confirmed WBPP 2.8.2: ImageRegistration is rejecting good images without justification

I ran WBPP on this dataset with build 1633 and experienced far fewer of the PCL exceptions, but there were still 3. The frame that failed to calibrate in Juan's test was corrupted by Automatic Cosmetic Correction. It calibrated just fine here with ACC turned off. No question, tracking failed on that subframe, yet registration succeeded.

Screenshot 2025-01-03 at 22.12.07.png


Screenshot 2025-01-03 at 22.15.37.jpg


Code:
[2025-01-03 23:23:49] * Loading target file: /Users/pane/NOT BACKED UP/vdb/debayered/Light_BIN-1_9576x6388_EXPOSURE-120.00s_FILTER-NoFilter_CFA/vdb_15_Light_120_secs_047_c_d_R.xisf
[2025-01-03 23:23:49] Loading image: w=9576 h=6388 n=1 Gray Float32
[2025-01-03 23:23:49] 5 image properties
[2025-01-03 23:23:49] 95 FITS keyword(s) extracted
[2025-01-03 23:23:49] Noise reduction:
[2025-01-03 23:23:49] Structure map:
[2025-01-03 23:23:49] Local maxima map:
[2025-01-03 23:23:49] Detecting stars: done
[2025-01-03 23:23:49] Fitting 4135 stars:
[2025-01-03 23:23:49] 610 PSF fits.
[2025-01-03 23:23:49] Minimum star size: 12 px
[2025-01-03 23:23:49] * Computing global linear transformation.
[2025-01-03 23:23:49] * Reference image: limiting to 762 brightest stars.
[2025-01-03 23:23:49] Matching stars: done
[2025-01-03 23:23:49] 220 putative star pair matches.
[2025-01-03 23:23:49] Performing RANSAC: done
[2025-01-03 23:23:49] ** RANSAC: Unable to find a valid set of star pair matches.
[2025-01-03 23:23:49] * Previous attempt failed - this is try #2
[2025-01-03 23:23:49] * Reference image: limiting to 250 brightest stars.
[2025-01-03 23:23:49] * Target image: limiting to 250 brightest stars.
[2025-01-03 23:23:49] Matching stars: done
[2025-01-03 23:23:49] 113 putative star pair matches.
[2025-01-03 23:23:49] Performing RANSAC: done
[2025-01-03 23:23:49] 49 star pair matches in 972 RANSAC iterations.
[2025-01-03 23:23:49] * Summary of model properties:
[2025-01-03 23:23:49] Inliers       : 0.434
[2025-01-03 23:23:49] Overlapping   : 0.764
[2025-01-03 23:23:49] Regularity    : 0.765
[2025-01-03 23:23:49] Quality       : 0.712
[2025-01-03 23:23:49] Root mean square error:
[2025-01-03 23:23:49] delta_RMS     :  0.218 px
[2025-01-03 23:23:49] RMS error deviation:
[2025-01-03 23:23:49] sigma_RMS     :  0.108 px
[2025-01-03 23:23:49] Peak errors:
[2025-01-03 23:23:49] delta_x_max   :  0.242 px
[2025-01-03 23:23:49] delta_y_max   :  0.636 px
[2025-01-03 23:23:49] translation   :     22.55 px
[2025-01-03 23:23:49] translation_x :     +8.62 px
[2025-01-03 23:23:49] translation_y :    +20.84 px
[2025-01-03 23:23:49] rotation      :     -0.04 deg
[2025-01-03 23:23:49] scale         : 0.9998
[2025-01-03 23:23:49] scale_x       : 0.9996
[2025-01-03 23:23:49] scale_y       : 0.9998
[2025-01-03 23:23:49] * Projective transformation matrix:
[2025-01-03 23:23:49]      +1.000367     +0.000447     +8.617279
[2025-01-03 23:23:49]      -0.000626     +1.000187    +20.838411
[2025-01-03 23:23:49]      +0.000000     -0.000000     +1.000000
[2025-01-03 23:23:49] * Computing distortion correction model.
[2025-01-03 23:23:49] * Star matching tolerance: 3 px
[2025-01-03 23:23:49] Iteration #1: 504 star pair matches, residual = 0.7704 px
[2025-01-03 23:23:49] Building DDM thin plate splines:
[2025-01-03 23:23:49] X: eps = 0.01 px, n = 504
[2025-01-03 23:23:49] Y: eps = 0.01 px, n = 504
[2025-01-03 23:23:49] Iteration #2: 504 star pair matches, residual = 0.0004 px
[2025-01-03 23:23:49] Building DDM thin plate splines:
[2025-01-03 23:23:49] X: eps = 0.01 px, n = 504
[2025-01-03 23:23:49] Y: eps = 0.01 px, n = 504
[2025-01-03 23:23:49] Iteration #3: 504 star pair matches, residual = 0.0003 px
[2025-01-03 23:23:49] * Distortion correction: Converged after 3 iterations.
[2025-01-03 23:23:49] * Best distortion model with 504 star pair matches and residual = 0.0003 px
[2025-01-03 23:23:49] Generating registered image
[2025-01-03 23:23:49] Building surface interpolation grid: done
[2025-01-03 23:23:49] DDM Thin Plate Spline / Lanczos-3 interpolation, c=0.30:
[2025-01-03 23:23:49] Registration successful.
 
Hi John,

After intensive testing and code analysis, today I have managed to reproduce this problem and fix it in build 1632 / Qt 6.8.1. Tomorrow we'll release an update to the ImageRegistration module that fixes this problem on macOS. Thank you for your patience.
 
Hi John,

Yes, I am sorry; I should have informed you. The update will be a new version 1.9.3 of PixInsight, where we have achieved a stable macOS version based on Qt 6.8.1. However, this new version still requires intensive and extensive testing work that we are carrying out right now. This necessarily implies a delay. I'll contact you (and other users) as soon as we have it ready for the final tests before an official release.
 
Back
Top