Overscan and Calibration

[mmirot]

How do I use overscan with calibration?
I have ability to acquire image with the overscan area with my camera.

Max

[mmirot]

you might want to read this thread. It is why I am asking.

http://pixinsight.com/forum/index.php?topic=2010.0

Max

[vicent_peris]

Hi,

I was just going to reply the other thread when I saw your message. :-P

Overscan correction works in PI by subtracting the pedestal defined by the median value of the overscan of each frame. This means that each calibration frame is overscan corrected, as well as each light frame.

You must set first the image area: the CCD area that contains the sensitive photosites. Then you define the overscan area; if your camera creates different overscan areas, define only one of them.

After defining the overscan area, you must define wich area is going to be corrected by that overscan area. In your case, this area will be the same image area of the beggining. As some other cameras may have more than one readout outputs, they have more than one overscan areas; see my article about NGC7331 to see what I mean.


Best regards,
Vicent.

[mmirot]

Vicent,

So this fixes the problem of bias drift that Nikolay reported. Correct?


Now I need to know how to fill in the overscan boxes better.

What is image region, source region and target?  I need  to have these explained these better than the mouseover.

I realize you are defining the pixel coordinates of the image area/region and overscan

I need an example if possible.

Thanks

Max

[vicent_peris]

So this fixes the problem of bias drift that Nikolay reported. Correct?/quote]

Yes.

Now I need to know how to fill in the overscan boxes better.

What is image region, source region and target?

Image region: The area where you have significative data from the sky. All the images will be cropped after the calibration process. This is important because you must trash any area with no significative data; this can be the overscan area (no real photosites), or highly vignetted areas. I. e. (1), at CAHA, our 65 mm filters leave a complete vignetted border of about 50 pixels wide. Is better to crop the image from 2048 pixels wide to 1950; this border becomes very noisy after calibration and degrades the statistics of the image. I. e. (2), if you have in your CCD an overscan area of 100x4096 pixels and it is at right of the frame, your total image size will be 4096x4196, but your values for image region (in case you don't have any vignetting) are 0, 0,

Source region: The area of the overscan. I. e., if you have in your CCD an overscan area of 100x4096 pixels and it is at right of the frame, you will put these values for source region: 4096, 0, 100, 4096.

Target region: The area on wich the overscan level is corrected. In your case, the target region is the same as the image region.

If your camera has more than one read outputs, you may activate more source and target regions. But this is not the usual case for amateur cameras.


Hope this helps,
V.

[mmirot]

Thanks Vicent crystal clear now.

I will start using this feature.

How do I register for Adler? Link

Max

[vicent_peris]

Actually overscan correction is done by suctracting the overscan pedestal, following what says all the bibliography. But I've found evidence of being a better correction by division! I will further investigate this... at Calar Alto we have old CCD cameras that are far more complicated to calibrate than an actual amateur camera... we're lucky to have such problems: if you don't have the problem, you cannot correct it! :-D

Never trust completely on the bibliography.


V.

[Niall Saunders]

at Calar Alto we have old CCD cameras that are far more complicated to calibrate than an actual amateur camera

That's why I 'like' my old DSI cameras - there is nothing sophisticated about them. You have to work very hard to get mediocre results - but to even achieve that, you really have to put a lot of effort into the actual 'learning' process (far more effort than you ever need to put into the 'doing' process )

In fact, by comparison, I fully expect to produce an APOD every time I open the observatory roof - once I upgrade at the end of the year

Cheers,

[NKV]

Vicent, one more time about overscan area for ImageCalibration please.

We should use for
Source region: light protected area ( it's Dark Reference Pixels ). I think, no.
or
Source region: no real photosites area ( it's Bias Reference Pixels ). I think, yes.

Looking to KAF-16803LongSpec.pdf I see description for Dummy Pixels / Internal Test Pixel / Dark Reference Pixels / Active Buffer Pixels.

Which one is Bias Reference Pixels?

Best regards,
Nikolay.

[Ginge]

Hi!

I'm definately missing something here after having read Vicent's post time and time again. I have an image from FORS1 with four separate readouts, but I feel I must have misunderstood how the overscan area is defined. Like you can see from this screenshot, my settings doesnt exactly do the trick. I have cropped away the black areas on the sides in the definition of the Image Area and tried to define one of the squares as overscan area (have also tested without the black area on the left, i.e. source start point x=17). Where have I gone wrong, and how do I use the information about overscan area from the FITS header?



Best regards,
Ginge

[vicent_peris]

Hi Ginge,

please attatch the screenshot to see what happens.

Regards,
Vicent.

[Ginge]

Hi Vincent! Refresh, I've fixed the bad link.

And when you see what I did you'll recognise just how WAY OFF I was here. I basically didn't know the meaning of the term Overscan at all. Am I correct that it is the black area on the right side of the image above? I defined it once more in ImageCalibration from the info on this page http://www.eso.org/observing/dfo/quality/FORS1/pipeline/pipe_gen.html. And am I right in assuming that the difference in illumination on the Flat above will be corrected if I enter the right integers in the Overscan portion of ImageCalibration?

I have entered this which goes through with no error code but the image remains uncorrected:


I am uncertain of the validity of my coordinates as my head has entered gimbal lock completely. Can somebody please compare my numbers to this info from FORS1 and tell me what do do?:

 

Best regards,
Ginge

[vicent_peris]

Am I correct that it is the black area on the right side of the image above?

Ginge, I cannot see the image above. But yes, overscan areas are always nearly black, because they represent the actual bias level.

I defined it once more in ImageCalibration from the info on this page http://www.eso.org/observing/dfo/quality/FORS1/pipeline/pipe_gen.html. And am I right in assuming that the difference in illumination on the Flat above will be corrected if I enter the right integers in the Overscan portion of ImageCalibration?

Yes, that should work.

I have entered this which goes through with no error code but the image remains uncorrected:

My web browser reports here a 404 error: document not found.

I am uncertain of the validity of my coordinates as my head has entered gimbal lock completely. Can somebody please compare my numbers to this info from FORS1 and tell me what do do?:

Could you please upload to the forum a process icon of your ImageCalibration instance?


Regards,
Vicent.

[Ginge]

Sorry for making a mess. Have changed links now and hope they work this time.

I addition you can download a FITS file of the flat frame I want to process here:

https://files.me.com/ginge70/hi5lpk

and the process icon here:

https://files.me.com/ginge70/idjx7l
Hope this works better .


Best regards,
Ginge

[vicent_peris]

Hi Ginge,

I attach the IC icon with the correct overscan settings.

OTOH, overscan doesn't correct the illumination difference between each quadrant in the flatfield, but this is normal. Each amplifier in the CCD has a slightly different electron to ADU conversion, and the flatfield contains itself this differences. So you must check if the quadrants disappear before calibrating the light images by overscan, bias and flatfield.


HTH,
V.