The inputs are compressed lossy, so a NODATA approach will fail to deliver completely correct results (and also if there is pure black anywhere inside the image).
Eventually we'll need support for a raster or vector mask (see the DETFOO GML sidecars)
The inputs are compressed lossy, so a NODATA approach will fail to deliver completely correct results (and also if there is pure black anywhere inside the image). Eventually we'll need support for a raster or vector mask (see the DETFOO GML sidecars)