WBPP Cosmetic Correction workflow help please

[maxchess42]

I am relatively new to Pixinsight and I am now trying to expand my cooled OSC workflow.
I want to use Cosmetic Correction with WBPP 2. I understand that CC must be used with calibrated images, but starting with a new set of lights I will not have these until I run WBPP. So I assume I need to do do two runs as follows:
-Standard run of WBPP to produced calibrated images
-Run CC on calibrated images and save process icon with desired setting.
-Run WBPP again using using Process Icon which will perform CC in the right place.

I assume that from then on I can use the same Process Icon with future images with the same sensor and duration.

Is this correct or am I missing something?
Should I redo the CC Process at intervals, ie will my sensor change?
Is CC worth doing with an OSC if I have 50 darks in my master (ie why doesn't the dark master remove Hot/Cold pixels?)

Thanks, Max

 

[fredvanner]

You are indeed missing something (but don't worry, you are not the first to miss this)!
If you look on the "Lights" tab of WBPP2, you will find a "Cosmetic Correction" panel:

To use this you must first run the CosmeticCorrection process, and set it up with the settings you want. I recommend the following:

Note: CFA checked; "Use Auto detect" checked; "Hot Sigma" checked and set to 3.0 (a good starting value - you can play with this). Everything else left blank.
Then drag the instance icon (blue triangle) to the desktop to create a process icon, which it would be sensible to rename (I called it ccCFA).
You need to do this before running WBPP, so the icon is ready on the desktop to be entered in the WBPP CC panel.
Then run WBPP, and the CC is done automatically at the appropriate point in the workflow.

 

[martingisborne]

You're not alone, Max, it's definitely not terribly intuitive! (But the results are definitely worth it).

 

[maxchess42]

Thanks for that, I think my confusion comes from two different ways of using Cosmetic Correction. Again please correct me if this is wrong.
I need a CC process defined for each sensor/duration combination. I can re-use this for other image sets with the same settings, much like a Master Dark.

1) The Auto way which you describe, where a CC process is created using "Auto Detect" with the parameters you define, but no Target Calibration Frames loaded. This process is then named and saved and used in the WBPP

2) A more refined way in which WBPP is used first just to create Calibrated images and a Master Dark and then CC is run using these calibrated images and loading the Master Dark. When the CC Master Dark box is ticked further options allow for adjustment of the Sigma value while viewing a Real Time Preview. so that a more accurate threshold can be set that eliminates Hot(or Cold) Pixels.
A defect list can also be created. Clearly I must then go back and re-run WBPP with this newly defined CC process so it can do everything in the right order. Although I end up running WBPP twice, from now on I can re-use my CC process on similar image sets.

One of the reference books I use (Keller) says you can use these together. Another book (Andreo) says that "Auto Detect" is used "instead of relying on a Dark Master frame."

Clear as mud, right?

 

[fredvanner]

Is CC worth doing with an OSC if I have 50 darks in my master (ie why doesn't the dark master remove Hot/Cold pixels?)

I confess I don't know the answer to this good question.
In the past, using either my Canon DSLR or my ZWO ASI183MC Pro, I found that a good master dark eliminated hot pixels (including amp. glow), and appropriate pixel rejection during integration eliminated occasional cosmic rays, aircraft, satellite trails, etc.
When I got my super new ZWO ASI2600MC low noise CMOS camera, I discovered that it has so little dark current that even quite long exposure darks are virtually indistinguishable from bias frames. But I was plagued with a completely new phenomenon of "warm pixels" - pixels that do not appear in the dark frames, but are present as "excess exposure" pixels in the light frames. It is these "warm pixels" that CC is really good at eliminating. WBPP is now sufficiently configurable that I almost always use it for my calibration workflow (only re-working if some problem appears). For safety's sake I take darks, dark flats, flats and lights, but with the 2600 the darks and dark flats are virtually identical. I believe these are mainly achieving bias subtraction, with little or no dark current correction. It is CC that is fixing most of the pixel artefacts in my images.
So for this camera at least, I need darks mainly for bias correction and CC for "warm pixel" correction.
Since lots of other users are finding that CC fixes pixels that "escape" dark calibration, I guess this is not just a feature of my camera!
I don't know what causes it, and I haven't yet seen any "authoritative" expert post on any astro forum that claims to explain it (there is sometimes a suggestion that this is a difference between CCD and CMOS cameras; I've never used a CCD camera, so for me it is a difference between "old" CMOS and "new" CMOS cameras). While this may not be the explanation, it appears that some pixels have anomalously high gain, so thay don't show up in dark exposures, but are anomalously bright in light exposures.
I'd be very interested to know if anyone has any more information on the cause of the ("warm pixel") behaviour.

 

[martingisborne]

In my more recent experiments what I've started to do is;

Throw everything at WBPP (all the lights, all the darks, flats, dark flats and bias frames), I let WBPP do its thing with little intervention from me and without doing CC. Then I have a quick look at the master integration for fun and some delayed gratification.

Then I start digging in, I use the calibrated files in Blink and toss some out, then I run the calibrated files through Subframe Selector and toss more out. Then I pick one of the remaining calibrated files and pull it into CC and use the preview to dial up or down, as appropriate, once I see it removing an appropriate amount of hot pixels I then make a process icon of it... and then use that in a fresh new WBPP using only the lights that haven't been tossed out. And I've started ruthlessly tossing data...

Caveats:
I am most definitely not an "authoratative" expert on PixInsight
I'm only on my second coffee of the day