roundel325
member
Reged: 10/09/07
Posts: 31
Loc: Healy, AK
|
|
I seem to get this diagonally patterned color noise in all my shots. It comes out very quickly when trying to stretch in PS. It's even faintly visible in the stack as it comes out of DSS. My camera is a Nikon D300. I've attached a 100% crop from a veil nebula shot I captured last night. 13x10min, 4xdarks, 15xflat, 15 degrees F ambient. I know 4 darks is probably not enough, but I have done stacks with 15 darks on other shots and don't see much if any difference.
How do I improve this? More subs? More darks? Fancy noise reduction in processing? Get a Canon camera? Or is this totally normal?
Thanks,
Jared
|
Mike Clemens
Post Laureate
Reged: 11/26/05
Posts: 4277
Loc: Wasilla, Alaska 61N
|
|
better temp match on darks, more darks & bias, larger dither between sets of frames, more light frames, median combines
|
jmX
super member
Reged: 04/23/09
Posts: 166
Loc: Orange County, CA
|
|
Are the streaks also in the direction the stars drifted between the first and 13th photo you took?
If so, I get the same thing with my Canon. The dark frame subtraction still leaves some noise for me, and if the drift is real slow in on direction I end up with that smeared noise. I end up heavily processing the "black" areas of my photos until its slightly more pleasant.
--------------------
Jon
C6N + CG5
Skywatcher Equinox 80 + CGEM
http://jmx.ls1howto.com
|
roundel325
member
Reged: 10/09/07
Posts: 31
Loc: Healy, AK
|
|
Yes! It does appear to line up with the direction of star drift. So does that mean it's mounting problem and won't be cured by more darks, etc? Do other stacking programs handle this better the DSS?
|
jmX
super member
Reged: 04/23/09
Posts: 166
Loc: Orange County, CA
|
|
I'm using nebulosity, so its not the software. It's normal for the mount to drift some, so it's not really a mount issue.
Mike Clemens will be the man to listen to here as far as how to work around it. I'm not sure "more light frames" is the answer, as I've had this with 60+ lightframes before. Its probably down to having more accurate dark frames, and/or using bias frames (which I've never done).
--------------------
Jon
C6N + CG5
Skywatcher Equinox 80 + CGEM
http://jmx.ls1howto.com
|
Mike Clemens
Post Laureate
Reged: 11/26/05
Posts: 4277
Loc: Wasilla, Alaska 61N
|
|
> I'm not sure "more light frames" is the answer
I think its fixed position sensor noise which is exacerbated into streaks by slight mount drift. Dithering the position of the camera/subject, then using a median type combine eliminates this, given more light frames.
|
roundel325
member
Reged: 10/09/07
Posts: 31
Loc: Healy, AK
|
|
Thanks Mike,
Is there a way to do this without physically moving the camera or scope between every frame. I generally am too much of a wuss to stand out in the cold (you know what I mean, I'm 64N) babysitting the rig. Plus, I need my beauty sleep.
|
roundel325
member
Reged: 10/09/07
Posts: 31
Loc: Healy, AK
|
|
Also, I just restacked the image with no dark frame subtraction and it looks pretty much exactly the same. Still streaky.
As a general newbie question, does the master dark frame get subtracted from each sub before stacking or from the entire stack?
|
jmX
super member
Reged: 04/23/09
Posts: 166
Loc: Orange County, CA
|
|
Darks get subtracted from each light frame, THEN they are stacked (at least in nebulosity).
Sounds like your dark frames were taken at a different temperature than the lights if they're having no effect. You took them with the same iso and exposure time right?
--------------------
Jon
C6N + CG5
Skywatcher Equinox 80 + CGEM
http://jmx.ls1howto.com
|
roundel325
member
Reged: 10/09/07
Posts: 31
Loc: Healy, AK
|
|
Yes. Darks had all the same settings. When I started the imaging run the ambient temp was 19 degrees F and this morning when I took the darks it was 15 degrees F. Is that too much of a temperature deviation?
|
s58y
Postmaster
Reged: 12/12/04
Posts: 5509
Loc: Eastern NY
|
|
One possibility is to try rotational dithering (as well as regular translational dithering) between subs. This is easier said than done, however, and can greatly reduce the FOV.
--------------------
Hutech 30D, SBIG ST-402 autoguider
SV80S, TV102iis
Old camera lenses: 800mm f/5.6, 180mm f/3.4
AP900, Barndoor tracker
http://www.pbase.com/s58y
|
Tonk
Post Laureate
Reged: 08/19/04
Posts: 4356
Loc: Leeds, UK, 54N
|
|
Try a sigma kappa combine with sigma at 1.75 to 1.50 with at least 5 iterations
If that doesn't do too much in the way of improvement - then try sigma kappa median
--------------------
Televue 85, GM-8/Gemini, Canon 40D (unmodded), Canon 450D (modded w/Astronomiks clip-ins - UV/IR, OWB)
Coronado SM60/BF10, Baader Herschel Wedge
Leeds Sky Clock Ripon Sky Clock
|
roundel325
member
Reged: 10/09/07
Posts: 31
Loc: Healy, AK
|
|
Thanks for the tip Tonk. Kappa-sigma didn't do much at either 1.75 or 1.5, but kappa-sigma median at 1.5 made a noticeable improvement. Still quite streaky, but an improvement. Tonight is supposed to be similar temperatures and cloudy, so I'll try to get 30 darks (5 hours!) and see if that kicks things in the right direction.
|
Psyire
scholastic sledgehammer
Reged: 06/24/07
Posts: 979
Loc: 55* North
|
|
Tonk, would you be able to give a quick explaination as to why that would make a difference in the noise?
--------------------
Celestron CPC 1100 XLT, Sky-Watcher Equinox 80ED
TV 31T5-Nagler, 8&13mm-Ethos
EarthWin Binoviewers w/ 24mm Panoptics
Elusive Photons.com
|
Tonk
Post Laureate
Reged: 08/19/04
Posts: 4356
Loc: Leeds, UK, 54N
|
|
Quote:
Tonk, would you be able to give a quick explaination as to why that would make a difference in the noise?
Ah - because it isn't noise! The streaks are residual hot pixels (the remaining difference between the lights and darl frame hot pixels) - this is still signal - albeit unwanted. These residuals are traveling slowly across the stack because of polar misalignment or flexure etc and as such don't exist at all the same pixels across the stack. Each pixels only sees the hotpixel for some of te time
Sigma-Kappa is a rejection algorithm and throws out outriders in the set of values for each pixel. These traveling residual hotpixels might be seen as outriders if you are lucky. Things that defeat this
1) Residual value is roo small - not seen as an outrider
2) More than one residual travels though the same pixel across the stack. If too many do this then the "outrrider" signature deminishes
1.75 to 1.5 might be too conservative - I would be tempted to try 1.25 to 1.0 (these are standard deviation numbers)
--------------------
Televue 85, GM-8/Gemini, Canon 40D (unmodded), Canon 450D (modded w/Astronomiks clip-ins - UV/IR, OWB)
Coronado SM60/BF10, Baader Herschel Wedge
Leeds Sky Clock Ripon Sky Clock
|
groz
Pooh-Bah
Reged: 03/14/07
Posts: 1078
Loc: Duncan, BC
|
|
Quote:
How do I improve this? More subs? More darks? Fancy noise reduction in processing? Get a Canon camera? Or is this totally normal?
More lights frames, or go to a darker sky.
We have shot with 2 dslr cameras here for most of the last couple of years. Chris with the Nikon D-50, and me with the Canon 350. Both show this kind of streaking, but the canon shows it MUCH worse than the nikon.
To get rid of it with my canon is really easy. 50+ light frames, and 25+ darks on any given target, and it really starts to go away. When I get 100 lights on a target, it's hardly visibile anymore. for chris with the D-50, cut those numbers in half, it's disappearing well after 25 lights, and, starts to be completely gone after 50 of them.
As an example, I did an M51 last winter, on the first evening I got 25x5 minutes, and, that ended up being a reasonably nice shot, but, the diagonal streaks really showed up trying to stretch the faint parts of the arms. On the second night, I added another 30 lights, and then it started to stretch really nicely. A couple nights later I added 50 more, and it turned into the best image I've ever gotten out of the 350, until a week later. A week later we got the kit out under the darkest sky it's ever been under, and wowowow, what a difference that made. 30 light frames in those condition, a spectacular shot of m101, and no sign of the diagonal streakies that I always have shooting from home.
|
roundel325
member
Reged: 10/09/07
Posts: 31
Loc: Healy, AK
|
|
Groz,
My skies are pretty dark. A 20 minute sub at 1600iso puts my skyfog hump half way up the histogram. I was under the impression that it's not the number of subs that matters, it's the amount of integration time. And that longer subs were actually better. I shoot 10 minute subs from my front yard and I just don't get enough clear nights to dedicate 50x10min to a single target. I see a lot of really clean images on this forum that were captured with a fairly small number of long subs. Like Mike's excellent NGC7331:
http://www.cloudynights.com/ubbthreads/showflat.php/Cat/0/Number/3387218/page/2/view/collapsed/sb/5/o/all/fpart/1#Post3390564
|
roundel325
member
Reged: 10/09/07
Posts: 31
Loc: Healy, AK
|
|
Well, I reprocessed the veil shot with 44x10min darks takeen last night to within 5 degrees F of the light frames. I see a small improvement, but still not great. 100% screenshot shows the difference. 44x darks on the left, 4x darks on the right, each frame stretched equally.
|
Tonk
Post Laureate
Reged: 08/19/04
Posts: 4356
Loc: Leeds, UK, 54N
|
|
Another tactic -
Divide your images into 3 stacks (called "one" "two" "three")- then deal out the 1st in "one", second in "two", third in "three" and then 4th back into "one", 5th into "two" etc
Stack each of these 3stacks separately using sigma-kappa-median, the combines the resulting 3 images using ordinary median stacking. This aids the rejection of residuals even further (subject to the same 2 caveats I gave above)
--------------------
Televue 85, GM-8/Gemini, Canon 40D (unmodded), Canon 450D (modded w/Astronomiks clip-ins - UV/IR, OWB)
Coronado SM60/BF10, Baader Herschel Wedge
Leeds Sky Clock Ripon Sky Clock
|
Mike Clemens
Post Laureate
Reged: 11/26/05
Posts: 4277
Loc: Wasilla, Alaska 61N
|
|
Thanks Roundel, too kind.
Looks like fixed pattern noise being dragged across a series of frames. Dither more, go out and spin your camera 180 degrees for a couple series.
|
Previous
Index
Next
Threaded
Edit
Reply
Quote
[Re: 