running into Flat subtraction issue

[HadesZ]

via the two photos you can see that in my Integrated image , my corners are somehow over subtracted it seems (lights in the corners) and in the other image, you can see what looks to me like evidence that somehow the bottom right of my master flat is also being subtracted from the top left of my stacked integrated image. how is this happening and what am I doing wrong?

ive also including in one of the images the steps im taking to calibrate the flats, and so on.

thank you for any help!

[HadesZ]

just to clarify, in the first image... the calibration is how I setup to calibrate the flats.. so taking our bias and dark signals. and the Integration in the same image is how I stacked the calibrated flats.

I do have a meridian flip involved in this data set (the lights obviously)  so I am wondering if this is somehow the issue.  I heard PI was great at auto correcting for this during stacking, which it has been, my final stacked image has pinpoint stars.. but something is obviously happening with the flat subtraction.

[pfile]

the meridian flip should not matter, unless of course you are using a rotator to undo the natural rotation the camera undergoes during the meridian flip.

the reason it does not matter is that calibration is performed before registration of the images. so even though the image of the DSO is upside-down relative to before the meridian flip, the spots and vignetting are exactly the same on both sides of the meridian. however if you are using a rotator then you do actually need 2 sets of flats as the camera's orientation to the telescope has changed.

what kind of camera is it and how were the lights and flats acquired? if you are using fits files anywhere in the flow there's the possibility that the images were read in upside-down relative to how they were written.

rob

[HadesZ]

the meridian flip should not matter, unless of course you are using a rotator to undo the natural rotation the camera undergoes during the meridian flip.

the reason it does not matter is that calibration is performed before registration of the images. so even though the image of the DSO is upside-down relative to before the meridian flip, the spots and vignetting are exactly the same on both sides of the meridian. however if you are using a rotator then you do actually need 2 sets of flats as the camera's orientation to the telescope has changed.

what kind of camera is it and how were the lights and flats acquired? if you are using fits files anywhere in the flow there's the possibility that the images were read in upside-down relative to how they were written.

rob

thank you for the reply Rob.  my Lights are .fit  ,  but everything I have saved post calibration, I have been using xisf.  should I be using all fits? or all xisf?

I am not using a rotator , and the flats were taking using a Spike-A flat field panel (LED)  (so no rotation, or other odd sky flat issues should be complicating the process)

maybe I need to give this all a shot with all the same file type ie - fits or xisf..... if there is a file type thats generally better to use please let me know

[pfile]

well the file format used from calibration on does not matter.

i should have been more specific - are the flats and light files as captured from the camera in the same file format?

rob

[HadesZ]

Yes they are Rob, everything captured from the camera is a .fit

not sure if I mentioned, im using a QHY10. incase this could present issues but.. dont see why it would.

I did try changing file types, still did nothing.  I even tried rotating the flats 180 degrees and then calibrated and stacked with lights, still giving me the issue. 

[pfile]

so i think the problem must lie in the calibration of the flats. how were the flats calibrated? bias? darks?

rob

[bulrichl]

so i think the problem must lie in the calibration of the flats. how were the flats calibrated? bias? darks?

rob

I am not sure about that. The second screenshot shows the left upper corner of the calibrated, registered, integrated light frames plus the right lower corner of the Master Flat. It looks like HadesZ supposed: the pattern of the Master Flat appears with inverted intensity and rotated by 180° in the integration. (By the way: during flat field correction the operation is division by Master Flat, not subtraction of Master Flat.)

It seems to me that the flat field correction of part of the lights (i. e. before meridian flip) has been performed correctly and for the other part (i. e. after m. f.) it is rotated by 180°.

What is your workflow for the calibration of the light frames? Were the light frames (before / after m. f.) calibrated altogether?

Bernd

[pfile]

well ok, sure, it is possible though sometimes these things (reading small spots in the images) can be very deceiving. i was going on the theory that if some of the inputs were FIT and some were CR2 that it's very easy for some of the frames to be flipped relative to the other since the fits orientation depends on the fits reader configuration in pixinsight. since they are all FITs the orientation should all be the same.

i also asked about a rotator because that's really the only way that an image can end up flipped when it's being exposed. without a rotator, flats just don't care about meridian flips because the flat models all the optical falloffs in the telescope and optical train; they have nothing to do with the orientation of the telescope to the sky. when you have a rotator and move it, you're changing the camera's orientation to the telescope, which is what matters for purposes of flats.

ergo i thought it best to move to different theories for why the flats are apparently overcorrecting the lights.

rob

[gvanhau]

Hello
Since the integration images include images of both meridian sides, it is perfectly posible that the spots are the effect of the rotation during image registration.

So I agree with Rob that the investigation should be centered on "why flats are over correcting".

One thing I would investigate is if the QHY10 is working linear; As far as I know you have gain and offset adjustments, if they are to far off, meay be you are working outside the linear part of the sensor.

Regards
Geert

[JGMoreau]

It is a known issue with the QHY10. I suggest using division with DBE to correct this.
Cheers,

Jean Guy Moreau
Quebec

[HadesZ]

so i think the problem must lie in the calibration of the flats. how were the flats calibrated? bias? darks?

rob

I am not sure about that. The second screenshot shows the left upper corner of the calibrated, registered, integrated light frames plus the right lower corner of the Master Flat. It looks like HadesZ supposed: the pattern of the Master Flat appears with inverted intensity and rotated by 180° in the integration. (By the way: during flat field correction the operation is division by Master Flat, not subtraction of Master Flat.)

It seems to me that the flat field correction of part of the lights (i. e. before meridian flip) has been performed correctly and for the other part (i. e. after m. f.) it is rotated by 180°.

What is your workflow for the calibration of the light frames? Were the light frames (before / after m. f.) calibrated altogether?

Bernd

Yes , all the light frames were calibrated together. Should I try and calibrate them separately? (Each Meridian flip)

[HadesZ]

It is a known issue with the QHY10. I suggest using division with DBE to correct this.
Cheers,

Jean Guy Moreau
Quebec

Havnt heard of this obviously,  what's the cause? Because I'm worried that this same thing is happening with my darks and bias!

There really needs to be a work around, otherwise... This camera is trash to me unfortunately :/

Can I use division with ABE? I've followed tutorials, tips, all the above on DBE, and I never, ever, get good results with DBE , ABE always out performs what I'm able to do with DBE.

A and thus far, ABE is not fully correcting the dust mote issue or the overall gradients PUT In by these flats frames. With the above result, I'd be better off not using them at all - which is also unexceptable to me.

Thanks again for all the info/help everyone!

[bulrichl]

Yes , all the light frames were calibrated together. Should I try and calibrate them separately? (Each Meridian flip)

No, I don't recommend to calibrate them separately.

I don't have an idea what's going wrong if your workflow is in the succession:

Calibration of the lights,
Debayering,
Registration,
Integration

(Only possibility seems to be that a flat field correction is performed after registration, but this is so out of place...)

I would take a closer look to the calibrated lights and compare thoroughly before / after flat field correction.

Bernd

[gvanhau]

Hello
May be you could try calibrating using the method I described in this link:
http://pixinsight.com/forum/index.php?topic=9312.msg59752#msg59752

At least it gives good results for me on my QHY8L.


Regards
Geert