eXcalibrator and PI

[Emanuele]

Hi all,

there's an incredibly nice tool called eXcalibrator which will use the DSS plates to give the weights of the different RGB channels for RGB combination.
This method is way better than the G2V calibration method using a G2V star.

The problem is that eXcalibrator needs the FITS files, with the header intact as it came out of the CCD: so I cannot use PI because it says it can't recongize the format (needs 8 bit signed, 16 bit unsigned, 32 bit float).

Anybody know what the problem might be? Maybe it is the FIT header?

Thank you
E.

[Juan Conejero]

Hi Emanuele,

PixInsight respects the original FITS headers. Certainly, it also adds some private FITS header keywords, mainly to identify itself and to support some features (like ICC profiles for example), but if you open a FITS image and save it with PixInsight, no keywords are lost. Anyway, you can always save and restore the original FITS header of an image:

1. Open the FITS file.
2. Select File > FITS Header
3. Save a process icon from the FITSHeader tool.
4. Process your image.
5. Apply the FITSHeader process icon you saved in step 3
6. Save your image as a FITS file.

On the other hand, PixInsight fully supports 8, 16, 32 and 64-bit FITS images (64-bit integer images are not writeable), and it generates fully standards-compliant FITS files. You can check it with NASA's FITS verifier application:

http://fits.gsfc.nasa.gov/fits_verify.html

So I suspect the problem here is not in the FITS headers, but somewhere else.

As a footnote, I personally don't like and don't recommend the G2V color calibration method (or any other method based on particular spectral types), as I think it is neither necessary nor reliable, but that's another (long) story.

[Nigel Ball]

Emanuele

I've been using eXcalibrator for some time now. I even maanged to calibrqate Canon RAW files using it. I had to split the CR2 file into R,G and B FITS then add a few FITS keywords but it worked okay.

Certainly the Red files has to contain certain of the FITS KEYWORDS - you'll get an error message saying 'PIXEL X - not found' or the like

Maybe you dont get this far?

What is the error you are getting?

Nigel

[Emanuele]

Juan,

thank you for explaining the fits handling. I might have an incorrect setting in my computers then. I have attached the fits header form of my M31 image. All the informations there are gone and just a few lines exists.

Nigel: when I add the Red FIT to eXcalibrator, it does not even open it, saying that it needs an 8 bit, 16 bit unsigned or 32 bit Integer.... something like that.
Am not sure how to fix that...

[Nocturnal]

Well that should be easy E Save the FITS file in a format that eX understands. I'm sure you've seen the option dialog when saving FITS files in PI? Experiment.

[Emanuele]

Will do, am trying as I type this....

[Nigel Ball]

Nigel: when I add the Red FIT to eXcalibrator, it does not even open it, saying that it needs an 8 bit, 16 bit unsigned or 32 bit Integer.... something like that.
Am not sure how to fix that...

Emanuele

eXcalibrator treats the RED FITS file differently to the Green and Blue. The Red requires a Plate Solve otherwise it will not load? Are you able to load the Green and Blue okay?

HTH

Nigel

[Emanuele]

Nigel, can i just use a single fit, or do I have to use the stacked masters for each channel?

[Nigel Ball]

Emanuele

You are calibrating the colour ratios between R,G and B so you must use the stacked combines that you are going to use in your final image

[sleshin]

Hi Emanuele,

I have tried eXcalibrator and like you ran into some problems. These are not PI related problems but rather relate to getting the files saved in proper format. I must say that I'm not convinced eXcal offers any advantage to the excellent Color Calibration module in PI. Nonetheless, I did figure out how to get eXcal working and the following is the work flow.

1. Need to use R, G, and B masters in eXcal. These need to be 16 bit unsigned fits format which is an available format in PI. If you are creating these masters in PI then you want to save each of them twice. Once as a 32 bit floating point fits for further processing and a second time as 16 bit unsigned fits to use in eXcal.

2. eXcal requires a plate solve of the image so it can figure out which stars in the field to use to calculate the RGB ratio. eXcal uses the red master for this so you have to do a plate solve on the red master. I use CCD Soft for this and initially had a problem, CCD Soft would not plate solve the master red image because the RA and DEC coordinates are not in the fits header of the master image. To solve this problem I loaded one of the original 16 bit fits images into CCD Soft so that CCD soft now had the  proper RA and DEC coordinates. Closed this red frame and then opened the master red which then will be successfully plate solved. Once solved save that image as a 16 bit unsigned fits.

3. Proceed to eXcal and use the master red saved after plate solve and the G and B previously saved.

Hope I got that right and hope it helps.

Steve

[Nocturnal]

Good grief that's a lot of work

Cheap b@st@rds like me can plate solve for free with astrometry.net  and Elbrus

[sleshin]

Should have been clearer. Don't believe CCD Soft is required for plate solving by eXcal so likely can use any program you want. I use CCD Soft, which was "free" with my SBIG camera, because it's the program I use for image acquisition along with CCD Autopilot so its set up do do plate solves for my system. If you mean that using eXcal is a lot of work then I agree and in fact I don't use it because I really like the Color Cal process in PI. Was just trying to offer some help to Emanuele.

Steve

[Nocturnal]

Of course Steve, I'm sure Emanuele appreciates it

[mmirot]

It is pretty easy if you are using one the automation programs to acquire images( ie CCDAP , ACP, CCDCommander)
They insert a plate solve WCS as part of the process.

The only problem with the PI method is that it does not account for atmospheric extinction.
However , I think it is fine 98% of the time. The other 2% will require an extra tweek

Vicent wants do something similar to Excalibrator using standard galaxies.
He and Juan don't think the G2V should be the standard candle for various reasons.

I am not sure it is really worth the time to develop such a tool. 
There is no reason to create a standard for deep sky color , it is all arbitrary conventions

Max

[Nocturnal]

Hi Max,

it's actually the other way around. G2V calibration does not take atmospheric extinction into account when a single G2V measurement is applied to an entire stack. When you calibrate an entire stack based on some assumption (such as all stars are on average white) then you are taking atmospheric extinction and filters and spectral response into consideration. Not separately but as an entire set of factors that affect color. You then correct till the colors are 'correct'.

This is why I always advise people to skip the RGB weighting effort that some folks seem determined to make. Just debayer and stack and let the image be green. Then calibrate and correct. No matter how smart the RGB weights are, you'll always have to correct anyway so you might as well skip the weights. That's why the debayer module doesn't have weights.