Reducing FPN

chripell

Hi, the site (http://www.zwoptical.com/Eng/Cameras/ASI120/features.asp) says no FPN but I guess it would be more fair to say low FPN. I'm 100% sure that "Column Correction" feature kicked-in (I saw frames without it and the column-wise FPN is much much worse and anyway I did some HW flips before acquiring). I'm attaching the mean of 56 10" darks with histogram equalized (so the contrast in really enhanced), a single frame 10" and the result of stacking 63 frames. Unfortunately there is some more vertical banding. I just subtracted the mean dark from the single frames.

I'm asking more experienced astro imagers: a flat field could help? It's quite some hours to build a flat field box and I don't see vignetting so I'm asking first before starting. :-) Other suggestions? The bars don't seem to be regular, so I don't see how one can use a notch Fourier filter. More darks?

Thank you very much!

PS. Haven't seen much FPN on the ASI120MC bu until now I worked only with 8 bit (Haven't written my debayer code yet).

PS. attachments don't work, it says "Sorry, the board attachment quota has been reached" so:

dark mean: http://farm3.staticflickr.com/2636/13078605354_9ef2e39495_o.png

a frame: http://farm4.staticflickr.com/3172/13078608084_fc49d657c9_o.png

final image: http://farm8.staticflickr.com/7328/13062814144_5bcf95e32f_o.jpg

ASI120MC example: http://farm3.staticflickr.com/2166/13066237494_c6458759d5_o.jpg

Sam@ZWO

yes, chris

the problem is the column correction is not prodicted

so eachtime the dark frame may different

that's the problem

chripell

This means that dark and normal frames should be acquired *without* stopping acquisition?

Sam@ZWO

I think more critical that dark and normal frames should be acquired *without* exposure change.

robbeh

Sam,

having the same issue I want to know exactly what you mean that "frames should be aquired without exposure change". I do not change exposure time whaen taking darks and lights, nevertheless I have this banding noise still persistent.

Thanks, Robert

chripell

Of course I don't change gain and exposure time between darks and normal frames. And I don't stop exposure either. Just change time from 10 to 1 and back to 10 to make focusing easier (with a feedback time of 1s instead of 10 it's much much easier). As far as understand from the DS this shouldn't retrigger column correction. Anyway I'm trying to understand what's going on.

riklaunim

If you really want to play with DS imaging with this camera then just shoot a "dark frame" AVI, stack it and use as master dark for object stacking. There is no extra "magic" here.

chripell

Of course I did it (well, I shot many separate PPM files instead of an AVI because they are much easier to handle) but id doesn't cancel completely the vertical bar pattern. Maybe it's an mistake of mine, but keep in mind that this neat little cameras do a lot of digital processing inside them (black level subtraction, vertical and horizontal noise reduction, both digital and analog gain just to cite some of them) so "things are not always as they seem". I have quite of lot of darks with exposure of 1,4 and 10s (gain 85) from Saturday, so I'm going to analyze them.

chripell

Some notes:

1) the vertical bar pattern is quite stable from a night to the other. Of course it's always better to do the acquisition in the same run, but evidently the correction calculated by the chip it depends mainly on the camera (the length of the feed line from the column to the amplifier I presume).

2) it's pure FPN, you can see it in the bias frame (T exposure = 64uS the shortest available). This is very good because I can try to minimize:

MD == master dark

MB == master bias

VAR(MD - k * MB) for the scalar k

In this way I can see if there is some amplification of the FPN that renders simple subtraction of the MD from the image ineffective as far as FPN is concerned. I'm pretty sure, like robbeh confirmed, that even if exposure and gain are the same, simple subtraction does *not* completely correct vertical bars FPN. To see it you have to work in 16 bit and do a gamma correction of the output, but it's there.

robbeh

Hi there,

last night I tried to take Sams words exactly as he said, and did not change the exposure between loght and darks. That means I took lights (series of single fits files in Firecapture) and at the end I did *not* stop capture process and just covered the scope to take the darks in the same run. And .. What should I say ... FPN was nearly completely reduced after dark frame substraction. Needs further evidence, though, but maybe thats the solution.

Robert

chripell

I'm starting to think that this is best it can be done (try to stretch gamma of the final stack in 16 bit mode). If there is still some defective column, it has to be eliminated with the "bad column" option of deep space stacker.

Sam@ZWO

thanks for you guys feedback.

@chris

I think there is bug of the camera driver when change from long exposure back.

this will cause the vertical line very seriously.

we will update the driver and let's you try

chripell

OK, thanks! :-)

chripell

With the latest driver the situation has improved indeed :-) In 2 nights I never had a "really really bad frame" (guess when column correction doesn't kick in) and the dark subtraction does a better job to FPN. See the attached M64, 90*10s light and 110*10s dark with my 10" f/4.7

robbeh

Sam, is or will there be an update for the SDK, too? Thinking of FireCapture ... Thanks

robbeh

Hi Sam,

I installed the latest driver and yesterday I tried DSO imaging with ASI120MM again, here are some 2 minute exposures of M96, taken with Firecapture (latest version, using built-in SDK). All images were consecutively shot in the same run. You see vertical noise appearing and disappearing in a random manner. What can be done about this?

Thanks Robert

Sam@ZWO

Hi Robert

what is the exposure time?

we are going to release a new version of our driver which should fix this problem

you can contact me: sam.wen(at)zwoptical.com

Sam@ZWO

here is the newest SDK

http://zwoptical.com/software/ASI%20SDK_Setup.EXE

you can install and put the ASICamera.dll to the directory of FC

robbeh

Sam,

thanks for the new driver / SDK. I will give it a try ASAP.

Exposure was 2 minutes, each.

Thanks, Robert