Dither Every X Images (minor bug report)

[lodrigj]

BYN 2.0.10 on Win10 Pro with Nikon D5300

Settings > Dither > Dither every x Images

Set to 40 images

Apply

Save

Go back to the setting and it stays at 10. Doesn't seem to want to accept any digits higher than 10.

I tried closing BYN and restarting, still goes back to 10

It just won't stay set.

Jerry

 

 

[admin]

This looks like the classic windows update messing up privileges.  To confirm this, change any other settings, then save, and see if they survive an application restart.  If they do not that is the issue.

The easiest way to fix this is to re-install BYN in a _separate_ folder.

Keep me posted please.

Regards,

[astroman133]

I don't think that this is a privilege issue. Even though the control allows entry of values up to 100 I can save values less than 10, but not values greater than 10 with BYE 3.1.18 RC1. The fact that values less than 10 are saved and that other settings are persisted as expected seems to indicate that permissions are fine.

[s3igell]

Not to disparage the apparent bug that you found.

What are you attempting to do where you would wish to only Dither after such a large number of Frames??

Much of the purpose of Dithering - allowing Stacking Algorithms to Average-Out the Shot Noise - is Defeated if a large-enough number of those Frames retain the Same Noise Patterns.  I would hazard a guess that any "Noise" seen in 40 Frames will be retained as "Signal".

Or, are you performing some plan of Massive numbers of Short Exposures, where 40 Frames at same Dither Position are still going to be only 1-2% of the Frame Count??

[admin]

1 hour ago, astroman133 said:

I don't think that this is a privilege issue. Even though the control allows entry of values up to 100 I can save values less than 10, but not values greater than 10 with BYE 3.1.18 RC1. The fact that values less than 10 are saved and that other settings are persisted as expected seems to indicate that permissions are fine.

Ah, you are correct.  I just checked in code and the maximum was 10.  I just changed that to 999 and should be in the next release. 

Like s3igell though... dithering every 10+ images is like not dithering at all IMO... but everyone's process is different and Jerry has a use case for 10+ so I made the change from 10 to 999.  This was a no risk change.

Regards,

[lodrigj]

Hi Guys,

Thanks for the fix Guylain. The bug really wasn't that big of a deal to me, I'm really kind of just messing around right now trying to debug a new setup.

I was shooting 250 x 30 seconds over three nights in my light polluted driveway at f/1.4 at ISO 100.  I was shooting 30 seconds because 10 second exposures would have caused too many dithers.

And even 30 seconds had the histogram farther to the right than I normally aim for. I probably could have shot 10 second exposures and easily been sky-noise limited. But then I would have had 750 frames, and since I couldn't dither every 40 frames like I wanted to, I had to push the individual sub exposure to have less dithers. I would probably only had 250 x 10s frames (instead of 750) if I dithered every frame throwing away 2/3 of the potential signal I could have gathered.

So I had dithering on at 10 frames, or every 300 seconds. But because I was shooting with a wide angle, I needed a really big dither, and it took a long time.  And I know about the random dither without guiding, but if I don't guide, I can see star elongation and trailing even at 30 seconds because of the crazy resolution of the new 75mm f/1.4 astrograph that I just got. And when I get my new full-frame camera (yes, this astrograph will cover full frame at f/1.4) with pixels half the size that I have now so I can get closer to correctly sampling the crazy resolution this lens can deliver, I'm going to see this tracking problem even more. So I have to guide. And I don't want to waste half the dark sky time trying to dither every 30 seconds.

It would be great though to have someone run the equation as to whether you get a higher s/n with more frames actually gathering signal, or pattern-noise reduction with more dithers but a GIANT loss of signal, because the dither overhead is so high and you then have only half the lights.

The true s/n increase comes from stacking more lights, not from dithering. Right? You might get a little increase in s/n from kappa sigma rejection with dithering, but I think the s/n increase is much bigger from more signal, maybe by an order of magnitude? 

I'd actually like someone with more math skills than me to actually run this equation and see if my gut feeling is correct. :-)  Because the light pollution shot noise would be the biggest noise term, and this is more complicated than my math-challenged brain can successfully calculate. I can instantly calculate a 20% tip in my head when presented the bill at a restaurant (Simply double the total  and move the decimal point one place to the left). But that's about it.

Now, if there was a way to stop guiding, and then do a really quick, big, random dither, and then go back to autoguiding, that would be perfect for my particularly peculiar setup, but I don't know how to do this with PHD and BYE.  But I've been out of the game for more than a year with health issues, so I'm just kind of catching back up seriously. 

So I apologize if there's a way to do this that I'm not aware of.

Jerry