Live stacking and dithering

[Chris_h]

I have mine set to 10 in SC, which equals to 30 pixels in the telescope. So during imaging, I end up with a "dither border" around the image. As expected.

BUT, I have noticed that the longer I image, the more these borders start creeping into the image.

When you start off, does SC anchor a center point, so it knows where to randomly move those 30 pixels around? Or does it dither off the last dithered image? If it dithers of a dithered image (of a dithered, of a dithered...), then that would explain why the edges move in.

Now if you have 6000 pixels to play with, it is less of a problem. But with smaller sensors, you will soon lose valuable real estate.

I ask, because I never had this problem with either SGP or NINA.

Thanks!
- Chris

[admin]

Hi Chris,

dithering is triggered by SharpCap, but the details are handled by PHD2. I think that the default in PHD2 is for the dither to be truly random, so you end up with a 'drunkards walk' effect, meaning that although there is no net drift in the very long term, there will be movements over time in all directions and the longer you go on, the more chance of a bigger movement in one direction or the other.

I wonder if this is just something you do not see in NINA etc because when you stack in other applications it often trims the borders automatically?

Worth checking out the PHD2 settings - it looks like there is a 'spiral' option for dithering as well as random. Never tried it though...

cheers,

Robin

[Chris_h]

Thank you, Robin.

I found this on the CN forum:

"The next random position is started from the last lock position as opposed to starting from a center position each time"

"For spiral dithering, PHD2 resets the lock-point in order to create an expanding pattern of rectangular movements. This involves keeping track of the last dither vector but not previous lock-points"

Spiral is probably worse, as dither points would be grouped more tightly together?

Would "Recenter" help with this, like if I set the drift limit to say 100 pixels? And would lives tack pause during this action?

Thanks!
- Chris

[admin]

Hi Chris,

in theory, recenter would limit the amount of drift, but I don't know if it will work.

Recenter will happen pause the stacking while it happens, but I don't think it will stop and restart the guiding, as I only envisaged it being used when you are *not* using a guiding application. In fact, I think you might actually be stopped from selecting both dithering and recentering at the same time.

Of course the obvious way to reduce the problem is to reduce the size of the dithering movements - if you reduce the dither step by a factor of 3 then it should reduce the drift by a factor of 3 (and the nature of the drunkards walk means that although it would still grow with time, a reduction by a factor of 3 should give you nine times as long before it gets back to the previous size). It's probably worth trying that to see if the smaller dither size leads to noticeable artifacts from hot pixels or not.

cheers,

Robin

[Chris_h]

Hi Robin,

Yeah, I will probably just have to reduce the dither and how often I do them. That, and restart the stack now and then.

Thank you again

- Chris