Unable to find an initial set of putative star pair matches

S

sandconp

Well-known member
I am using WBPP to try and get a stacked master light and I keep getting this error message:

Error: Unable to find an initial set of putative star pair matches.

I have also gotten some RANSAC errors but eliminated them and I have tried increasing the noise reduction from 0 to 5 in Image Registration but still cannot get a stacked master light. I am imaging M51 which does not have a lot of stars to begin with. I used a gain of 200 and I binned the capture in APT 2x2.

I have not had this problem in the past but I did use a different method of capturing my images by placing the ZWO camera on the back of the Telescope versus using a Hyperstar where the camera attaches to the front. There aren't a ton of stars but there are enough to stack so I don't understand the problem.

I am willing to post my source files if somebody would like to help me solve the problem.
 
Last edited:
If you could upload a few frames somewhere and post a link I'll have a look. I have this problem quite often with relatively narrow FoV images in areas with few stars. However, if the image is half-way useable, I usually manage to find some working configuration of StarAlignment. This is one of my more recent "sledgehammer" settings:
1618816195531.png

Noise is not usually my problem - it's usually an image with just half a dozen big fuzzy stars (because that's all there were). One thing I've learned to avoid is winding the sensitivity up (i.e. lower value) in the hope of finding more stars - it often just finds more noise. If you can only see a dozen stars, don't select the sensitivity to find more than 100.
 
Sadly no. You have to run SA stand-alone for this (WBPP simply doesn't expose enough of the SA interface). I have proposed a possible way of doing this (adopting the same approach as CosmeticCorrection), but so far with no take-up.
However, it is not difficult. Run WBPP with registration and integration unselected, then pick up the files from the appropriate wbpp output folder (depending on whether or not you are OSC, and whether or not you select CosmeticCorrection). WBPP still does all the tedious calibration work.
 
I think that I might of found my problem is that I did not have the gain set high enough. I normally use a Hyperstar which is F/2.0 and I was using the camera at the back of the telescope with a focal reducer which was F/6.3 using the same gain setting as my Hyperstar which was 120.

Here is my dropbox link if you want to take a look.

Dropbox - File Deleted - Simplify your life

www.dropbox.com www.dropbox.com

So I could take the files from WBPP and take them right into Star Alignment?
 
Running WBPP with the following registration setting:
1618863787259.png

succeeded in registering and integrating 47 of the images.
Running SA stand-alone on the contents of the "...\WBPP\calibrated\light\debayered\" folder, with the settings I gave in an earlier post, registered all of the images.
I haven't looked too closely, but the WBPP integration (with 47 images) looks cleaner than the integration with all the images, so it would be worth checking the images out with Blink before deciding which ones to integrate.
1618868009279.png
 
Last edited:
I am not sure why mine looks so awful after using the same settings? Attached is the master light after I stretched it and did a background extraction.

In the control panel of WBPP in the calibration settings on the lights, if I don't have optimize master darks checked, nothing works so I check this option and I get the attached results.

What I am doing wrong?
 

Attachments

  • M51S.png
    M51S.png
    141.9 KB · Views: 50
  • M51ES.png
    M51ES.png
    215 KB · Views: 59
It is fairly clear that your subs have not been calibrated (bright amp glow at right edge indicates no dark subtraction) or registered (misaligned stars), so either WBPP has failed completely or you simply integrated the original uncalibrated subs. This could be a simple misunderstanding: WBPP does not modify your original images, it generates new calibrated images, which it places in a set of folders under the top level folder you specify (I usually call it wbpp). You uploaded 15 x 120s Dark, 19 x 0.29s Flat, and 59 x 120s Light. To process flats properly, you should really have either bias frames or Dark Flats (because the Flats will have a bias pedestal that must be removed somehow). Since you have neither, I did the next best thing, and calibrated the flats with the 120s Darks, with dark optimisation (not really recommended, but the best option with this data), so my WBPP control panel looked like this:
1618904999949.png

With the registration settings as in the previous post.
Try that and see if you can get it to work (the final integrated result is placed in the "master" subfolder with filename "masterLight-BINNING_2-FILTER_NoFilter-EXPTIME_120.xisf").
 
Did you use just the first 47 frames or just didn't use the bad ones? My guiding got messed up towards the end. Could a few bad frames mess up WBPP this much?

I will give it another try and let you know the results. You had posted some images earlier that did not look that bad and I am trying to repeat what you did. I realize that my data isn't that good but you were able to produce something decent despite the bad data.
 
The subs are not bad. The 47 image integration was the one done automatically by WBPP (i.e. "masterLight-BINNING_2-FILTER_NoFilter-EXPTIME_120.xisf"), which uses the 47 frames that WBPP successfully registered. The one with all 59 frames was after I registered them all using stand-alone StarAlignment (on the frames calibrated and debayered by WBPP, as in the "...\debayered" folder). With a bit of polish the WBPP version looks something like this:
1618925660710.png
 
On inspection you have a few clouds drifting across, and they upset the tracking towards the end. However, the 50-odd salvaged frames are enough for a quite presentable image.
 
I noticed on the log sensitivity that I had .80 instead of -.80 and that made a big difference, Not sure why I was able to stack 56 of the 58 images maybe because I am running an older or newer version of PixInsight compared to you. Thanks again for your input it's much appreciated !
 
If you blink through the registered images you will find some (near the end) are seriously messed up - guide failures combined with significant cloud obscuration. Your integrated image will be significantly better if you exclude these. It turns out that the 47-49 images that WBPP originally wanted to integrate are about right (that's what my image above is based on).
While you can sometimes "beat StarAlignment into submission", if it is having trouble it is often telling you something, so you should always Blink the subs and see if there is an obvious problem. You will often get a better image integrating the best 10 subs, if the remaining 40 are seriously inferior.
 
Last edited:
Actually, I processed the WBPP integration first, then Blinked the subs to see why WBPP had failed to register them (not really the right workflow - I should have looked first - but I was mainly trying to help you get your workflow right). I might have done even better if I'd rejected a few more frames, but I really wanted to show you that the frames you had were enough to produce a decent image with a bit of processing.
 
I am fairly new to PI and have been having issues with WBPP and Star Alignment failing to register a lot of decent subs. I tried your above suggested settings and they worked great. Thank you, fredvanner.
 
Great support you are sharing @fredvanner - helped me big time on my learning curve into PI.

But now I have found a nut I cannot crack:
I can't get a series of faint images taken at f/10 to process and am close to giving up 11hrs of exposure - so seeking help here

Setup was:
- object M51
- scope 8" Newtonian with Baader field flattener and 2x Barlow (I know this does make it very faint, yet I wanted to try)
- camera Zwo ASI2600mc pro
- filters dualband L-Enhance and Askar D2 as I need to cut out light pollution under a bortle 5-6 sky
- exposures 135x with 300s and 600s exposure times on gain 100 and 200 (gain and exposure did not much improve SNR)
- calibration frames: 20x bias, 20x darks, 20 flats for each exposure + gain + filter combination relevant

WBPP exits every time on not finding enough putative stars (which I could fix with your tipps here) and not being able to register the images.
Either I am not using WBPP/PI not good enough or my images are too faint.

Any ideas what I could try differently - or are my images just too faint?

Attached is the WBPP log + link so sample images https://drive.google.com/drive/folders/1gmgaj__iVaYCCqPURlTRksTQXgPBOdxV?usp=sharing

Arny
 

Attachments

  • ProcessLogger2.txt
    23.1 KB · Views: 30
Last edited:
The lights in your sample look OK. Lots of hot pixels, so they definitely need dark calibration. CosmeticCorrection will also reduce this (the ASI2600MC is my favourite camera, but it does have "warm pixels" which CosmeticCorrection always fixes). Could you post some samples after calibration (e.g. from the WBPP "debayered" folder), to check what is causing the problems for registration. This is a region with few bright stars, so StarAlignment may need some extreme configuration.
 
fredvanner said:
The lights in your sample look OK. Lots of hot pixels, so they definitely need dark calibration. CosmeticCorrection will also reduce this (the ASI2600MC is my favourite camera, but it does have "warm pixels" which CosmeticCorrection always fixes). Could you post some samples after calibration (e.g. from the WBPP "debayered" folder), to check what is causing the problems for registration. This is a region with few bright stars, so StarAlignment may need some extreme configuration.
Click to expand...
I haven't gotten my head around cosmetic correction you suggest here and got lost last I tried - could you kindly steer me to a good starting point?
 
Last edited:
You must log in or register to reply here.
Back
Top