Because SLIC is randomly seeded, we'll see slight variations in outputs at the pixel level. We should use raster-tester compare (https://github.com/mapbox/raster-tester) to compare, as it allows for down / upsampled comparison + comparison within a threshold.
@dnomadb I also broke the SLIC test in two: one is a test that we don't find nodata and the other is a test that we do find lossy nodata on the edge of a noisy background.
Because
SLIC
is randomly seeded, we'll see slight variations in outputs at the pixel level. We should useraster-tester compare
(https://github.com/mapbox/raster-tester) to compare, as it allows for down / upsampled comparison + comparison within a threshold.