1.8RC7 Segmentation violation

[astropixel]

Hi.

1.8RC7 ImageIntegration PI crash/Segmentation Violation.

PI produced the error shown in the attached screenshot. This was the only time that PI did not crash, during 4 attempts to integrate 31 bias frames - usually at 7% II froze then crashed.

I noticed this first with BatchPreprocessing and tried the ImageIntegration tool verify the problem.

The bias frames are 16bit tiff, converted from FujiFilm RAF files. SilkyPix was used to convert the files. I have used SilkyPix before, to convert RAF files to be processed in PI, without any problems.

Another image set of cr2 files integrated without problems when I last used PI about 3 - 4 weeks ago. Not sure what has changed since.

Regards,

Rowland.

[papaf]

A bug was recently discovered about this. Windsorized sigma integration can't cope with too many absolute black (ie 0 value) pixels in the image. This should be corrected in the next RC.
This however shows you have "bugged" files to begin with. Even biases should be made of pixels with some value, albeit low. So something must be wrong with the files.

Make this test, which was at the time suggested to me by Juan: open one of those biases and apply this PixelMath expression:

$T == 0

and see how many pure black pixels you have.

[astropixel]

Thanks papaf. Results using Pixel Math show fewer dark pixels. Will have to track down the prolem.