PixInsight Forum (historical)
PixInsight => General => Off-topic => Topic started by: astropixel on 2011 February 07 14:49:13
-
What is the best way to deal with the milky background associated with some DSLR cameras in PI. I notice that with a significant non-linear stretch my images reveal uneven whiteness across the image. To a large extent I have minimized the effect with Windorized or Percentile clipping, but the residual makes it impossible to stretch and obtain brightness and colour. Background extraction tools have not helped, and I have tried various different approaches during non-linear processing with little success.
Basically, this artefact prevents getting the best out of significant data.
-
are you sure this is a camera artifact and not due to light pollution? i'm not sure i've ever seen uneven whiteness but i definitely get some strange gradients sometimes...
-
Could you post an example image. I think I dont see this problem with my Canon EOS 40D.
Georg
-
Thanks for taking an interest in this.
I extracted the L* channel to highlight the issue. I will resist the temptation to think out loud, except to say that this translates to a very muddy background.
-
Have you tried not calibrating your subs to eliminate the possibility that the Master Calibration files are not causing this problem?
Cleon
-
I'm hesitant to say that it's a calibration problem. I don't want it to be >:( The initial stretch produces a very flat image - evident in the RGB and extracted L*. But this is only achievable with Windsorized clipping on the large set - 247 lights, and Percentile clipping on smaller experimental sets - 3 lights.
Integration with Linear clipping, the problem was evident from the first stretch - I assumed to have selected the wrong clipping algorithm.
I have no reservations concerning the bias and dark frames - they are easy to acquire and process. But, I dread the thought that it may come back again to the flats that I am producing???
-
I wonder if this is the general background "dust" in this region of Orion. See http://de.wikipedia.org/wiki/Datei:Orion_Belt.jpg for example.
Georg
-
Georg may have the answer. Just what is the the total time in the 247 lights?
Do you examine the three channel list created with II after integration?
Cleon
-
I did think of dust, but discounted it on the basis of the total exposure time - just 67 mins, made up of 10, 15, 20 and 30 second exposures - f/3.2, iso800.
The other problem was differential flexure. Exposures are unguided, but alignment was unusually good and I should have dithered the frames periodically - could this contribute to the problem? Originally the image was streaky, but I managed to process a lot of this out. Maybe it shows up with non-linear processing and more extensive stretching.
If it is dust, then this is all too funny - I've spent hours attempting to eliminate good data. :'(
-
Here is a masked version just to illustrate the extent and nature of the problem. Is this recoverable?
-
check this widefield - the entire area is full of dust...
http://www.flickr.com/photos/31986095@N05/5281522077/
it does show up even in shorter exposures. and it looks... kind of milky when the signal is low...
-
Comparing images, I see faint evidence of dust areas, but overall I think this is an artefact - possibly the result of trying to squeeze to much out of the image. I've just about given up on this set.
-
Here is the same field - I don't think you have milkyness...I see dust ;-)
http://www.flickr.com/photos/21315110@N00/5343866972/
Don't look too closely at my picture...it is not a very good one...
-
Very nice David. Thanks everyone, I concede - dust! Stunning area - just needs more data.
This time I started out with ACDNR following the first non-linear stretch.
Given my skills, this is as much as I can get out of it - a long way to go, I understand.
Comments/feedback most appreciated.
-
I think you've got the start of a good image there. I like the colour in the Flame and the Orion nebula. Stick with it and get more subs.
Regards
Dave