WBPP Frame Selector (new): how do you decide which frames to reject?

Pleiades Astrophoto PixInsight Acquisition Calibration Processing Stacking Deep Sky Galaxy How-to Nebula
5 replies•164 views
Pablo Gafo avatar
Pablo Gafo avatar

Hi everyone,
I’m new to the AstroBin forum, but I’ve been doing astrophotography for quite a long time. Glad to be here 🙂

Recently, WBPP in PixInsight has implemented the Frame Selector, and I’d like to learn how to use it properly. Until now, I’ve always done a fairly basic manual selection (FWHM, elongated stars, etc.), but with all the new metrics available I’m not sure which criteria are really critical and which are not.

My main question is:
Which parameters do you personally use to decide which frames to reject and which ones to keep?

For example:

  • What FWHM, Eccentricity, or Star Count values do you consider acceptable?

  • Do you always reject frames with poor SNRWeight, even if they’re not terrible?

  • How do you deal with frames affected by thin clouds or variable transparency?

  • Do you use the Frame Selector only to remove the worst subs, or also for weighting?

My main targets are deep-sky objects (nebulae and galaxies), and I’m trying to find a good balance between final image quality and not throwing away too many usable frames.

Any practical explanations or real-world workflows would be greatly appreciated.
Thanks in advance!

Well Written Engaging
Brian Puhl avatar
Brian Puhl avatar

The few times I’ve used SFS for culling data, the results were extremely negligible. I will sometimes do seperate stacks by culling the lower 50% of the data based on FWHM. My resulting FWHM went down by barely 0.1 arc second. WBPP already does a very good job weighting exposures for the final stack.

Blink is a far more useful tool for rejection.

- Clouds = reject. Any clouds, whatsoever, indicated by that glow around a star is gone.

- Oval stars due to bad guiding/lost stars = reject

-Abhorrently fat stars (bad seeing) = reject

Everything else, I let WBPP figure out.

Helpful
Pablo Gafo avatar
Pablo Gafo avatar

Thanks, Brian — that’s very helpful.

I agree that Blink seems much more effective for real rejection, and that WBPP weighting already does most of the heavy lifting. Out of curiosity, do you ever use Frame Selector thresholds at all, or purely as a diagnostic?

Well Written Respectful Engaging
Brian Puhl avatar
Brian Puhl avatar

Pablo Gafo · Feb 4, 2026, 10:33 PM

Thanks, Brian — that’s very helpful.

I agree that Blink seems much more effective for real rejection, and that WBPP weighting already does most of the heavy lifting. Out of curiosity, do you ever use Frame Selector thresholds at all, or purely as a diagnostic?

I’ve used SFS purely for FWHM rejection. The benefit is very little, but there is some to be had. The thing is, this day and age with BlurX in play, what little you gain seems pointless by the time BlurX does its thing. You end up with the same result.

If I have wide swings in FWHM’s it would probably hold more benefit. I could see it being highly beneficial for much larger scopes than mine. The thing is, those large changes in FWHM are quite apparent when I blink, so in my workflow, I catch them there long before.

Eccentricity is another value to monitor, however in my case, I’m often undersampled. My sharpest data will yield the highest eccentricity simply because my pixel FWHM’s will be around 1.5-1.8 which I don’t believe is enough pixels to properly sample. On soft nights, poor seeing, my eccenticity values will be ‘very good’ by measurement, simply because there are more pixels to measure. Any oval stars caused by guiding issues I will catch during the blink process, and if there are only a couple, I’ll just leave them. Pixel rejection will do it’s work if theres only a few, and they’ll get rejected in the final stack.

My takeaway from it all is SNR is more important than sharpness in most cases. This is a debatable topic, so take my advice with a grain of salt. I rely on blink for culling my data. If I want to measure the quality of my data from night to night, I use SFS. I’ve been trying to track my nightly seeing conditions over the year.

Helpful Insightful Respectful Engaging
Michael Sleeman avatar
Michael Sleeman avatar

I usually (at least for Galaxy) make one master with mostly all subs (except for the worst FWHM outliers) and one sub (only in Luminance; lets call it “sharpMaster”) where I make a quite harsh cut-off (depending on the total integration time; but about 10-30% of the best frames).

I then blend the “sharpMaster” with the normal Lum-Master together to raise the resolution of the inner areas / cores of galaxies, where noise does not matter that much.

Yes, we have BlurX - but high-res subs still make a very big difference which you can definitely see and notice.

Helpful Insightful Engaging
alpheratz06 avatar
alpheratz06 avatar

As already mentioned , I use a blend of blink and WBPP.

When I have a large amount of subs, i always blink the batch first to prune any images that fit in the following criteria :

  • density shift (high clouds, etc)

  • tracking issue (elongated stars)

This , to my opinion, save calibration and quality evaluation time.

For the rest , I trust Wbpp using fwhm criteria. The last version offers a quality feedback before stacking, thus allowing to eliminate manually others poor subs.

Clear skies

Helpful Concise