Dark offset bounces around in SER

[ChrisR Oz]

Hi Robin,

I was trying to collect SER files of Saturn and its moons in SharpCap, at various exposures, with the idea of building a HDR Composition to show both Saturn and perhaps 7 moons. Perhaps misguided - we'll see. The 5 ms and 100 ms exposure SER files look as they should. But the 1s exposure file is very strange.

Firstly, 1s seems to be upset by Feature (COG) Tracking, with the FPS dropping about x10. But that's a story for another post ... However, I mention it here as Feature Tracking was interfering in some way.

The 1s SER file shows two types of frame. The first is like normal SER files (e.g. the 5ms, 100 ms done here) having a Brightness offset plus background peak seen in the histogram (left below). The odd ones, not seen elsewhere, have a truncated histogram, where the background/offset peak seems to have been subtracted by a few % (right below). These bad frames seem to outnumber the good ones by about 10 to 1.

Two frames in 1s SER file

SER-1s-odd-1.JPG (81.74 KiB) Viewed 7461 times

I checked that I did not have any feature active that may have been modifying frames on the fly. The attached camera settings file shows no active: Flat, Dark, Background Subtraction, Banding Suppression, neutral colour Balance, etc. And I see the same effect for 1s SER acquired in RAW8 or RAW16. However, High Speed mode was still enabled for all.

Unfortunately, as all has been going so well lately, I am still on version 4.0.9268.

Any suggestions?
Cheers, Chris.

[admin]

Hi Chris,

thanks for posting the camera settings - I was hopeful that maybe something in there would be able to take the blame for this, but nothing obvious.

I guess the first thing we have to try to work out is whether the images are being manipulated inside SharpCap or before they get to SharpCap. I am suspecting before it gets to SharpCap, since the code for things like feature tracking and other processing is designed to make it hard for me to make a mistake and pass through processed frames to capture... Not impossible, perhaps, but not an easy mistake to make. Also the fact that everything works at lower exposures but not at 1s hints in that direction - there is really no difference in the processing of the frames depending on the exposure.

ZWO cameras switch over from live mode into still mode at 2s exposures in the latest SharpCap, so that shouldn't be the issue...

I noticed that you are capturing at 16 bit, but that the histograms shown are up to 255 - I presume that is just an axis scaling issue on SERPlayer rather than any real issue. Also, looking at those histograms it is hard to tell if the peak level drops when the histogram moves left - ie is everything being offset by -1000 ADU (or whatever), or is it being rescaled.

I did a quick test on an SER capture on the closest camera I have (ASI071MC) in 16 bit, ROI, SER format, 1s exposure and saw no issues - I was not running feature tracking though. It may be worth having a closer look at the frames by converting the SER to FITS using PIPP and then poking around the FITS files with FITSliberator or similar.

cheers,

Robin

[ChrisR Oz]

Hi Robin,

I used PIPP to convert the SER to FITS (no Debayer or processing) and looked at the same two frames (134, 135) in FITS Liberator and ASI FitsView. While the ASI FitsView agrees with the SER Viewer histograms for these frames, FITS Liberator does NOT. It shows the truncated histogram for both, Checked the FITS headers in both programs and they are the same. No indication of an odd BZERO or something like that. There is something odd in the frame data that confuses SER Viewer/ ASI FitsView sometimes and FITS Liberator all the time. Screen grabs of all these attached below ... The plot thickens ...

Cheers, Chris.

[admin]

Hi Chris,

could you share the original SER file with me please? I can have a poke around and see if I can spot anything odd in the data. It's odd though since there is no per frame data in the SER file other than the pixel values and the timestamp...

cheers,

Robin

[ChrisR Oz]

Hi Robin,

The SER file can be found here: https://cloudstor.aarnet.edu.au/plus/s/kVEMk6onaeMSPmo

Cheers, Chris.

[admin]

Hi Chris,

I'm not sure how useful this will be, but having looked at your SER file and camera settings, I think I would conclude the following

1) The effect is real - some frames have a brighter background than the majority of them, moving the histogram right.

2) This also reduces the size of the histogram peak at pixel brightness=0, which changes the look of the histogram further due to auto-scaling of the vertical axis

3) Frames 113/114 show this nicely, with the average pixel value in 114 being 500 ADU higher than 113. See screenshot below..

4) The bulk of the frames - the darker ones - are badly black level clipped (lots of zero pixels), although this probably will not affect the type of imaging you are doing

5) Nothing you have set in your camera settings should make SharpCap change the image data, so the different data is most likely coming straight from the ZWO camera/SDK with these glitches.

6) It seems like a brightness related issue (black level/offset in other brands) - either the brightness you have set is not being applied to most frames, or an incorrect, higher brightness is being applied to some.

There have been updates to the ZWO SDK since the version you are using in SharpCap 4.0.9268, so I think the first thing to do would be to try to update SharpCap and see if that helps. Beyond that, it may be worth trying changing to 2s exposures with a lower gain - 2s pushes ZWO cameras in to the still mode option of the SDK, which could, just possibly, bypass some error of this sort.

cheers,

Robin

Capture.PNG (115.38 KiB) Viewed 7351 times