Culling or rating light frames

[billdan]

I was wondering how others grade their subs especially when you have 50+ to go through.
Even though I use Astroart 5 to stack and process, I always register my light frames in DSS first and go by the reported score or star count that DSS gives. It makes it easy to reject the lowest half dozen, but this method is not infallible as it doesn't take into account sat trails or dew.

Is the only way to open each sub, stretch and then reject if bad? This could take forever if you had 200+ subs to sort through.

Regards

Bill

[glend]

DSS will do the rejection for you (based on score) if you just vary the percentage in the processing setup box - the default seems to be 90%. I'd like it to actually reject on star shape as well, or is it doing this already?. It seems to be based on star count and not much else.

[billdan]

I'm not sure what algorithm DSS uses to give a score, I've had subs with a lower star count but still gives a high score compared to others. So I think DSS also takes into account background noise as well (passing clouds).

Bill

[Atmos]

I typically start by running all of the frames through CCDInspector, it can be a quick and easy way to cull ones that have a really soft focus (higher FWHM) or elongation higher than I'd like.

Sometimes only taking out the really bad ones and leaving some of the okay ones isn't a bad thing too. If you have a lot of subs to work with and you get a good rejection going when stacking, it can improve the SNR without effecting star shape.

[strongmanmike]

I use the trusty Eyeball feature outside Astroart to cull subs :thumbsup:...never trust algorithms to do what the trained eye can do better ;)

Mike

[LightningNZ]

I use DSS to rank all the frames, first by FWHM then by star count, and tend to find I can remove the real outliers at the bad end quickly. Then I often flick through each one to see if there's some anomalies missed by the program.

[Shiraz]

I use the "subframe selector" script in PI to reject anything with too high a FWHM. Then visually inspect what is left using "preview images" in Nebulosity - that applies a stretch to the displayed data on the fly, so that you can see what is actually there - it is useful to get rid of bright frames and other occasional problems.

[RickS]

I do a quick visual on each sub which really doesn't take long (used to use CCDStack but mostly I use the PixInsight Blink process these days.) This weeds out any cloud affected subs or guiding catastrophes.

Next, I use the PI SubframeSelector script to analyse FWHM, Eccentricity, SNR, etc. in each sub. I toss out the really bad ones and plug in a formula to calculate a weight for each sub. By default, PI image integration is noise weighted but I like to include FWHM and Eccentricity in the weight calculation. This trades off some SNR for improved detail.

Cheers,
Rick.

[ericwbenson]

And I use the Stack command in Maxim, add the folder, analyze command, then sort by FWHM, elongation or brightness. You can quickly step through the stack using the arrow keys to view a medium stretched image and hit the space bar to manually deselect/select a sub.
You could use the blink command instead, but you have to load all the images into memory. The stack command operates off disk files, so really big stacks are never a problem. With an SSD and file caching load times are really negligible.

So there you have it:
CCDInspector
Nebulosity
PI
Maxim
DSS: the only free tool

Best,
EB

[billdan]

Thank's everyone for your wisdom, it seems we all doing similar strategies

i.e use software to get rid of the really bad ones and then follow up by a visual inspection of the rest.

Cheers
Bill

[Atmos]

It is an expensive slippery slope!