Why Dither?

[avandonk]

Dithering is moving the position between exposures randomly by several pixels.
I use Guidemaster to do this and for my 300mm lens setup I have the setting set to 30 pixels as the guide scope has a focal length of 1250mm.

Here is an animated gif of crops of the same stack of twenty exposures without dithering, with and without dark correction.

2MB
http://d1355990.i49.quadrahosting.com.au/2011_09/ND.gif

Note that even the dark corrected stack just simply stacks the 'holes'. The fridge my Canon 5DH is in maintains a constant temperature within 0.2 degrees C. These images were taken with a fridge temperature of 8C so the hot pixels would be more obvious.

All these images were unifomly processed to show any noise.

Here is the dithered stack of twenty frames animation. The image with the D in the top left corner is the dark corrected stack. The other had no dark correction.


2MB
http://d1355990.i49.quadrahosting.co...11_09/Dith.gif

Notice how even the dithered stack that is not dark corrected has less residual hot pixel noise and is better than the undithered dark corrected stack.




Bert

[ballaratdragons (Ken)]

Excellent example Bert.

Thanks for going to the effort to show the benefits of Dithering. I've never seen it displayed so well before

The difference is amazing.

What would you suggest as a minimum movement? 3 pixels? 5 pixels? or should it be more like 10+ pixels?

[leon]

Well that certainly is different Bert, at least one can see the advantages with your process.

Leon

[LucasB (Lucas)]

I use BackyardEOS for my capture and have often wondered if I should set up dithering in PHD. Now I am convinced that it's a must by this great demo!
Thanks for posting.
Lucas

[jjjnettie (Jeanette)]

Cheers for this. Excellent.

[sadia]

Excellent example Bert. I see you mention your guide scope focal length and how much you dither by, so is there a relationship between focal length and how much one should dither?

[avandonk]

I have the number of pixels to dither set to 30 pixels in Guidemaster. The guidescope has a FL of 1250mm and the image train 300mm. So this is about 7 pixels at the image sensor. The bigger the dither range the better the smoothing.

Here is the full frame of the test area in Grus. I chose it because it was so nondescript only to find a planetary in the field. I wondered at first why a donut would be bright!

3MB. FoV is 7 X 4.6 degrees.
http://d1355990.i49.quadrahosting.co..._09/dithff.jpg

The planetary is IC 5148.

There are better images here
http://www.iceinspace.com.au/forum/s...ghlight=ic5148
and here
http://www.iceinspace.com.au/forum/s...ad.php?t=65658


I have been using dithering for some years but never really did a proper test. It is very useful to pause sometimes and work out if you could do it better.

You lose some time as you have to delay between images for the guiding to settle down to the new position. I use about 20s.

Bert

[coldlegs (Stephen)]

Dithering clearly does a great job of getting rid of the hot/dark pixels but I thought it was also good for improving resolution. Am I wrong?
Stephen

[avandonk]

Originally Posted by coldlegs

Dithering clearly does a great job of getting rid of the hot/dark pixels but I thought it was also good for improving resolution. Am I wrong?
Stephen

Here is an animated gif of crops from dithered and not dithered 20 frame stacks. 1MB

http://d1355990.i49.quadrahosting.co...11_09/D_ND.gif

and one with more contrast etc 2.5MB

http://d1355990.i49.quadrahosting.co...1_09/NB_D_.gif

The dithered image is 'tighter' especially in the very dim stars. Dithering does improve one shot colour as the Bayer matrix effectively lowers colour resolution to less than half of the luminance.

Bert

[avandonk]

The randomness of the dithers can be seen in this screenshot of DSS in the dx and dy columns.


Bert