ICEINSPACE
Moon Phase
CURRENT MOON
Waning Crescent 32.6%
|
|
14-05-2016, 01:54 PM
|
|
Registered User
|
|
Join Date: Feb 2006
Location: Sydney
Posts: 18,002
|
|
Removing a column defect in image processing
I have a camera which has a couple of minor columns that form a dark line vertically. These are quite common with CCD cameras.
I use CCDstack for calibration. Mostly these just dark subtract out but I am finding sometimes they persist.
Has anyone got some advice about best practice to remove these? CCDstack has a bad pixel map function which I used briefly yesterday and it did not seem very user friendly or effective.
Greg.
|
14-05-2016, 02:06 PM
|
|
ze frogginator
|
|
Join Date: Oct 2007
Location: Sydney
Posts: 22,068
|
|
Quote:
Originally Posted by gregbradley
I have a camera which has a couple of minor columns that form a dark line vertically. These are quite common with CCD cameras.
I use CCDstack for calibration. Mostly these just dark subtract out but I am finding sometimes they persist.
Has anyone got some advice about best practice to remove these? CCDstack has a bad pixel map function which I used briefly yesterday and it did not seem very user friendly or effective.
Greg.
|
Hi Greg, are you using an OSC or mono? Process is a little different.
To make a bad pixel map in CCD Stack, load one sub, a 20min or longer exposure dark frame. Open the bad pixel map dialogue box, click on create map, then go to the reject range algo in the rejection standard drop down menu. Select a range to reject until you paint all the hot pixels and your column as well. When you're happy with the map, enter a name and save it. I usually use the name of the camera and a timestamp like <camera>_20160514 . To apply the map, load your sub then apply it. Tick the box to set the painted pixel to 0. Next step is to interpolate those pixels as they're all black in your sub. Then you're done. You might want to redo your map every 6 months or so, like your dark library. PixInsight has a cosmetic correction process that will do the same.
|
14-05-2016, 02:58 PM
|
|
Registered User
|
|
Join Date: Apr 2010
Location: ardrossan south australia
Posts: 4,918
|
|
have a look at https://pixinsight.com/forum/index.php?topic=2869.0
I used to use Sander's software to clean up column issues with my QHY8 - from memory it worked well. Haven't used the PI script.
Last edited by Shiraz; 14-05-2016 at 08:53 PM.
|
14-05-2016, 03:23 PM
|
Registered User
|
|
Join Date: Jun 2013
Location: Lake Macquarie
Posts: 7,096
|
|
Seems too complicated in PI to me. My mono Canon has a bad pixel column, a bad row, and a bad hot pixel and DSS cleans it up perfectly. You need flats that show the bad column, and darks that show the hot pixel area, and other than that its just ticking the cosmetic cleaning you want DSS to do. It finds bad columns and rows automatically and replaces the bad pixel with an average of the ones on either side. If you dither between exposures it gets even easier. Anyone with sensor flaws should be dithering imho.
|
14-05-2016, 03:23 PM
|
|
Really just a beginner
|
|
Join Date: Jun 2009
Location: Brisbane
Posts: 3,040
|
|
PI cosmetic correction. You know you want to make the change Greg ;-)
DT
|
14-05-2016, 03:59 PM
|
Registered User
|
|
Join Date: Nov 2011
Location: Sydney
Posts: 1,454
|
|
+1 for using Pi. Takes a couple of mouse clicks then integrate with batch processing and you can save then reuse the process settings for future images.
Where's the emoticon for fan boy?
|
14-05-2016, 04:03 PM
|
|
Mostly harmless...
|
|
Join Date: Jul 2008
Location: Brisbane, Australia
Posts: 5,735
|
|
Or just dither more, as I expect you're no doubt fully aware of for DSOs. Perhaps not so helpful for widefield/landscape of course.
|
14-05-2016, 08:37 PM
|
|
Galaxy hitchhiking guide
|
|
Join Date: Dec 2007
Location: The Shire
Posts: 8,278
|
|
MaximCcd has an excellent bad pixel mapping and removal routine (from memory as sub function of its image calibration menu)
You can define partial or full rows or columns, as a numeric entry or simply by using the mouse cursor, as well as define hot or cold pixel cut-off's. While my current 16803 has no bad columns, there are some hot pixels which are non-linear that it cleans up in the blink of an eye.
Last edited by Peter Ward; 14-05-2016 at 08:45 PM.
Reason: typo
|
14-05-2016, 09:28 PM
|
|
Registered Rambler
|
|
Join Date: Jan 2014
Location: Melbourne, Australia
Posts: 399
|
|
Hi Greg,
If you don't mind me asking -- which camera is this? Is it a whole column defect or just partial?
|
15-05-2016, 08:57 PM
|
|
Registered User
|
|
Join Date: Feb 2006
Location: Sydney
Posts: 18,002
|
|
Quote:
Originally Posted by multiweb
Hi Greg, are you using an OSC or mono? Process is a little different.
To make a bad pixel map in CCD Stack, load one sub, a 20min or longer exposure dark frame. Open the bad pixel map dialogue box, click on create map, then go to the reject range algo in the rejection standard drop down menu. Select a range to reject until you paint all the hot pixels and your column as well. When you're happy with the map, enter a name and save it. I usually use the name of the camera and a timestamp like <camera>_20160514 . To apply the map, load your sub then apply it. Tick the box to set the painted pixel to 0. Next step is to interpolate those pixels as they're all black in your sub. Then you're done. You might want to redo your map every 6 months or so, like your dark library. PixInsight has a cosmetic correction process that will do the same.
|
Thanks Marc. I will try your process. I did do most of those steps but not the check box to set the painted pixel to 0 so I got some dark rectangles where the correction took out too much.
Rick went over the cosmetic correction tool in PI so I may also try that if no luck with CCDstack.
Did you find you had to do a 20minute dark for it to work?
Quote:
Originally Posted by Shiraz
|
I'll look that up. Is that the cosmetic correction script for PI?
Quote:
Originally Posted by glend
Seems too complicated in PI to me. My mono Canon has a bad pixel column, a bad row, and a bad hot pixel and DSS cleans it up perfectly. You need flats that show the bad column, and darks that show the hot pixel area, and other than that its just ticking the cosmetic cleaning you want DSS to do. It finds bad columns and rows automatically and replaces the bad pixel with an average of the ones on either side. If you dither between exposures it gets even easier. Anyone with sensor flaws should be dithering imho.
|
Thanks for the tip.
Quote:
Originally Posted by DJT
+1 for using Pi. Takes a couple of mouse clicks then integrate with batch processing and you can save then reuse the process settings for future images.
Where's the emoticon for fan boy?
|
Quote:
Originally Posted by DavidTrap
PI cosmetic correction. You know you want to make the change Greg ;-)
Yes I already went to one of Rick's tutorials. I tried once since and no luck but I am sure I can get it to work. One issue was the lack of feedback on applying calibration. With CCDstack you see the result immediately.
DT
|
Quote:
Originally Posted by Peter Ward
MaximCcd has an excellent bad pixel mapping and removal routine (from memory as sub function of its image calibration menu)
You can define partial or full rows or columns, as a numeric entry or simply by using the mouse cursor, as well as define hot or cold pixel cut-off's. While my current 16803 has no bad columns, there are some hot pixels which are non-linear that it cleans up in the blink of an eye.
|
Thanks Peter. So far I have not used Maxim. I still use CCDsoft or Sky X.
Quote:
Originally Posted by RobF
Or just dither more, as I expect you're no doubt fully aware of for DSOs. Perhaps not so helpful for widefield/landscape of course.
|
Yes dithering is probably what changed. Although I found the lines vanished with regular darks. These were longer exposures and the darks don't match exactly for length.
Quote:
Originally Posted by Eden
Hi Greg,
If you don't mind me asking -- which camera is this? Is it a whole column defect or just partial?
|
ST8300.
Greg.
|
16-05-2016, 09:46 AM
|
|
ze frogginator
|
|
Join Date: Oct 2007
Location: Sydney
Posts: 22,068
|
|
Quote:
Originally Posted by gregbradley
Thanks Marc. I will try your process. I did do most of those steps but not the check box to set the painted pixel to 0 so I got some dark rectangles where the correction took out too much.?
|
The step to interpolate the missing pixels back with the neighbouring ones is the important one. Otherwise you end up with holes in your subs and they become lines after you've registered and stacked them so best to take care of all the cosmetics before debayering makes them bigger or registering makes them worse.
|
16-05-2016, 11:39 AM
|
|
Registered User
|
|
Join Date: Feb 2006
Location: Sydney
Posts: 18,002
|
|
Quote:
Originally Posted by multiweb
The step to interpolate the missing pixels back with the neighbouring ones is the important one. Otherwise you end up with holes in your subs and they become lines after you've registered and stacked them so best to take care of all the cosmetics before debayering makes them bigger or registering makes them worse.
|
Thanks Marc.
I will be implementing this next I sit down to process the mountain of images I have at the moment.
I probably have 2 weeks solid processing's worth of data at present. So I want to get it all perfect and not have to redo stuff later.
Greg.
|
Posting Rules
|
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
HTML code is Off
|
|
|
All times are GMT +10. The time is now 10:14 PM.
|
|