Strange lines in OSC stacked image

dchamberlain

Well-known member
Nov 24, 2017
51
1
Hi,

I am using a ZWO ASI1600MC-Pro on an Astro-Tech 14" RC. I took 53x180 sec subs of M81 the other night and the subs looked fine, aside from some tracking issues. The subs were taken at 200 gain/50 offset using SGP. I took flats, dark-flats, and darks, 40 of each. Ran them through WBPP using galaxy settings and the result I got was an image of the galaxy with colored streamers, almost like rain in a windstorm, running through the image (see attached non-linear jpg).

Can someone tell me what is the cause of these streaks and how to get rid of them?

Thanks for any advice!

DaleM81.jpg
 

pfile

PTeam Member
Nov 23, 2009
5,321
64
usually this is caused by incomplete correction of hot pixels together with differential flexure in your system + no dithering.

if you look at the calibrated subs, do you see hot pixels? if you load the registered subs into blink and run them as a movie do you see hot pixels marching in diagonal lines?
 
  • Like
Reactions: ngc1535

ngc1535

Well-known member
Feb 1, 2014
506
54
Yeah... walking fixed pattern "noise." The key is to do random direction dithering. It is likely these are unguided exposures that over time drift due to polar alignment error (which as Rob notes could be flexure, even a stiff system not polar aligned will do this).

On another note of interest- sometimes when things are calibrated well and everything is working in processing it actually *reveals* issues that would not otherwise be seen when things are *not* done properly or working. There have been some other issues with people having subtle flat errors, that likely would not have been visible had the calibration process and everything else been *incorrect*. Things correctly done can lead to the "next level" of issues...just like this case. :)
-adam
 

dchamberlain

Well-known member
Nov 24, 2017
51
1
usually this is caused by incomplete correction of hot pixels together with differential flexure in your system + no dithering.

if you look at the calibrated subs, do you see hot pixels? if you load the registered subs into blink and run them as a movie do you see hot pixels marching in diagonal lines?
I did run the registered subs through blink and it does show the noise moving through. But if I run the uncalibrated subs through blink it looks fine, so I can only assume the subs calibrated by the darks are putting the noise in? Now there is no question my tracking was off, so my next question is for a permanent pier setup how often do I need to polar align?

Dale
 

pfile

PTeam Member
Nov 23, 2009
5,321
64
well maybe because they aren't moving around, you may not be able to easily identify the hot pixels in blink on the unregistered, uncalibrated images. if you zoom the blink window in maybe you will be able to see them better. or, you should be able to open a couple of your images on the desktop, zoom in and pixel peep both an uncalibrated image and a calibrated image and take a look at what's going on with respect to hot pixels.

it would be very unusual for hot pixels to be introduced by dark subtraction, since, well, you're subtracting the dark from the light. you'd have to subtract a negative number in order for the result to come out more positive than the input... and pretty much by definition neither the dark or the light will have negative numbers in them.

the problem might not be tracking or polar alignment. a polar alignment error would manifest as field rotation, so if you don't see a steadily increasing angle in the black areas of the registered subs, the problem probably is not polar alignment. it could be tracking related if you are not guiding.

but if you are guiding with an external guidescope, and the guidescope is not rigidly attached to the telescope, or the guide camera is not rigidly attached to the guidescope, then the problem is likely to be differential flexure. note that the problem could also be that the focuser is sagging or the imaging camera is not rigidly attached to the focuser. the flexure could happen in the guider or in the main imager, and it's the difference between the two that causes the drift. as far as the guider is concerned it's doing an excellent job keeping the star centered, but the main camera and the guide camera are ever so slightly moving with respect to one another.

anyway you might be able to solve this with more aggressive pixel rejection parameters during ImageIntegration. or you can use CosmeticCorrection to try to remove the hot pixels from the calibrated subs, then register the cosmetically corrected subs and integrate them.
 

dchamberlain

Well-known member
Nov 24, 2017
51
1
well maybe because they aren't moving around, you may not be able to easily identify the hot pixels in blink on the unregistered, uncalibrated images. if you zoom the blink window in maybe you will be able to see them better. or, you should be able to open a couple of your images on the desktop, zoom in and pixel peep both an uncalibrated image and a calibrated image and take a look at what's going on with respect to hot pixels.

it would be very unusual for hot pixels to be introduced by dark subtraction, since, well, you're subtracting the dark from the light. you'd have to subtract a negative number in order for the result to come out more positive than the input... and pretty much by definition neither the dark or the light will have negative numbers in them.

the problem might not be tracking or polar alignment. a polar alignment error would manifest as field rotation, so if you don't see a steadily increasing angle in the black areas of the registered subs, the problem probably is not polar alignment. it could be tracking related if you are not guiding.

but if you are guiding with an external guidescope, and the guidescope is not rigidly attached to the telescope, or the guide camera is not rigidly attached to the guidescope, then the problem is likely to be differential flexure. note that the problem could also be that the focuser is sagging or the imaging camera is not rigidly attached to the focuser. the flexure could happen in the guider or in the main imager, and it's the difference between the two that causes the drift. as far as the guider is concerned it's doing an excellent job keeping the star centered, but the main camera and the guide camera are ever so slightly moving with respect to one another.

anyway you might be able to solve this with more aggressive pixel rejection parameters during ImageIntegration. or you can use CosmeticCorrection to try to remove the hot pixels from the calibrated subs, then register the cosmetically corrected subs and integrate them.
I believe I found the problem. After integrating the light subs without calibration I saw no noise from hot pixels. Remembering anytime I have a problem with integrating images it always seems to point to the dark frames, I decided to take a closer look at the individual dark frames. I found that the cooler stopped cooling midway during dark frame capture. Started out with -15C and ended up with 8.5C. Now I am retaking the darks but keeping tabs on the sensor temps.

Then I will try again with preprocessing.

Stay tuned!

Dale
 

pfile

PTeam Member
Nov 23, 2009
5,321
64
i wonder if your lights suffered the same fate - if the lights are super high temp and the darks were an average of low temp and high temp, that would leave undercorrected hot pixels in the calibrated result.

in fact if the uncalibrated lights have really hot hot pixels they may have been rejected during integration whereas with the bad dark-calibrated frames you sort of had 'warm' pixels left over in the calibrated images and they weren't outlying enough to be rejected.

rob
 

dchamberlain

Well-known member
Nov 24, 2017
51
1
After creating a new set of 180-second darks and running them through calibration, debayering, registration, and integration, I came up with the following image which I am generally pleased with. Now I am going to break out my OAG and guide camera and try to get tracking to be a bit better. But I will be paying much closer attention to the darks from now on.

Thanks for everyone's help. It led me to the right place.

Dale
 

Attachments

  • Like
Reactions: pfile