshort]

I recently purchased a fast (f2.77) 7" scope and I want to use my QHY163M at a low gain in order to maximize dynamic range and full well capacity.  I have tried taking darks at gain 0 and gain 5 and when I try to integrate the darks into a master PI gives me an error message:

"Zero or insignificant signal detected (empty image?)"

There is data in the image but I am unable to make a master dark.  What setting do I need to change to allow integration?

Edit:  It appears as though exposures greater than 5min stack fine in PI.  But 5min is too long with this camera, especially with such a fast scope.  How do I integrate shorter exposures at low gain?

[pfile]

do you have weights set to "Don't care"? and normalization set to "no normalization"?

rob

[joelshort]

do you have weights set to "Don't care"? and normalization set to "no normalization"?

rob

Yes on both counts.

[bulrichl]

Did you also try the integration of the darks with 'Evaluate Noise' disabled?

Bernd

[joelshort]

Did you also try the integration of the darks with 'Evaluate Noise' disabled?

Bernd

Evaluate noise was also disabled.

[bulrichl]

Then I guess it is advisable to upload a set of 3 dark frames which produce this error message in order that Juan can take a look at it.

Bernd

[joelshort]

Then I guess it is advisable to upload a set of 3 dark frames which produce this error message in order that Juan can take a look at it.

Bernd

Here you go:  https://drive.google.com/file/d/1y76xCk3LY9X3W_lg_ZmbiA2XAPSYI6TL/view?usp=sharing

[bulrichl]

With 'No Pixel Rejection', these darks are integrated flawlessly. However when I set e.g. 'Percentile Clipping', the error message appears.

On further inspection it turns out that dark 022 has a totally different median (= 4720 ADUs) than darks 020 and 021 (= 620 and 624 ADUs respectively). I guess this is the culprit. You could load all darks with 'Blink', output a Seies Analysis Report and take a look whether there are any more outliers. Exclude the deviating darks from the integration. Hope that helps.

Bernd

[joelshort]

With 'No Pixel Rejection', these darks are integrated flawlessly. However when I set e.g. 'Percentile Clipping', the error message appears.

On further inspection it turns out that dark 022 has a totally different median (= 4720 ADUs) than darks 020 and 021 (= 620 and 624 ADUs respectively). I guess this is the culprit. You could load all darks with 'Blink', output a Seies Analysis Report and take a look whether there are any more outliers. Exclude the deviating darks from the integration. Hope that helps.

Bernd

Dark 22 isn't the problem, although that is an errant one that I will remove.  I have 29 other darks and all of those others ones have the expected 620 ADU yet they do not integrate with my standard settings.  However I just discovered that if I set the Pixel Rejection (1) to "No Normalization" (with Windsorized Sigma Clipping) then the darks integrate.  Is that advisable?  I need some form of rejection to eliminate gamma ray strikes etc.

[bulrichl]

Yes, for the integration of darks, flat-darks and bias frames the setting 'Normalization' in the 'Pixel Rejection (1)' section has to be set to 'No Normalization', see Vicent Peris's excellent tutorial:

https://www.pixinsight.com/tutorials/master-frames/

Why did you say that an outlier-dark is no problem? Something obviously went wrong during the capture of dark 022, and I definitely would exclude it from the integration.

Bernd

[joelshort]

Yes, for the integration of darks, flat-darks and bias frames the setting 'Normalization' in the 'Pixel Rejection (1)' section has to be set to 'No Normalization', see Vicent Peris's excellent tutorial:

https://www.pixinsight.com/tutorials/master-frames/

Why did say that an outlier-dark is no problem? Something obviously went wrong during the capture of dark 022, and I definitely would exclude it from the integration.

Bernd

Dark 22 IS a problem and I have removed it.  But that wasn't the cause of this problem of PI not integrating these darks.  If I use my normal Dark integration settings then the other 29 darks wouldn't integrate either. 

Somewhere along the way I probably got some advice regarding creating a master dark that said to use normalization, or I just copied my LIGHT integration settings which does use normalization.  I'll adjust my process icons accordingly so this doesn't happen in the future.

[pfile]

i think somewhere along the way someone may have concluded that canon DSLR darks do need to be normalized in some fashion due to the tricks the DIGIC processor plays with the raw data. i remember some threads about that here. but for CCD/CMOS bias/darks, they should essentially just be added together (no scaling, normalization or anything) and outliers rejected. in theory each frame should be very similar to the others so there's no need to massage them even for outlier rejection.

rob