Nasty issue

[Piotr]

I found an issue. I called it Nasty issue. It involves nasty cutting of Lambert Conformal Conic in poles. In the configuration -45 and 60 standard parallels:

[daan]

I put limits on the poleward extent for Lambert under extreme conditions. Acceleration of distortion reaches such an extreme toward the pole for the parameterization you chose that Geocart is not able to perform useful analytics there and cannot correctly image raster elements—in fact, at that degree of magnification, quantization error in the source image causes rendering artifacts. Geocart also can’t properly display graticule elements because the rapidly accelerating distortion exceeds the tolerances Geocart has set for poor behavior in a projection in order to protect itself against floating-point precision losses, especially to ensure that it always finds an outer boundary.

For any typical use of Lambert, this is not a problem. For the upcoming release, I went ahead and adjusted the clipping formulation to bring it closer to Geocart’s absolute tolerances, but it will still clip for the parameters you use, about half as much.

— daan