How important is the quality % in Autostakkert! ?

[StuartT]

Tonight I was blessed with a crystal clear night and a waxing gibbous moon. So good was the seeing that my videos (5000 frames) were pretty good from first frame to last. So it makes me wonder what AS!3 is actually using to grade quality? This was not the usual situation where the few frames at the start were sharp and the rest were mush. Nevertheless, AS!3 graded only a few of them above 90% quality.

This made me wonder... what if I just ignore the quality analysis and stack them all?

The results were interesting. From the same video, I tried stacking

• a) only frames rated with quality >90%;
• b) all frames.

b is less noisy than a (as I would expect given the much higher number of frames stacked). But there doesn't seem to be any advantage to having such a high quality selection in a.

Which makes me wonder if there is a point to the "Analysis" stage in AS!3

Discuss.

 

 

[Altocumulus]

Pulls up armchair and gets out the popcorn for the debate.....

[neil phillips]

10 hours ago, StuartT said:

Tonight I was blessed with a crystal clear night and a waxing gibbous moon. So good was the seeing that my videos (5000 frames) were pretty good from first frame to last. So it makes me wonder what AS!3 is actually using to grade quality? This was not the usual situation where the few frames at the start were sharp and the rest were mush. Nevertheless, AS!3 graded only a few of them above 90% quality.

This made me wonder... what if I just ignore the quality analysis and stack them all?

The results were interesting. From the same video, I tried stacking

• a) only frames rated with quality >90%;
• b) all frames.

b is less noisy than a (as I would expect given the much higher number of frames stacked). But there doesn't seem to be any advantage to having such a high quality selection in a.

Which makes me wonder if there is a point to the "Analysis" stage in AS!3

Discuss.

 

 

There is a big point. And it is. Not all nights have superb seeing. infact most do not. So quality estimation, is in fact very very important. And the higher the focal length the more important it becomes. Your focal length seems quite low here. And coupled with the superb seeing. its no surprise what your finding. But don't go thinking this will apply to all situations. As it will not. Quality estimation can make a absolutely huge impact on a final result. But its estimation is far from perfect. That would be the only gripe. 

[Altocumulus]

I suppose that even when all the frames look good, overall there may be some minor differences the algorithm can identify. Don't forget it's all on the premise of 'lucky imaging'. As for the the quality graph, eveything's probably relative.

Have you looked at the individual frames, you might find one that's even better than your stacked final

[neil phillips]

Btw Stuart you are making good improvements with your capturing and processing. Both images look fine. the top ever so slightly less soft. Sharper appearance. Looking on a 30" 4k monltor

There is no advantage overstacking  Only stack what you need to control noise. No benefit will come with more than is needed, other than controlling noise with more sharpening. Which can only go so far. Because even if a image isn't noisy, too much sharpening will still make the image look unnatural. Stacking more than is needed especially poorer quality lower back on the graph. Will just produce softer  slightly blurry images

Edited March 15, 2022 by neil phillips