resampling dropout

[bolsen@…]

I have a set of 1/2 ft imagery, striped 8 bit rgb geotifs.

I used gdalwarp to resample to 2, 4, 10 & 20 foot geotifs.  I used both bilinear
and nearest neighbor samplings.

The first test set I did ended up tiling them.  All the 20 foot ones had a large
block dropping out in the top left of the image.  About 1/3 of the 10 foot had
the same artifact.

I reran these to output striped geotifs.  The artifacting changed to both
missing scan lines at the bottom and sometimes single or multiple garbage scan
lines appearing near the bottom of the image.

The 2 & 4 foot resamples didn't seem to have issues other than a missing scan
line at the bottom (which is probably acceptable).

One thing of note is that many of the input tiles are not nicely divisible by
the resample, in fact many of them have odd dimensions.