ASI1600MM Pro: impossible to get good flats? - Forums

Topic starter

Sep 16, 2022 · Edited: Sep 16, 2022

Hello everybody! I'm creating this topic here as the thread on Cloudy Nights is not getting much attetion. I'm sure some of the more experienced astrophotographers might know the fix to this issue, but until now, no one could find something significant that figured my flats out to calibrate properly.

The issue is: over / undercorrection on all 7 filters (camera is an ASI1600MM Pro). I already tried several solutions but had no sucess what so ever. I made a very big list of what I already tried and some details of my setup on the CN post here. I also uploaded all my raw data so everyone can inspect it here. It has been 2 years now that I live with this issue and all pictures taken by myself on my gallery suffer from this terrible over / undercorrection, meaning I need to use DBE on very high tolerances (that takes away a lot of signal) to eliminate the mess flats are creating. But I need to overcome this as I want to improve my pictures, I just can't subract signal from dozen of hours invested on a project that should go deep.

Details I posted on CN:

I tried making all types of flats: white t-shirt or blank and iPad, white t-shirt and sky flats, just sky flats, iPad flats, iPad and paper sheet, white wall, and maybe more stuff that I don't remember now.
I tried short exposures (<1s), medium exposures (>1s and <5s), and a bit longer exposures (>5s), that didn't have any effect.
I aim for 50% histogram (32768) in 16bit mode.
I use gain 139 and offset 50 (that's unity gain) for all my captures and never tried other settings. Sensor temperature at -10°C for all frame types.
L filter and Ha seem to be fully corrected or al least very close to the desired result. RGB and SII-OIII are very problematic, with special metion to G filter. I have the ZWO New 1.25" LRGB 7nm SHO filter set for the 1600.
It doesn't look like I have a parasite light source case. I took them during the day and night and both had the same problematic result. Obviously I usually take them when it's dark as an extra caution, but it doesn't help it.
It does happen on other telescopes.
I tried 0.3s bias before, that never worked. Today I shoot flat darks at the exact flat exposure to match each filter, doesn't work either (and I'm almost sure the light - dark - flat - flatdark gain and offsets settings are the same).
I already watched the Adam Block series that cover the exact problem I have, and the solution he shows at part 3 doesn't apply to me as well. The partial solution in part 2 does work though but doesn't satisfy me, because as himself states: 'it doesn't solve the root cause'.
I don't enable dark optimization.
I'm sure flats are calibrated correctly by WBPP (I'm using PixInsight), and that it's matching the calibration frames accordingly.
I tried calibrating everything by hand, I get the same result as WBPP.
The mean background value of my master flat dark does match the mean value of my master dark.
I already bought a flip flat panel (for my observatory), and I'm almost sure that won't solve any problem either, but just automate the making of my nightmare. I'll update the topic if that changes anything at all.
My flats worked once in my lifetime when I was at a astrophotography party, but unfortunately I don't have the original files anymore, aside from the stacks. I did them exactly how I do them today at home: paper sheets and iPad as a white light source using the values I quoted before. But again, I have no clue of what factor might have changed that made it work, maybe it was some divine intervention after all LOL.

If someone is able to help, it is very welcome any suggestions.

masluigi

Sep 16, 2022 · Edited: Sep 16, 2022

I use 1 second flat and dark flat no bias
4 layers of t-shirts for lrgb - nothing for narrowband
gain 139 offset 50

masluigi

Sep 16, 2022

luminous graphic tablet for lighting

andrea tasselli

Frequent contributor

Sep 16, 2022

Whats the scope?

Mauricio Christiano de Souza

Topic starter

Sep 16, 2022

masluigi:
I use 1 second flat and dark flat no bias
gain 139 offset 50

I do the same thing, I generally use 1 second, gain 139, offset 50 and temperature at -10°C. However, even using longer or shorter exposures doesn't make any difference.

Mauricio Christiano de Souza

Topic starter

Sep 16, 2022

andrea tasselli:
Whats the scope?

Today I'm using a Takahashi FSQ-85EDX, but I noticed this issues on all telescopes that I used with this camera. That includes a GSO 8" f/5 Newtonian and the Meade Series 6000 80mm Triplet Apo. The Takahashi has not very significant vignetting, but it still requires flats.

andrea tasselli

Frequent contributor

Sep 16, 2022

Can you show us your flats, G for example?

andrea tasselli

Frequent contributor

Sep 16, 2022 · Edited: Sep 16, 2022

And, BTW, how do you take sky flats? When, especially.

Mauricio Christiano de Souza

Topic starter

Sep 16, 2022

andrea tasselli:
Can you show us your flats, G for example?

Mauricio Christiano de Souza

Topic starter

Sep 16, 2022

andrea tasselli:
And, BTW, how do you take sky flats? When, especially.

I don't take sky flats anymore, I just take iPad with paper sheets flats.

Andy Wray

Sep 16, 2022 · Edited: Sep 16, 2022

I have EXACTLY the same camera and filters as you and do not have issues with flats. My flat setup is:

* 2 white T-shirts
* Cheap A3 LED tracing panel from Amazon
* I use NINA flats wizard (and peviously APT's flats wizard) and target 22,000 ADU
* I do not use Bias frames at all, but I do take dark flats
* I use Pixinsight WBPP for all my processing

The only time I have had problems with flats is when my corresponding darks/dark flats were taken at a different offset to the flats/lights. That would immediately cause the over/under-correction.

FWIW: I take my flats/dark flats at gain 0 and my lights/darks at gain 139, both at offset 50 (all at -10 degrees C). I'm sure you know this, but you can check the offset that was used in the FITS header.

I hope this helps.

Helpful Respectful Supportive

Mauricio Christiano de Souza

Topic starter

Sep 16, 2022

Andy Wray:
I have EXACTLY the same camera and filters as you and do not have issues with flats. My flat setup is:

* 2 white T-shirts
* Cheap A3 LED tracing panel from Amazon
* I use NINA flats wizard (and peviously APT's flats wizard) and target 22,000 ADU
* I do not use Bias frames at all, but I do take dark flats
* I use Pixinsight WBPP for all my processing

The only time I have had problems with flats is when my corresponding darks/dark flats were taken at a different offset to the flats/lights. That would immediately cause the over/under-correction.

FWIW: I take my flats/dark flats at gain 0 and my lights/darks at gain 139, both at offset 50 (all at -10 degrees C). I'm sure you know this, but you can check the offset that was used in the FITS header.

I hope this helps.

That is what annoys me a lot! I have seen maybe hundreds of people with the same setup and they don't have any problem at all while I try being extra cautious!
I didn't know you could use different gain settings, I thought the gain and offset had to be the same.
Why the 22,000 ADU target and not the exact middle (close to 32,000)?

Andy Wray

Sep 16, 2022 · Edited: Sep 16, 2022

Mauricio Christiano de Souza:
That is what annoys me a lot! I have seen maybe hundreds of people with the same setup and they don't have any problem at all while I try being extra cautious!
I didn't know you could use different gain settings, I thought the gain and offset had to be the same.
Why the 22,000 ADU target and not the exact middle (close to 32,000)?

Your dark flats don't look right. I just looked at one using Image Inspection in Pixinsight it showed a minimum value of 32 ADU. With an offset of 50 that should be nearer 800 ((50/4096)*65536). i.e. it looked like your flat darks had been taken at an offset of nearer 5 rather than 50.

Hope this helps.

Ref the ADU: I just picked that up from somewhere when I was in the same situation as you and have stuck with it. I don't think it matters too much.

Ref the Gain settings: you definitely MUST match the gain, offset, exposure and temperature settings between darks and lights. The same applies to flat darks and flats. I do not believe the gain needs to be the same between lights and flats though.

Lastly, you could use pixelmath to alter your master flat (the one I looked at was green) by adding a value to it and then re-integrating the lights against that flat. The formula would be something like:

$T+((800-32)/65536)

My Flat darks actually have a minimum value of 730 ADU, so you may also want to try:

$T+((730-32)/65536)

Helpful Supportive

andrea tasselli

Frequent contributor

Sep 16, 2022

From the flats I don't see any obvious indication of light leak. Usually this sort of issue is due to a gain/offset factor when used for calibration. If you think the flats can remove all sort of gradients from your lights you are mistaken. You'll always have some residual gradient, the more light polluted your skies are the worse it would be (even more so for the G channel).

Helpful Insightful

Mauricio Christiano de Souza

Topic starter

Sep 19, 2022

andrea tasselli:
From the flats I don't see any obvious indication of light leak. Usually this sort of issue is due to a gain/offset factor when used for calibration. If you think the flats can remove all sort of gradients from your lights you are mistaken. You'll always have some residual gradient, the more light polluted your skies are the worse it would be (even more so for the G channel).

Yes, I know flats job isn't removing light pollution as well, but the problem is this over and undercorrection of vignetting. If the images were calibrated correctely, the final result would be an image with a smooth background gradient.

Mauricio Christiano de Souza

Topic starter

Sep 19, 2022

Andy Wray:
Mauricio Christiano de Souza:
That is what annoys me a lot! I have seen maybe hundreds of people with the same setup and they don't have any problem at all while I try being extra cautious!
I didn't know you could use different gain settings, I thought the gain and offset had to be the same.
Why the 22,000 ADU target and not the exact middle (close to 32,000)?

Your dark flats don't look right. I just looked at one using Image Inspection in Pixinsight it showed a minimum value of 32 ADU. With an offset of 50 that should be nearer 800 ((50/4096)*65536). i.e. it looked like your flat darks had been taken at an offset of nearer 5 rather than 50.

Hope this helps.

Ref the ADU: I just picked that up from somewhere when I was in the same situation as you and have stuck with it. I don't think it matters too much.

Ref the Gain settings: you definitely MUST match the gain, offset, exposure and temperature settings between darks and lights. The same applies to flat darks and flats. I do not believe the gain needs to be the same between lights and flats though.

Lastly, you could use pixelmath to alter your master flat (the one I looked at was green) by adding a value to it and then re-integrating the lights against that flat. The formula would be something like:

$T+((800-32)/65536)

My Flat darks actually have a minimum value of 730 ADU, so you may also want to try:

$T+((730-32)/65536)

I'll take new flats, flat darks and darks tonight. I hope this time I'm sure the gain and offset values are correct.

Well Written

Douwe79

Sep 19, 2022

I’m a 1600 mm user also:

never had problem with the following method:

- I use a dimmable computer screen with white background
- I attach white t-shirt in front of scope
- I use flat aid in APT to calculate the optimal length of the flats

good luck

Mauricio Christiano de Souza

Topic starter

Sep 20, 2022

Hello again everyone. Here's my update: the problem was really the darks and flat darks (they had light leaks). I decided to take new ones inside the fridge, it seems like it helps a lot with cooling and is a very dark enviroment inside, and they worked beautifully. I'm down now to correct light pollution gradients only.
Thanks everyone that pointed out something was strange about my calibration data.

Helpful Respectful