BPP error: Cannot integrate less than three frames

[cplloyd42]

Hi,

I'm still trialing PI and have successfully used the BatchPreProcessing script on a number of Canon CR2 image sets, until one of three I generated last night. The above error occurred on the middle set of the three, the other two generating no problems.

I am using master bias, dark and flat files generated by PI. The set of images comprises 12 CR2 files and the error occurs in the light frame registration section. Here is the start of this processing step:
"*********************************************************************
* Begin registration of light frames
*********************************************************************

StarAlignment: Global context
Loading reference image:
Y:/Astrophotography/PI_work/PI_output/calibrated/light/debayered/IMG_11183_c_d.fit
Reading FITS: 32-bit floating point, 3 channel(s), 5202x3465 pixels: done
Computing extreme sample values: done
Structure map: done
Detecting stars: done
648 stars found.
* Limiting to 500 brightest stars.

Registering target image 1 of 12
Loading target file:
Y:/Astrophotography/PI_work/PI_output/calibrated/light/debayered/IMG_11181_c_d.fit
Reading FITS: 32-bit floating point, 3 channel(s), 5202x3465 pixels: done
Computing extreme sample values: done
Structure map: done
Detecting stars: done
1346 stars found.
* Limiting to 500 brightest stars.
Matching stars: done
79 putative star pair matches.
Performing RANSAC: done
*** Error: RANSAC: Unable to find a valid set of star pair matches.

*** Applying error policy: Continue on error."

I have tried 2 different Registration Reference Images, and have set the Clamping threshold = 0.3 (the default), 0.2 and 0.45, all with no visible improvement. The other Image Registration parameters have been left at the default values (Maxm nbr stars = 500, Noise reduction = disabled, Use triangle similarity = ON).

Would someone be able to point me to a solution or work around, please?
Thanks,
Paul.

[Phil Leigh]

This is normally caused by a problem with the chosen reference frame having non-round stars or other problems...

[georg.viehoever]

Its difficult to say without seeing the images. Maybe you upload your reference frame and one of the images that does not registers. Then someone will certainly have a look.
Georg

[cplloyd42]

Thanks, Phil and Georg, for getting back to me on this. I'll upload several images if you would let me know how you do this - I'm obviously not looking in the right place. Also, is there a protocol for naming such uploads? I was thinking of zipping them into a single file with either my forum id, or this thread number, or both as the file name.

It's possibly of interest that I had photographed this field about a week ago, but with half the exposure time, and these images stacked without a problem. This might indicate that you are on the right track, Phil, with your suggestion of non-round stars. However, as I mentioned in my problem description, I photographed 2 other fields during the same session, one before and one after this problem field, both with the same exposure time, and these stacked without a problem. Curious...!

Thanks in advance,
Paul.

[georg.viehoever]

Thanks, Phil and Georg, for getting back to me on this. I'll upload several images if you would let me know how you do this - I'm obviously not looking in the right place....

Use your favorite file service, for instance dropbox, and publish the links and some explanations here.
Georg

[cplloyd42]

Sorry, Georg. I have neither. Is there another option?

Paul.

[cplloyd42]

Georg,

I think I should suspend this posting.

I have had a good, magnified view of these images, viz-a-viz the images of the two successful stacking/registering runs and all are showing elongated stars. The two successful runs were of fields fairly well south (and I'm in the southern hemisphere) while the unsuccessful series was of a field near the celestial equator. I suspect that the star streaking of the two successful fields was somewhat less than the near-equatorial one, and presumably within the tolerance of the registration process. Thus, I strongly suspect that Phil Leigh's suggested cause is true.

From my perspective, these streaked stars are not acceptable. So, if I want to expose for this length of time (2 mins), I'm going to need to use autoguiding. This I will look into before I raise this problem again.

Thanks for your help so far, and my apologies for troubling you. I would like, though, to know how to upload files to a PI server, or elsewhere, for future reference.

[Phil Leigh]

You go to www.Dropbox.com and set up an account (it's free) then upload your files to your dropbox folder... then you publish the links to the dropbox folder here.

Glad to hear you found the problem...

[Raymond]

Here is maybe another hint that could help you (in the case of elongated stars). Try to change the default settings under:
StarAlignment - StarDetection
> Peak Response and/or
> Maximum Distortion
Hope, this helps
Raymond

[cplloyd42]

Thanks, Phil, for your advice. I've never used a Dropbox, and would prefer not to, but it seems this forum doesn't provide an ftp site for file uploading - I'm surprised about that. It's like some on-line account registration systems requiring a mandatory mobile number - forgetting that not everyone has a mobile, and may not want to have one, or has one but doesn't want it to be used in this way.

Paul.

[cplloyd42]

Raymond,

Thanks for your suggestion.

Do the Process>ImageRegistration>StarAlignment parameters get used in the BPP? I've tried a number of different settings (e.g. Maximum distortion = 0.2 {was 0.5} and Peak response = 1.00 {was 0.80}) with no discernible improvement. I would have thought that, as I was able to stack the two more southerly (and hence, less streaked) image sets with just the default settings, I wouldn't have needed to change these parameters all that much to be successful.

Just in case the above parameters are not used in the BPP, I also tried to use the Process>ImageRegistration>StarAlignment tool to do the stacking by loading the *_c_d.fit files and opening one of these files as the reference image. However, all I could get it to do was process the reference file, so I suspect that I'm doing something wrong. Unfortunately, there's no help file available as yet for this tool.

Paul.

[Andres.Pozo]

PixInsight does have a file server. You can read this message for instructions about how to get an account.

Thanks, Phil, for your advice. I've never used a Dropbox, and would prefer not to, but it seems this forum doesn't provide an ftp site for file uploading - I'm surprised about that. It's like some on-line account registration systems requiring a mandatory mobile number - forgetting that not everyone has a mobile, and may not want to have one, or has one but doesn't want it to be used in this way.

Paul.

[cplloyd42]

Andres,

Thanks for that. I'm glad that PI has some facility like this - it's what I'm used to as all other forums I'm a member of have this. If I read the comments in that link correctly, though, I believe that this service might be for just full licence holders. I'm pretty sure I'll be buying in (I've about 3 weeks left to go with my evaluation licence) so will wait until then to make an account.

Paul.