Cosmetic Correction with Batch Preprocessing and dark scaling

[RalphPaonessa]

If I run Batch Preprocessing or Weighted Batch Preprocessing with different exposure times (120, 480, and 1800 sec for later HDR), I can only select one Cosmetic Correction instance, which I typically create from a master dark, adding a Defect List for bad columns in my KAF-8300 sensor.

I add master darks of 480 and 1800 sec and I choose Optimize dark frames, and I assume BPP and WBPP will scale the 480 sec dark for the 120 sec, and use 480 for 480, 1800 for 1800.

But I don't know if I can use CosCorr based on a single dark exp time for lights of different exposures.

Is there a way around this, or should I just batch process the different times separately (with the same registration reference, but different CosCorr instances)?

To make matters worse, the camera was rotated 180 deg after meridian flip, so I have both E and W lights and flats, so I need to batch process east and west separately.

BTW, I know that there's no reason to rotate, but there's a glitch in my automation software that rotates even if I tell it not to! (CCDAutoPilot, TheSkyX, and MoonLite NiteCrawler.) In the future I will just disconnect the rotator after framing my target.

Thanks,
Ralph

[ngc1535]

I know people like to use darks to characterize the hot pixel population... but  would simply using Auto detect with an appropriate (perhaps conservative) sigma threshold get you what you want? Sinch this is a statistical threshold (based on neighboring pixels) it is independent of dark frame vagaries.
You might consider comparing the number of pixels you are adjusting based on a fixed pattern defined by the dark compared with the computed number of pixels from the statistical detection.
I find that by using AUto Detect- I can be conservative in my threshold and let the later rejection through Image Integration take care of the rest. This assumes there are many images that are nicely dithered. Otherwise, being more aggressive with CC is sometimes warranted.

The bad columns and things could be a subsequent image container process. Instead of using a Defect List- create a Defect Map (an image) and then simply apply the DefectMap process to all the images. Using the Defect List ties you to CC. DefectMap separates the handling of hot pixels (CC)  from other fixed pattern things which it takes care of. You have the added bonus of specifying the method of pixel substitution.

-adam

[RalphPaonessa]

I know people like to use darks to characterize the hot pixel population... but  would simply using Auto detect with an appropriate (perhaps conservative) sigma threshold get you what you want?

Yes, I'd forgotten about the Auto Detect option. That would make more sense.

The bad columns and things could be a subsequent image container process. Instead of using a Defect List- create a Defect Map (an image) and then simply apply the DefectMap process to all the images.

I will try a Defect Map. If I use (Weighted) Batch Preprocessing, where in the workflow would I do this? It can't be done on the registered images, because they will not line up with the map. It seems it would be best to do it before BPP on the virgin light frames, unless that causes other problems?

Is the assumption that the Defect Map is treating defects that don't vary with exposure time; is that how bad columns and hot pixels behave? If that is the case, I can then add the different exposure times (for HDR) to WBPP simultaneously and it will sort them out. My bad column is a new development, so I'm only just learning how to deal with it.

If this works, I will run WBPP with two data sets, for the 180 deg rotation after meridian flip. (Note that the data is dithered.)

Thanks,
Ralph

[RalphPaonessa]

I will try a Defect Map. If I use (Weighted) Batch Preprocessing, where in the workflow would I do this? It can't be done on the registered images, because they will not line up with the map. It seems it would be best to do it before BPP on the virgin light frames, unless that causes other problems?

Running DefectMap before WBPP did not work. It removed the bad column from the raw subs; but then it reappeared as a dark line after calibration. I should have realized that subtracting the dark, which still had the column defect, would create that dark line where the column was. 

Another possibility: If I abandon DefectMap and use Cosmetic Correction with (a) Auto Detect, so that it is not tied to particular dark exposure, and (b) Defect List for my bad column, I think that will solve my problem, and still let me use WBPP.

Ralph

[ngc1535]

You use Defect Map after calibration (BPP/WBPP) and after CC (which is initiated in BPP/WBPP) and before registration.
Hot pixels will not be fixed in general- so AutoDetect is good.
Columns and such will typically persist- perhaps forever.
-adam