Dithering

[Spookyer (Brett)]

Hi all, I gather one should dither a bit to improve results though I don't really understand why. I can't find how much. What are the factors that determine what the dithering parameters should be for a given setup?

thanks
Brett

[RobF (Rob)]

Just imagine you've got a couple of whopping dust spots on your camera lens, then you take a shot. The image is compromised where the dust/gunk is. If you move your camera a little in any direction then snap again, then continue moving and snapping........then later when you combine the images the stacking algorithm will discard the "damaged" areas that should now be in a mirnority of frames.

The downside is you lose some of your frame size when you crop the final overlapping montage of multiple images.

On the other hand, if you kept tracking perfectly at the one spot all night, snapping away, you'd never "see" well what was behind the dust/gunk.

Hot and cold pixels can also be similarly dithered out, not just large scale dust. Hope that helps.

Then, there is drizzle.....

Rob

[RickS (Rick)]

Brett,

<deleted description of dithering since Rob already covered this>

You need to choose a dither amount so that most of the time you don't get the same pixels coinciding. I like to pick an amount around 3 or 4 pixels on the main sensor so there's at least a 9 pixel square area for the dither and 8 out of 9 times a random dither will be in a different position to the previous sub. It's not really quite this simple because dither distance can be a fraction of a pixel but it's a good enough approximation. Note that I said 3 or 4 pixels on the main sensor. Some programs measure the dither using guide sensor pixels so you may need to adjust to suit the difference in image scale.

You don't want the dither amount to be too big because:

• it will take longer for the guiding to settle after each dither
• you'll lose some pixels around the edge of your image

The easy way to see if your dither is working is to register a bunch of subs and then blink compare them. You should see the hot pixels wandering around the image while the stars and DSOs stay still.

Cheers,
Rick.

[Spookyer (Brett)]

Thanks for that guys. Very helpful.

[Spookyer (Brett)]

Looks like in SGP you set the dither in the guider setup. The manual says:

"Check the “Dither” option and then choose the dither amount. We use “PHD speak” for these terms, but in general, use “Very High Dither” or “Extreme Dither” for shorter focal lengths and the smaller dither options as you increase focal length. Finally you need to define the settling options. "

So for my scope with focal length at 740mm with reducer or 925mm without I should maybe go for "Very High Dither" ?

Is .2 reasonable for a settling option?

Brett

[rcheshire (Rowland)]

Berry and Burnell recommend 12 pixels for DSLRs. 10 - 15 is OK. I guess with longer focal lengths keeping the target within the sensor boundary becomes a consideration.

The reasons for dithering are manifold. A distinct advantage is improved subpixel sampling as well as hiding less than optimal image calibration/reduction artifacts. Typical of poorly temperature matched darks.

[RickS (Rick)]

Brett,

I have no idea what the SGP settings mean so I can't recommend anything there.

You'll see the effect of the dithering in PI as the results from StarAlignment are displayed in the process console. It prints the delta X and delta Y (and rotation angle) required to register each sub to the reference image. So long as I'm seeing a decent amount of variation there and numbers that are usually at least a couple of pixels I know the results will be fine.

Cheers,
Rick.