Help needed with ASI 1600 Flats over correcting - DSS problem.

[Stub Mandrel]

I'm having trouble with flats over correcting in DSS. This hasn't been a problem for me with DSLR images unless I made silly mistakes.

I'm using an ASI1600MM, all frames (lights, darks, flast, dark flats taken at -20C), flats are taken using a flat panel with an extra layer of diffuser, rotated between frames (24 of them). I've greatly reduced the sample images in size and these are 8 bit too.

Here's the unstretched master flat, clearly not over or under exposed (I checked the histogram):

Here's a stretched master flat which looks sensible enough:

Here's the result of a stack, with a quick stretch in DSS, note the over darkened centre:

I can't help noticing that the dark area looks suspiciously 'square'...

For completeness, stack without flats:

I've tried stacking with and without darks and with and without dark flats, and with and without checking 'set the black point to zero'.

The only things I can think of are that:

1 - I used TIFF this time for Lights, Flats and Bias, I didn't have this issue when I used FITS (sharpcap generates a master dark that's a FITS).

2 - Dark subtraction may have been switched on when taking the lights.. but if this was the problem not using a dark should surely have solved the issue?

Has anyone got any suggestions?

[david_taurus83]

Hi Neil. The stretched master flat looks too bright in the middle. Try taking them again and output everything in FITS. I have made the mistake in Sharpcap before. Spent a whole day doing a dark library to find out after the output files were png or something daft.

[Stub Mandrel]

3 hours ago, david_taurus83 said:

Hi Neil. The stretched master flat looks too bright in the middle. Try taking them again and output everything in FITS. I have made the mistake in Sharpcap before. Spent a whole day doing a dark library to find out after the output files were png or something daft.

Hi David, still set up. I've taken another set as FITS. Also, I've done them without the dewshield - previously I've always done them without it. Fingers crossed.

[Stub Mandrel]

I'm not 100% sure, but I suspect the problem is that my image has virtually no detail (conditions were far from good) and that I'm just stretching it far beyond what the data will allow.

[wimvb]

If you suspect the calibration process of the master flat, try calibrating one flat frame with the master flat, and the master flat dark. This should give a completely flat flat.

15 hours ago, Stub Mandrel said:

I've greatly reduced the sample images in size and these are 8 bit too.

I guess that this means that the originals are NOT 8 bit, and that lights and calibration frames are all the same bit depth.

[Stub Mandrel]

2 hours ago, wimvb said:

If you suspect the calibration process of the master flat, try calibrating one flat frame with the master flat, and the master flat dark. This should give a completely flat flat.

I guess that this means that the originals are NOT 8 bit, and that lights and calibration frames are all the same bit depth.

Yep, everything is 16 bit. (12 bit really).

[Stub Mandrel]

3 hours ago, wimvb said:

If you suspect the calibration process of the master flat, try calibrating one flat frame with the master flat, and the master flat dark. This should give a completely flat flat.

Looks like the process is OK, allowing for the fact my LED flat screen probably has a gradient (I rotate it to eliminate this) the result is pretty even, certainly no big dark (or light) spot in the upper middle:

Linearly stretched to show the very smooth residual gradient, this is just because the LED panel is brighter on one side - bear in mind that it's always worked fine for my other camera (and last week with this one!)

Edited December 19, 2019 by Stub Mandrel

[wimvb]

Next you could check your dark master, but this is also a likely candidate:

9 hours ago, Stub Mandrel said:

I'm not 100% sure, but I suspect the problem is that my image has virtually no detail (conditions were far from good) and that I'm just stretching it far beyond what the data will allow

If the problem persists, it's vey likely something in the calibration process, but otherwise atmospheric conditions could be to blame. Are all subs affected equally, or are some subs worse than others?

[Stub Mandrel]

On 19/12/2019 at 23:11, wimvb said:

Next you could check your dark master, but this is also a likely candidate:

If the problem persists, it's vey likely something in the calibration process, but otherwise atmospheric conditions could be to blame. Are all subs affected equally, or are some subs worse than others?

It seems all subs are affected equally.

All I can conclude is that, for some reason, the flats just aren't matching the subs.

[alexbb]

That happened to me too some time ago. There might have been multiple reasons for this, but I tried to do everything right. Same temperature for lights and dark, same temperature for flats and dark flats. I also see that you're using a newtonian so make sure you don't have any light leaks. I cover the end of my newtonians always (lights, calibration) and try preventing light leak through/around the focuser too.

I suspect you might have some light leaking into the OTA.

[david_taurus83]

I still think those flats arent right Neil. See below a sample of mine. First 2 are a single flat unstretched and stretched. Second 2 are master flat unstretched and stretched. Also note how narrow the foot of my histogram is compared to yours, so theres a large deviation of pixel values in your flats. In both my examples, my 4 corners measure circa just under 21000 ADU and the centre is just over 22000 ADU. I suspect that your example will have a much bigger difference between corners and brightest part in the centre. Its the same camera but the difference is I used a refractor. FWIW I always struggled to take propoer flats on my old 200P.

[Stub Mandrel]

5 hours ago, david_taurus83 said:

I still think those flats arent right Neil. See below a sample of mine. First 2 are a single flat unstretched and stretched. Second 2 are master flat unstretched and stretched. Also note how narrow the foot of my histogram is compared to yours, so theres a large deviation of pixel values in your flats. In both my examples, my 4 corners measure circa just under 21000 ADU and the centre is just over 22000 ADU. I suspect that your example will have a much bigger difference between corners and brightest part in the centre. Its the same camera but the difference is I used a refractor. FWIW I always struggled to take propoer flats on my old 200P.

I think its a bit misleading to use the histogram from DSS, it looks skinnier in other programmes.

I suspect it's either something to do with SharpCap settings OR an issue with my subs. I lean towards the second as I didn't have this issue the first time I use the ASI1600.

The big difference is thAt this is narrowband, the images with the dark spot are with an L filter. Ironically the flats for both sessions were taken with the L filter, so the problem session should be better matched!

See - no problem I can see here (aside from framing and slightly soft focus):

[Stub Mandrel]

Here we go, comparison of flats from 9th and 17th of December  - you can see I got rid of the dust bunnies for the second run... both auto stretched.

The one on the left was used for the heart nebula image above. This makes me pretty confident it is something about my subs, rather than a problem with the flats.

[Stub Mandrel]

On 21/12/2019 at 10:51, alexbb said:

That happened to me too some time ago. There might have been multiple reasons for this, but I tried to do everything right. Same temperature for lights and dark, same temperature for flats and dark flats. I also see that you're using a newtonian so make sure you don't have any light leaks. I cover the end of my newtonians always (lights, calibration) and try preventing light leak through/around the focuser too.

I suspect you might have some light leaking into the OTA.

Could be, but I've been using this setup for three years with a DSLR and haven't had the same issue (I have had nights when the subs have been too poor to get anything though).

[Stub Mandrel]

I think I found the problem.

Looks like I had dark subtraction on in Sharpcap when I took the lights, but not the flats (or the flats had additional subtraction of the wrong dark).

Last night I took my lights with dark subtraction, then took flats with correctly timed darks subtracted and stacked without darks or bias and it worked fine 🙂