Dithering seems unidirectional and affects histogram

[Alan_]

I have been using SharpCap for a while and it is working well. I have started to use dithering but it seems to be moving the image in "positive" directions in RA and DEC. This means more and more seems to get lost in the frame.

1. Is this the case or have I just been unlucky?
2. In live stacking the borders where dithering crops frames affects the histogram and adds extra peaks, can a ROI be set for the histogram?
3. What is the best way to reset the image centre if dithering is walking the frame so the framing of the subject becomes undesirable?

See image below where dithering is every 5 frames. For the M31 image: left and bottom are getting lost due to dither, right and top show no such borders and always have an image overlap. This suggest dithering is always "positive".

dither.png (401.67 KiB) Viewed 3716 times

Thanks,
Alan

[admin]

Hi Alan,

I see what you mean about the loss in the bottom/left of the image. Since you are using PHD2, almost all of the fine configuration of what dithering does is in the PHD2 application. SharpCap tells it when to dither and how far the maximum movement should be, but nothing about the direction. I'd therefore check the PHD2 dithering settings to see if anything is amiss, also look at the PHD2 log to see what it is recording about each dither operation in terms of direction/size.

SharpCap now has a recenter option (right hand side of the guiding page), but that is not designed to be used when guiding with PHD (or an MGEN guider) because the guiding should make it unnecessary. SharpCap can be set up to do its own dithering using the mount movement (with no guiding active), in which case the recenter option becomes available.

One test would be to turn off dithering in SharpCap and just let PHD2 guide - make sure that you are not seeing the same drift just with guiding (not quite sure how, but the test would either prove it is related to dithering or disprove it).

cheers,

Robin

[Alan_]

Hi Robin,

Thanks for the rapid response and hints.

I think I have figured out that the dither jumps are quite big and that is part of the problem. I set dither to 20 pixels as I thought that was for the main imaging camera. But I think Sharpcap probably does not tranlate this to arc-secs for the command to PHD2 and has no knowledge of the PHD2 settings? So my guess it is "guide camera" pixels. Due to the scale difference (focal length of 120mm guide and 400mm main camera) the 20 pixels dither at the guide camera would give 66 pixels in the main camera image. I have reduced the max. dither size to 5 pixels as that seems reasonable.

I'll try this and watch the PHD2 movements next time I get a chance. Dithering worked as it should and eliminated walking noise.

During live stack the histogram includes the dither borders which tend to be dark, so auto stretch does not work as well. Not a big issue as I can move the guides to the right place. I suspect there are borders at the top and right it is just that that would require an image display area bigger than the sensor pixel x,y size. So top and right dither zones get cropped autouatically.

So I understand what is happening now, thanks! I am very pleased with SharpCap by the way!

Regards,
Alan

[admin]

Hi,

yes, you're quite right that when dithering is in terms of pixels, it is all in terms of pixels on the guide camera - SharpCap has no idea about the guid camera PHD is using, and PHD has no idea about the main camera being used by SharpCap or whatever other application is capturing. 5 pixels sounds like a better starting point, you may even get away with 2 or 3 pixels, as that will still be enough to move the main camera 6 to 10 pixels and should deal with the walking noise.

Fixing up the live stack histogram is something on the TODO list - I guess it's not that hard really, just trim a border off the image before measuring the histogram. It's probably enough to take 5% off each side as a starting point.

cheers,

Robin

[Alan_]

OK, thanks.

I will reduce to a dither to 3 as that will give an average dither box of ~18x18 pixels rather than 120x120 pizels that I set originally! So should no longer lead to significant cropping.

Not a big deal on the histogram, it is just a nice button to hit every now and then to see how things are going. I restack outside of Sharpcap later but it is a great help to see the stack content and evolving detail so I can assess the exposure and required integration time.

Alan