bug in smile geometric correction

[Robin Leadbeater]

there appears to be a bug/ "undocumented feature" in the way ISIS applies a smile correction (for example used in place of slant when selecting the ALPY 600 option). It can produce a shift in the dispersion direction at the Y location of the spectrum as here for example with the ALPY 200. (The lamp lines were almost straight and vertical so the radius was large and the required correction small which may be a contributing factor)

ISIS_smile_correction_bug.png (42.72 KiB) Viewed 439 times

It is a problem in circumstances where ISIS is given the position of a lamp line to use as a reference (for example in "filemode" or when using a predefined dispersion equation with a lamp line to correct for any offset). The line in the geometric corrected image is not in the same place as in the raw image so ISIS can select the wrong line and give an incorrect calibration.

The workround is to first process the lamp image with geometric corrections applied, note where the X position of the line moves to and enter that as the X coordinate in the calibration tab instead of the position in the raw image.

Cheers
Robin

[Tom Love]

This possibly explains why I have so often had difficulty getting accurate calibration with the Alpy, whereas I've never really had much of a problem with the Lhires.

I have also often found the smile correction very volatile, sometimes getting quite different values with different lines selected from a given image. I now usually extend the ISIS window vertically as much as possible, seeing more of the image, so you can select the maximum vertical range on a line which seems to help it to achieve a more consistent smile fit.