How do you revert to previous revision?

[ftherrmann]

I read the post about the star alignment change.  It appears to be "less" robust than before.

The new release also generates a "checksum" error when trying to load preprocess scripts that worked in the last release.

Processing script file: C:/Program Files/PixInsight/src/scripts/BatchPreprocessing/BatchPreprocessing.js
Source code MD5 checksum mismatch: Expected d82158a1331dc17652ccfb50dc71451d, got add306faef287dac3e538b907534b1c2
0.405 s

Here is a link to two zipped subs:  https://www.dropbox.com/s/nwi075imkuq2v69/Blue%20to%20Clear.zip

Try using the preprocess to upsize, and register the blue to the clear.  Doesn't work on my system.  The star align does however work.  These subs were taken on a really BAD night of seeing, however I see no reason why the align function in the preprocess would have a problem.

[ftherrmann]

Here's the results that I get from the preprocess:

Loading reference image:
C:/Users/User/Desktop/Astro SSD/M66_Project.data/M66-30_Clear_1x1_272E700s.00708.fit
Reading FITS: 16-bit integers, 1 channel(s), 3358x2536 pixels: done
Structure map: done
Detecting stars: done
33 stars found.

Registering target image 1 of 1
Loading target file:
C:/Users/User/Desktop/Astro SSD/M66_Project.data/calibrated/light/cosmetized/M66-30_Blue_2x2_272E600s.00700_c_cc.fit
Reading FITS: 32-bit floating point, 1 channel(s), 1679x1268 pixels: done
Computing extreme sample values: done
Structure map: done
Detecting stars: done
51 stars found.
Matching stars: done
20 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.

===== StarAlignment: 0 succeeded, 1 failed, 0 aborted, 0 skipped =====

*********************************************************************
* End registration of light frames
*********************************************************************


Other than the size difference the main difference in the original subs is that the blue is slightly dithered from the clear and there is a slight rotation. ie.  It's practically registered already.

Throw these two files at the star align with the default settings and you get a successful alignment as follows:

StarAlignment: Global context
Loading reference image:
C:/Users/User/Desktop/Astro SSD/M66_Project.data/M66-30_Clear_1x1_272E700s.00708.fit
Reading FITS: 16-bit integers, 1 channel(s), 3358x2536 pixels: done
Structure map: done
Detecting stars: done
33 stars found.

Registering target image 1 of 1
Loading target file:
C:/Users/User/Desktop/Astro SSD/M66_Project.data/Original Subs/M66-30_Blue_2x2_272E600s.00700.fit
Reading FITS: 16-bit integers, 1 channel(s), 1679x1268 pixels: done
Structure map: done
Detecting stars: done
36 stars found.
Matching stars: done
11 putative star pair matches.
Performing RANSAC: done
* Previous attempt failed - this is try #2
useScaleDifferences=true
Matching stars: done
*** 0 star pair matches found - need at least six matched stars.
* Previous attempt failed - this is try #3
useScaleDifferences=false
* Reference image: Limiting to 30 brightest stars.
* Target image: Limiting to 30 brightest stars.
Matching stars: done
13 putative star pair matches.
Performing RANSAC: done
* Previous attempt failed - this is try #4
useScaleDifferences=true
* Reference image: Limiting to 30 brightest stars.
* Target image: Limiting to 30 brightest stars.
Matching stars: done
*** 0 star pair matches found - need at least six matched stars.
* Previous attempt failed - this is try #5
useScaleDifferences=false
* Reference image: Limiting to 15 brightest stars.
* Target image: Limiting to 15 brightest stars.
Matching stars: done
14 putative star pair matches.
Performing RANSAC: done
14 star pair matches in 1 RANSAC iterations.
Summary of model properties:
Inliers     : 1.000
Overlapping : 1.000
Regularity  : 0.778
Quality     : 1.000
Root mean square error:
delta_RMS   :  0.247 px
RMS error deviation:
sigma_RMS   :  0.062 px
Peak errors:
delta_x_max :  0.837 px
delta_y_max :  0.127 px
Transformation matrix:
     +0.4995     +0.0045     +7.3661
     -0.0048     +0.4996    +20.0543
     -0.0000     -0.0000     +1.0000
scale       : 2.002
scale-X     : 2.002
scale-Y     : 2.002
rotation    :     -0.53 deg
dx          :     +7.37 px
dy          :    +20.05 px
Generating registered image
Homographic Projection / Lanczos-3 LUT interpolation, c=0.30: done
Registration successful.
Writing output file: C:/Users/User/Desktop/M66-30_Blue_2x2_272E600s.00700_r.fit
Writing FITS image: 16-bit integers, 1 channel(s), 3358x2536 pixels: done

===== StarAlignment: 1 succeeded, 0 failed, 0 aborted, 0 skipped =====
2.714 s


So,,,  What's the difference?

[pfile]

one question - have you set the output directories to be *inside* a PI project hierarchy? not that it would cause this to fail but it seems a little dangerous. PI probably does not expect to find files inside of a project that it did not put there and i'd hope that it would not decide to delete them (say for instance if you save the project with "Purge Existing Files" checked.)

anyway it's hard to use BPP without darks, bias or flats. your camera seems to be based on an 8300M (which i also have) so i was able to use my bias and darks which of course completely mismatch. if i add a random flat of mine it will undoubtedly mess up the image. i'd have to make a fake flat with a constant illumination profile to proceed. anyhow, without flats i got as far as registering the clear against itself, which succeeded, but it died before registering the blue image.

to be fair the stars are pretty soft in these images and there are very few of them. also it is possible there is an incompatibility between the new SA and the BPP script with respect to the SA settings which causes images such as this to fail.

did you try unchecking "use triangles" in BPP to see if that has an effect? the default in StarAlignment is to use pentagons which may account for some difference (however, when i manually configured SA to use triangles, i was still able to register the two images.)

finally, have you looked at the calibrated images? there is a very slight possibility that the calibration completely messed up the images. i have not seen this except when trying to use calibration masters from MaximDL, but i suppose it's worth checking that the calibrated images look similar to the uncalibrated images.

rob

[ftherrmann]

Don't worry about the calibration.  Just get them registered.

Yes all the files are in the folder that PI created.

Yes I unchecked triangles.   Am I the only one here that has a problem with the fact that a preprocess script created in the previous version won't even load in this new version?   This also isn't the first time this has happened.  It's crazy.  Come out with a minor release and break the most important parts of the program!!!  Insane!

Rob,

The images calibrate just fine in star align.  Pretty simple..

Fred

[ftherrmann]

If someone would just send me a link to the previous installer I'd be really happy!!!!!

Carlos, Juan,,, WHERE ARE YOU?

[pfile]

if you have access to macosx i have a whole bunch of old versions.

p.s. you have to appreciate that no one can reproduce your issue with BPP without the calibration masters.

rob

[pfile]

by the way, this is what i would do in your situation.

use BPP to calibrate the images only.

open StarAlignment, pick a reference frame, load all the calibrated lights, set the output directory to the "registered" directory, and let it rip.

95% of the work that BPP does is to sort out all the flats and darks and apply them properly, so doing the registration step manually seems like a viable option.

rob

[georg.viehoever]

...95% of the work that BPP does is to sort out all the flats and darks and apply them properly, so doing the registration step manually seems like a viable option....

Thats probably the right thing to do.
Georg

[georg.viehoever]

The difference between StarAlignment and StarAlignment inside BPP is the "Maximum Stars" parameters. Its default is "Auto" in StarAlignment, and 500 in BPP. If you change it to 500 in StarAlignment, it will also fail to register. My guess it that BPP will work fine once you change it to 0 there (which appears to be the equivalent of "auto").

Many of the stars in your image are fairly "soft". The star detector in Staralignment does not see many of the stars. You may want to use the "Detect Stars" working mode in StarAlignment to tune it to find more stars. For instance, increasing Peak Response should help. You may also have to balance Noise Scales and Hotpixel Removal as well to avoid the artifacts that are in your image.

Georg

[ftherrmann]

You don't need calibrated subs in order to register the images.  Stars still look like stars if there's a gradient or not.  I managed to register the un calibrated subs here with star align so you should be able to also. 

You should also be able to get the preprocess script to register them since your not having any problems.

I'm using a PC so the iOS stuff won't do me any good.

How do I get in touch with Carlos?

[ftherrmann]

Georg,

Setting the "Maximum" number of stars or zero or Auto in the PP Script fixed the problem..   YEA!!!!!   IT WORKS!!!!

The fix makes absolutely NO SENSE,, but it works!!!!   

If the setting said "Minimum" number of stars for a registration it would make a good deal of sense!!

Thanks for the help!!!

Fred

PS:  No more upgrades without a working installer sitting on at least two drives!

[pfile]

that's interesting, because the log posted here shows only 36 stars detected. a high limit of 500 would seem to be moot in the face of that. it must have some other influence on star detection.

rob

[pfile]

You should also be able to get the preprocess script to register them since your not having any problems.

reading comprehension, though. i am trying to tell you i can't run BPP without calibration masters. BPP does not have the option to skip calibration and just do registration (as far as i know.)

rob

[Carlos Milovic]

BTW, I don't have older releases stored. In fact, most of the time I only have development releases, that are not of public access... so if you can't find a release in the software distribution center, I would not be able to help you.

[ftherrmann]

No problem Carlos!   From now on I keep several copies of the installer before I upgrade...