Dithering

[nightster]

I used dithering when using DSLR to help battle the noise. Should I dither now with my new low noise Sony chipped CCD? And if so, how aggressively should I move the mount between subs?

[steppenwolf]

All sensors have hot pixels (even Sony's !) so dithering is a very powerful tool for eliminating them from your images if you use the correct stacking method. I use a maximum deviation of 2 pixels.

[nightster]

Thank you, I shall continue to dither.

And I must say, my wife ordered the 3rd edition for me as a Christmas gift and I absolutely love the book.  Thank you for writing it.

[steppenwolf]

And I must say, my wife ordered the 3rd edition for me as a Christmas gift and I absolutely love the book.  Thank you for writing it.

My pleasure, Jeff

[pyrasanth]

I tried a dithering experiment. I took 3 frames of RGB with a 3 pixel dither. When they were not calibrated but stacked as a colour picture I could see the hot pixels in each frame clearly showing how the dither had moved the frame. The hot pixels showed as colored dots. When several of the images were stacked with CCD stacker the hot pixels were completely removed- dithering really works so is worth doing! 

[ollypenrice]

There is always noise so, at the cost of a couple of pixels on the edge of the image, do it!

Olly

[julian489289]

Hi.  I have watched Tony HALLAS DSLR Astrophotography on You tube.  He mentions dithering but rather than 2 or 3 pixels he refers to 2 or 3 star diameters to dither.  Is this right and how do you change how much the program dithers?

[julian489289]

I have just found in APT, the dithering parameters.  It gives dithering distance 1 thru' 5.  I take it these are pixels?

Also it lists dithering stability which is set at 0.70, settle time 15, timeout 180 & dither on # images 1.  Are these setting rights?

[Yoddha]

When PHD is used the Dithering Distance is a scale because PHD don't have information about the guide and can't calculate the steps in pixels. Just experiment which value gives good amount of moving. If 5 is still small then in PHD from the Brain you can increase the step with "Dither Scale"

[julian489289]

Thanks Ivo.

However, I went out last night to have a session but unfortunately PHD2 kept crashing so I could not try these settings out.

I did manage though to identify the issue and it is to do with the compatibility of APT & PHD2 rather than the hardware.  Rather than take time up on this thread, I have emailed you direct Ivo to see if you can assist.  Hope that is OK.

Regards

[Yoddha]

Mails are always welcome

Did you send it or will make it later?

[Yoddha]

Got it

[Droogie 2001]

Just finished watching the excellent video on using Backyard-EOS where they discussed the dithering options within the interface.

https://www.youtube.com/watch?v=z3gkw8bx7Aw

I have switched this on as well as enabling the Server element in PHD but now have to wait for an opportunity to test it.

May question is if the light images are deliberately moved during each shot does this not effect the final image for stacking? I guess the answer is no other wise people wouldn't use it but I thought the idea was to get a stable stack of light images and not ones all of alignment.
It almost like introducing field rotation into you lights?

Thanks

[glowingturnip]

no - the stacking software will first align the frames, so it nudges each individual light frame so that the stars all line up with one master reference frame, then it stacks them all together, so effectively it 'undoes' the dithering.

The advantage is that any hot pixels etc will be in a slightly different place in each light, so after a sigma-clipping stacking routine, they will disappear.  It can also help with some cameras that exhibit banding effects which also get cancelled out.  Dithering when used with drizzling can also give a better level of resolution for under-sampled images.

Basically you should always dither if you can.

It doesn't introduce field rotation, but it does introduce a translation between each light.

[Droogie 2001]

Thanks for the clear explanation, that makes sense to me.
I will so what results I can get from dithering.

[nightster]

Would it be possible to dither the images after the fact? If the images are simply nudged left right up or down, could a script be built to take a stack of un-dithered images and move them, randomly a few pixels before the images are stacked to allow the Sigma Clipping to detect them as hot and remove them?  Is there a difference if this is done in software vs physically moving the hardware?

[nightster]

Nevermind, thought about it more. The hot pixel doesn't move, only the stars move. Stacking realigns and the orphaned pixel has nothing to line up to, is considered an outlier and removed. 

[wxsatuser]

Thanks for the clear explanation, that makes sense to me.

I will so what results I can get from dithering.

If you have a modern Canon DSLR you should not have to worry about hot, dead or stuck pixels.

As long as you do a forced sensor clean before the imaging run the hot/dead/stuck pixel list is updated in the RAW file.

The RAW converter should ignore all these pixels on conversion.