Imaging with Dither

[dhalliday]

Maybe questions like this are "Off Topic" here...

But I am reading about the value of dither,but very little about how to accomplish it.
I have purchased Nebulosity,and will try to have it apply dither to my SSAG system.
Could folks comment on this,or better ways to achieve this?
I use a DSLR,image mostly at 1000-1280mm,guided,and often get linear "streaking" which I am told dither will prevent...
Thanks

Dave

[georg.viehoever]

Hi Dave,

if the "streaking" is similar to the bright/dark horizontal bands that my (and may other) Canon EOS camera produces, you may want to try the Canon Banding Reduction Script that comes with PI 1.5.9. . I sometimes read that people recommend to use dithering to get rid of the banding, but in my case this did not help. Thus the software solution...

Georg

[Simon Hicks]

Dave,

If the streaking is always in exactly the same place on each image then dithering will help.....when you align and stack then you are effectively moving the streaking around relative to the stars and therefore you will average it out to some extent.

However, I think the Canon banding is NOT always the same...its more random, i.e. if you take two bias frames (or darks) and do a massive hitogram stratch so you can see the banding then it won't be in the same position, it will change in a random way from frame to frame. If this is true (and it should be easy to check) then dithering will not help.....you just need to take lots of subs, or use the Canon Banding Reduction Script.

The dithering will help get rid of things (defects) that always appear on the same part of the CCD no matter where it is looking.

Cheers
         Simon

[Harry page]

Hi

I dither all the time ( I think I do, or I might not  )

As said dithering will remove all FIXED pattern noise Ie Hot pixels , when used with a sigma clip routine,,  Very handy indeed


Harry

[dhalliday]

Well,at the risk of shaming myself ...
here is a (particularly BAD) example;

http://www.flickr.com/photos/daveh56/3868338463/sizes/o/in/set-72157622188544040/

So I do not think this is the dreaded "banding" problem....
When I stack my lites with some form of "SigmaClip" (median,OR exclude) in DSS things improve (quite a bit,maybe 70%..)...but what sigma/kappa etc etc to use,etc is all Greek to me...but I can play with that part...
My own personal theory is that this is caused (somehow) by my autoguider correcting for bad PA...and in each frame I see my stars slowly moving a bit...(in the direction of the streaking...) ??

Sort of like slow motion field rotation (ie stacked F rotation...) ??...I don't know..!!
I am further confusing myself as I write this...

ANYWAYS...I am told/hoping/pretty sure, that dither will fix this...(?? "pattern noise"..)

But HOW to (best) dither is my question...Nebulosity (software) has been suggested,but I am not 100% convinced...
Since everyone here seems fairly advanced...I feel a bit like a kid asking Bill Gates for advice about how to set up a lemonade stand...
(LOVE these Emoticons...)

But I am asking,and again, its mostly about HOW folks achieve dither,and how strongly they believe in it..
(And why I don't see more about it...)
PS I understand that the basis of "Drizzle" is a dithered image...

Dave

[georg.viehoever]

Hi Dave,

I agree, this is fairly different from my beloved Canon Banding Problem. I have no idea what could be the cause of this.
Georg

[Simon Hicks]

Hi Dave,

I agree with Georg...this is nothing like the Canon banding.

I have seen something like this (though not as severe) only once. I did a stack of Comet Lulin which was moving quite fast against the background stars. I stacked in DSS on the comet and stars....so it held the comet still, stacked and then stacked on the stars....and somehow combined them to give pinpoint stars and a non-streaked comet. However, it left this sort of trailing in the depths of the background.

So it must be due to a small drift in the tracking during the exposures so that the first and last frame are slightly shifted diagonally with respect to each other. But on top of that there is something that is on the CCD that must be constant....and therefore is being seen as streaks when you stack.

Have you looked at a single sub stretched to the same extent as the image you have posted? Could you post one for us to look at.

Cheers
         Simon

[dhalliday]

Here is one of the RAW frames,stretched,as a JPEG
http://www.flickr.com/photos/daveh56/4087398456/sizes/l/


Thank goodness for flats !!
Ant ideas ??

[Simon Hicks]

Hi Dave,

That is strange. Firstly your sub is a much wider field of view, and a rotated version of the original one you posted....took me a while to work that out....but I got there in the end.

From what you have said it definitely sounds like you have some sort of drift over many frames, and this is somehow contributing to what you are seeing. But there must also be something on the CCD that is being streaked by the drift. So maybe you need to play with the sigma clip values, try to make the rejection a bit stronger or try some of the other rejection methods in DSS. Also, maybe play around with the hot and cold pixel rejection parameters in DSS....again try stronger rejection and change the pixel size. It may require you to do a few more stacks in DSS to experiment, but it could be worth it.

Good luck!

Cheers
         Simon

[papaf]

It looks quite clear to me. You have constant drift, and there's even evidence of it in the single raw frame you provided. Obviously, when you stack some of them, the drift will be longer and the noise from the camera will be "appended" forming a longer streak.
Dithering should help in your case, but I would surely investigate what causes the drift in the first place. For example, what OTA are you using to shoot and guide with? From the focal lenght, it seems you're maybe using an SCT? If so, mind that those are particularly prone to flex problem, most of which are not addressable. I know because I've hit the same wall in the past. I have a C8 which I cannot use to take photos for this reason. I only use it for star parties now...

Fabio

[georg.viehoever]

Hi,

I am not so sure if drift is the problem. The stars have a nice round form, as it should be. Only when you enlage the RAW picture, some of the dimmer exemplars (not the bright ones) seem to have small tails. And those tails point into different directions, which should not be the case with drift.

Strange...

Georg

[papaf]

Really? The drift is apparent to me already in the zoomed view. Also, it seems the trail goes from up-left to low-right, just like the cropped version noise goes... Am I missing something?

[georg.viehoever]

Fabio, we are both looking at http://www.flickr.com/photos/daveh56/4087398456/sizes/l/, aren't we?
Georg

[Nocturnal]

Measuring drift/differential flexure is easy. Dave should have seen this page float by a few times on the DSS Y! group:

http://deepskystacker.wikispaces.com/Measuring+differential+flexure

It's hard to fix a problem without knowing the cause. This technique can be used to exclude DF from the list of possibilities.

[papaf]

Yes, Georg, I was looking at that one too. It's not big, but some star trails are visible from top left to bottom right. It doesn't seem the exact same angle as the zoomed view, but some rotation could have taken place during stacking, maybe?

Fabio