May 27 2026 APP 2.0.0-beta45 has been released !
Fully Multi-Threaded LNC, many improvements for the registration engine, platform upgrade, and further tuning of internal memory consumption and memory release back to OS.
Apr 14 2026: Google Pay, Apple Pay & WeChat Pay added as payment options
Update on the 2.0.0 release & the full manual
We are getting close to the 2.0.0 stable release and the full manual. The manual will soon become available on the website and also in PDF format. Both versions will be identical and once released, will start to follow the APP release cycle and thus will stay up-to-date to the latest APP version.
Once 2.0.0 is released, the price for APP will increase. Owner's license holders will not need to pay an upgrade fee to use 2.0.0, neither do Renter's license holders.
Hi,
What's the meaning of kappa low and high for winsorized averaging? My understanding of winsorized averaging would be that one provides a percentile (maybe upper and lower) at which to clip and replace the clipped values with these percentiles.
I assume that kappa is usually the standard deviation of a list of values in APP?
Thanks in advance!
Björn
Hi Björn @barnold84,
Kappa is the number of times the scale/dispersion factor is applied for rejection. Scale/dispersion is usually the standard deviaton. APP uses a weighted standard deviation since you integrate with weights.
Kappa low, is the kappa that is applied for rejection on all pixels in the pixel stack that are below the central value. The central value is usually the average or median of the pixel stack.
Kappa high then is the kappa that is applied for rejection on all pixels in the pixel stack that are higher than that central value.
The reason to discriminate between the two sides relative to the central value is is that the pixel value distribution usually is not symmetric and that most rejection usually needs to be done on pixels with too high values, like hot pixels for instance.
Winsorization is replacing all pixels values that are higher and lower than a fixed threshold by their threshold to get a winsorized central value and scale/dispersion. The rejection kappa's are then applied to the pixel stack using the winsorized central value and scale/dispersion value.
Let me know if this answers your question or if you have any others,
Mabula
