Quick Advices to Work with the new SpectrophotometricColorCalibration tool

I'm getting an error when applying SPCC:

Robust linear regression (R/G): 0%
*** Error: Insufficient data: only 1 sample(s) are available; at least 5 are required.
Reading swap files...
2654.501 MiB/s
<* failed *>

I've attached a screenshot. I have no idea what setting(s) might address this - I would greatly appreciate some help, thanks!

More info: it's a 2x Drizzled image from WBPP; the non-drizzled image gets processed by SPCC just fine (see second screenshot). Is there a workaround?

Screen Shot 2023-04-26 at 10.44.38 AM.png
 

Attachments

  • No Drizzle.png
    No Drizzle.png
    596.6 KB · Views: 50
Last edited:
In2Driving said:
I'm getting an error when applying SPCC:

Robust linear regression (R/G): 0%
*** Error: Insufficient data: only 1 sample(s) are available; at least 5 are required.
Reading swap files...
2654.501 MiB/s
<* failed *>

I've attached a screenshot. I have no idea what setting(s) might address this - I would greatly appreciate some help, thanks!

More info: it's a 2x Drizzled image from WBPP; the non-drizzled image gets processed by SPCC just fine (see second screenshot). Is there a workaround?
Click to expand...
Here's the No Drizzle screenshot:
No Drizzle.jpg
 
fredvanner said:
Although the Optolong L-Pro is on the "Curve Explorer" (CE) list, it is not obvious (from the documentation) how to use it. Is this the correct process:
  • in CE, find and select the best match to your camera R filter
  • in CE, find and Ctrl-click the Optolong L-Pro ("PAN") filter
  • click on the View attachment 16545 icon to combine and create and name a new combined R filter
  • repeat for G and B filters
  • select the combined filters from the R, G, B dropdowns
Click to expand...
I have been trying this for the L-Pro but the new filters never show up in the drop down menus. Any idea what I am doing wrong?

Len
 
Are there any plans to include other light pollution filters to the CE list such as the Baader Moon & Skyglow filter?
 
fredvanner said:
I'm afraid the GAIA and APASS processes are more recent than "Inside Pixinsight".
It is worth undersanding how they work. They both replace on-line searches using the VizieR interface. Instead they implement efficient local searches of a downloaded version of the key databases with files in the proprietary .xpsd format.
To use these local databases, you must first download the necessary database files, which you can place anywhere you like on your system.
The PCC process requires the APASS data files DR9 and/or DR10 (about 3GB for both files). Because the coverage is not complete in either of these databases, I suggest that you download both if you intend to use PCC. PCC is superseded by the SPCC process. This uses a GAIA database with enhanced spectral data, the DR3/SP files. The full data set comprises 20 files totalling about 63GB; some trial and error may be required to find just how many you need, which will vary from image to image. The ImageSolver script should use the GAIA DR3 files, which are different to the DR3/SP files. These comprise 16 files totalling about 40GB.
Once you have downloaded the files, you must run the APASS process (to locate the APASS DR9 and DR10 files), and the GAIA process (to locate the DR3 and DR3/SP files). The process for specifying the file locations is the same for the APASS and GAIA processes:
  • first, select the "wrench" icon on the bottom bar of the process.
  • In the pop-up preferences window, select the data release dropdown and select which data release you are going to specify (e.g. "Gaia DR3/SP")
  • now click the "select" control and navigate to the folder with the required files
  • now select all the files required (using the multiple file select funtion of your operating system)
  • select "open" (in Windows, or the equivalent control)
  • the selected files will now be displayed in the database files window of the preferences dialogue
  • Repeat the above process for any other downloaded data release (e.g. "Gaia DR3")
  • you can then close the APASS / GAIA process - you do not need to execute it (globally or otherwise)
Executing the APASS or GAIA processes requests a database search (with the parameters specified in the dialogue). The resulting list of sources can be output to a file, if a file is specified. If no file is specified, a summary is output to the process console. This can be usefull as a quick "reality check" of how many sources there are in a region.
Click to expand...
Thank You for this Fred.

Helped me work out why the PCC wasn't working - I'm pretty much a newbie to PI and PCC stopped working when I made the purchase from the trial version - and I could not work out why. I'm guessing the trial version didn't specifically need the offline files.

One more issue I'm having - I don't seen to have access to SPCC even though I'm using the 1.8.9.-1 update. My current GAIA has no dropdown for DR3/SP files - it just has Gaia DR2 and Gaia EDR3. Am I missing something very obvious here?

Thank again
 
Ok, I want to give a try to SPCC. I have a EOS 500D with the the LPF-2 filter removed and an Astronomik L2 UV/IR block filter.
Since the 500D is an OSC, i choose Ideal QE Curve. But what should i choose for the filters ?
The Canon EOD 500D R, G and B filters proposed seems to be for an unmodded 500D.
Should I use the Canon Full Spectrum filters ? And combine them with a UV/IR block filter, which are not proposed in SPCC.

And another question : in the SPCC's doc it is recommended to always use drizzle with SPCC :
Another important point when working with OSC cameras is to always use drizzle. Besides the fact that only a drizzle integration can provide optimal results by avoiding interpolation, some de-Bayering algorithms may modify color proportions at small scales, where interpolation of missing color data in CFA patterns plays an important role. This is the case with VNG. While large–scale structures remain untouched, point sources (like stars) will be altered. And, what is worse, this color shift depends on the sizes of the stars so that each image can be altered differently. This affects stellar photometry so that de-Bayering can lead to unexpected erroneous results, both with PCC and SPCC.
Click to expand...
So I guess I have to drizzle my images first ?

And a last question : I have an image with a nebula which takes the whole frame. How can I perform Background Neutralization ? I have nowhere to draw a preview.
 
Last edited:
milesinthesky said:
Ok, I want to give a try to SPCC. I have a EOS 500D with the the LPF-2 filter removed and an Astronomik L2 UV/IR block filter.
Since the 500D is an OSC, i choose Ideal QE Curve. But what should i choose for the filters ?
The Canon EOD 500D R, G and B filters proposed seems to be for an unmodded 500D.
Should I use the Canon Full Spectrum filters ? And combine them with a UV/IR block filter, which are not proposed in SPCC.

And another question : in the SPCC's doc it is recommended to always use drizzle with SPCC :

So I guess I have to drizzle my images first ?
Click to expand...
In practice, it is unlikely that your choice of filters will make a noticeable difference in the color calibration. You can likely get away with any of the OSC filter choices and be fine.

Personally, I have almost never seen the slightest improvement from using drizzle given even a small degree of oversampling. I think the argument is more theoretical than practical. YMMV, but I'd suggest experimenting. And certainly, not bothering with it while you're trying to figure out your best SPCC settings.
 
milesinthesky said:
I guess I have to drizzle my images first
Click to expand...
You don't have to drizzle. If you have enough well-dithered frames you are advised to drizzle (x1).
The reason for using x1 dither with an OSC is that it avoids the interpolation required in the debayer process - but the difference is not dramatic, particularly if the image is oversampled (in which case the inherent PSF spread has much the same effect as the debayer interpolation).
 
cloudbait said:
In practice, it is unlikely that your choice of filters will make a noticeable difference in the color calibration. You can likely get away with any of the OSC filter choices and be fine.

Personally, I have almost never seen the slightest improvement from using drizzle given even a small degree of oversampling. I think the argument is more theoretical than practical. YMMV, but I'd suggest experimenting. And certainly, not bothering with it while you're trying to figure out your best SPCC settings.
Click to expand...
Ok I will try with EOS 500 et Full Spectrum filters. Thanks you !

fredvanner said:
You don't have to drizzle. If you have enough well-dithered frames you are advised to drizzle (x1).
The reason for using x1 dither with an OSC is that it avoids the interpolation required in the debayer process - but the difference is not dramatic, particularly if the image is oversampled (in which case the inherent PSF spread has much the same effect as the debayer interpolation).
Click to expand...
I didn't even know that x1 drizzle was a thing.
I have a lot of dithered frames and they're undersampled. So I guess I have a good reason to drizzle in my case. Is there a benefit to x2 drizzle or should I only use x1 drizzle ?
 
milesinthesky said:
Is there a benefit to x2 drizzle
Click to expand...
If your image is significantly undersampled (e.g. if running FWHMEccentricity gives an FWHM < 2 pixels) then x2 drizzle can increase the effective resolution of your image. The improvement will depend on just how undersampled you are. Note that this refers to the actual sampling (not the theoretical resolution). In the UK even images with sub-pixel thoretical resolution are almost always limited by the seeing conditions - hence my advice to use FWHMEccentricity to check your actual sampling PSF. You will also need more well-dithered frames for x2 drizzle (I don't think there is any documented way of estimating the number of frames required, but I'd aim for at least 50 for x2 drizzle).
 
fredvanner said:
If your image is significantly undersampled (e.g. if running FWHMEccentricity gives an FWHM < 2 pixels) then x2 drizzle can increase the effective resolution of your image. The improvement will depend on just how undersampled you are. Note that this refers to the actual sampling (not the theoretical resolution). In the UK even images with sub-pixel thoretical resolution are almost always limited by the seeing conditions - hence my advice to use FWHMEccentricity to check your actual sampling PSF. You will also need more well-dithered frames for x2 drizzle (I don't think there is any documented way of estimating the number of frames required, but I'd aim for at least 50 for x2 drizzle).
Click to expand...
I just checked, my FWHM is 2.3px on Moffat4 stars. I will give a go to x1 drizzle then.
And to return to SPCC, what would you do if you had an image with only nebula for the Background Neutralization ? On my NGC7000 image I only have nebula or dark nebula, but no free background to draw a preview for the BN.
 
What you are trying to correct with BN is any small channel offset left after the SPCC linear scaling. Try with no BN first, but if there is a clear background colour problem, I would pick a small dark patch of dark nebula - anything that you think is not coloured.
 
You must log in or register to reply here.
Back
Top