• During the June and October 2004 commissioning runs, extended periods with very good seeing (0.5" and less in Ks) allowed us to judge the instrinsic image quality of LIRIS. Under such conditions LIRIS significantly undersamples the PSF with its 0.25"/pixel scale.
  We found that the PSF can have a triangular appearance during excellent seeing conditions. This is illustrated by the following three images, that were obtained in J, H and the Ks filter and have 0.4" image seeing. The images are stacks of many dozen individual exposures. The appearance of the PSF is virtually identical in the raw images. It is currently not clear whether this appearance is caused by the telescope or by some optical element in the camera. When the seeing gets worse than about 0.6" this triangular shape is washed out.
  
  

J	H	Ks
Click on the images to see them 3x enlarged.

Field distortion

• Optical field distortion is very significant in the corners of the field of view covered by LIRIS. A full third order polynomial correction is necessary in order to correct for it. The image below illustrates the distortion field (click for full resolution):
  
  Shown is the difference between a single exposure and the same exposure, resampled with a proper two-dimensional, third order polynomial (including cross terms). Thus, the black dots indicate stellar positions before the resampling, and the white dots show the stellar positions after correction.
  The pattern is not radially symmetric.
   
• Correction of the field distortion is mandatory for extended targets and many other applications. This is due to the relatively large dither patterns that are usually used in near-IR observations for proper sky background subtraction and flat-fielding.
  If the distortion is not corrected, very elongated stars and multiple images appear especially in the image corners, where the distortion terms grow very rapidly. This holds even for compact dither patterns that can be applied on largely empty fields. We illustrate this in the example below, where once a second-order polynomial (left panel) and then a third-order polynomial (right panel) was used for the correction:
   

  Second-order distortion polynomial	Third-order distortion polynomial
  Click on the images in order to see them full-sized. Correction with the second-order polynomial is clearly insufficient. Compare the lower right corners of these two images in the magnification below.
  The image seeing in the right image is 0.8".

  
   
• We do not state any distortion coefficients here since these are subject to permanent change (rotation of the instrument, filter change, zenith distance etc). Instead, by the time LIRIS goes to the WHT again in period 2004b (October), we will have a fully integrated reduction pipeline (THELI, Erben & Schirmer 2004, in prep.) available that determines the required distortion correction from the data itself. The examples shown were created with a pre-release version of THELI, which takes advantage of those wide dither patterns.