[Closed] Astro Pixel Processor 1.063  


Quasar Admin
Joined: 2 years  ago
Posts: 1797
August 30, 2018 22:00  

Astro Pixel Processor 1.063

  • IMPROVED, Calibration, the calibration engine can now create and process data calibration with 32bits masters. These masters will be of datatype 32bits normalized floats in the range of [0,1]. If this is used in data calibration, the calibrated data will be of the same datatype as well. In most cases, calibration in 32bits is not needed however. Testing shows no difference in noise and SNR values in the integrated results.

32bits Master GUI option

  • IMPROVED, Star Analysis, the minimum star size setting is an important setting if you have lots of stars in your images (>2500) or if you have a lot of stars that have a Full Width at Half Maximum larger than 3 pixels. In those cases, it will help a lot in registration if you increase the minimum star size area. Internally, the star analysis module now better scales for images with different star FWHM sizes. The tooltip for this parameter is updated as well.

StarAnalysis minimumStarSize

Example, showing the benefit in registration of complicated widefield data : registration of only 2 widefield images of a Canon 6D with a Samyang 14mm F/2.8 objective showing a final field of view of 180 degrees. The individual images are just full of stars! If you set the minimum star size at 6 or lower, registration is not robust in this case, thus is due to both the high star count and optical distortion present in the Samyang objective. But if we now increase the minimum star size, to a value of 10-20, registration becomes very robust for this complicated data. Even the optical distortion margin does not need to be increased to help registration allthough this is an extreme example. Since this field of view is larger than 120 degrees, we need to use the calibrated projective registration model and project the data with the equirectangular projection. (Rectilinear only works up until a field of view of 120 degrees ).

StarAnalysis MinimumStarSize helps Registration

  • IMPROVED, Bad Pixel Map creation, the algorithms to detect bad pixels and especially bad columns are improved, which was needed for sensors with a lot of bad columns. Below is an extreme example of an exposure with a Red Filter of the Helix Nebula, shot with iTelescope T32 using a FLI CCD camera with a lot of cold/defect columns on the sensor. Shown first is a Masterdark of 24x600s, used to create the Bad Pixel Map with bad column detection enabled with a kappa value of 1.5, which is shown next:

MasterDark FLI CCD 24x600s
BPM FLI CCD Bad Columns

Next are shown the single Red filter exposure of 180 seconds, and the calibrated exposure with only the Bad Pixel Map and the MasterDark to illustrate the effectiveness of the Bad Pixel Map and the correction of the data on the bad columns, of which there are a lot ! in this extreme example:

Helix Nebula RedFilter FLI CCD 180s
Helix Nebula RedFilter FLI CCD 180s calibrated with BPM


  • UPGRADED, Fits Frame Loader, the fits frame loader now supports 32bits floats and 64bits double images in Bayer CFA format.
  • FIXED, FITS data calibration MaximDL captured images, MaximDL has a FITS header tag to indicate if the image data in the FITS file is flipped in both X & Y axes (or rotated 180 degrees) due to an equatorial mount pier flip in an image session. 

HDU1 - FLIPSTAT= 'Flip/Mirror'
HDU1 - SWMODIFY= 'MaxIm DL Version 5.24 140126 0KJR5' / Name of software that modified
HDU1 - the image
HDU1 - HISTORY Edit Flip
HDU1 - HISTORY Edit Mirror

I don't know what the actual argument from MaximDL is to do this, but to make data calibration work properly, the data needs to be flipped back, so the right pixels are calibrated using the calibration data which is not (and never should be) flipped. So If APP encounters this header tag in a MaximDL fits file, it will flip the data back to it's original state ensuring correct data calibration.

  • IMPROVED, Dark Frame Scaling, the dark frame scale factor calculation has been improved. It still creates 3 scale factors for the entire ADU range of the MasterDark. However, on some sensors, especially CMOS sensors, it was in some cases a bit unstable. The new calculation is more robust and will give more reliable scale factors, which is possible since several scale factors are calculated. Clear outliers are removed and on the resulting factors we can fit the scale factors giving a dark scale curve over the entire ADU range. Be aware, dark scaling is only possible if both a MasterDark/MasterDarkFlat and a MasterBias is supplied for calibration.
  • FIXED, Calibration, Adaptive Data Pedestal, could give an error if it was enabled and no MasterDark was loaded for calibrating the light frame. This function requires a MasterDark in calibration. This is now fixed.
  • FIXED, FITS metadata interpreter, the fits metadata interpreter was too flexible for some header tags, which sometimes caused issues when loading FITS files created by external applications. This is now properly fixed for all FITS header tags that APP needs for processing. One of the main issues reported was that that shot data was reported in the instrument column of the frame list panel, and this created problems in the creation of Bad Pixel Maps.
    • APP 1.062
      APP1062 error

    • APP 1.063
      APP1063 correct

  • FIXED, PreviewFilter, in case of loading an image which is clearly clipping on the left side of the histogram ( the black point) the preview filter could fail to show a proper preview when using any of the DDP presets. The data was strongly overstretched in that case. See the following 2 images with the problem and fix:

PreviewFilterBug APP1062
PreviewFilterFix BlackClipping APP1063

  • FIXED, Fits Metadata, in case of zero values for gain, APP reported the Egain value is present. It will now correctly show the zero gain value is the Fits tag is present in the header.
  • FIXED, LNC, in some cases a LNC calculation will fail and output a null pointer exception. The main cause of this is usually severe bblack clipping of the data. The LNC analysis wasn't able to deal with this properly, giving a null on analysis and that caused the null pointer. The analysis is adjusted and fixed so it shouldn't happen again. If there are a lot of pixels with 0 values (black clipping), analysis will only focus on the non-zero value in analysis.
  • FIXED, LN Rejection filter, if a Local Normalization Outlier Rejection filter was chosen without performing LNC in multi-channel and/or multi-session mode and you created the individual channels and/or sessions and the ALL integration, the LNC RejectionWorker failed on the ALL integration and reported a null pointer exception. The root cause is found and fixed and the LNC Rejection Worker will now work.

LNCRejectionWorker Bug

  • FIXED, Flat-field Calibration with Bayer CFA patterns other than RGGB, there was a bug in normalization of flat frames for Bayer CFA sensors. This bug manifested itself if the Bayer CFA pattern was not the default pattern of RGGB. In most cases the resulting masterflat was still okay to be used, but if you created a MasterFlat out of flats with quite different histograms/exposure times then the problem was really severe, no good flat-field calibration was possible.

Example of a MasterFlat from a Canon EOS 5D Mark II with a GBRG CFA pattern, first 2 images is with the bug, notice that the resulting MasterFlat has an odd histogram and that, when zoomed in, the pixel structure is not smooth, due to having normalized the CFA pixels with an incorrect CFA pattern :

MasterFlat GBRG BUG histogram is BAD
MasterFlat GBRG BUG zoomedIn


Next 2 images is with the bug fixed, notice that the histogram is now smooth and consistent with a single flat frame and, when zoomed in, the MasterFlat's pixel structure is now smooth:

MasterFlat GBRG BUG FIxed histogram is GOOD
MasterFlat GBRG BUG Fixed zoomedIn smooth

  • FIXED, GUI Calibration Details, if something was adjusted in the calibration Masters, Dark Scaling, or the Adaptive Data Pedestal and you tried to directly load the l-calibrated image viewer mode, the new calibration details were NOT yet directly applied still showing the calibration with the old calibration details. This was a synchronzation issue in the User Interface. Now the calibration details are checked and re-created if needed, and the l-calibrated image viewer will now directly apply the new calibration details to the light frame.
  • IMPROVED, MEMORY USAGE, some internal memory parameters have been adjusted to improve processing throughput and reduced memory consumption. APP will now again release memory in between processing steps.
This topic was modified 2 months  ago by Mabula Haverkamp - Admin

Main developer of Astro Pixel Processor and owner of Aries Productions