help.md

Thanks for looking into this

Copy pasting from the original gdal2tiles.py script:

https://github.com/OSGeo/gdal/blob/84e622a2def7d0e7d37f349c95adcf1339e9a3ab/swig/python/gdal-utils/osgeo_utils/gdal2tiles.py#L2837-L2840

rx = int((ulx - geotran[0]) / geotran[1] + 0.001)
ry = int((uly - geotran[3]) / geotran[5] + 0.001)
rxsize = max(1, int((lrx - ulx) / geotran[1] + 0.5))
rysize = max(1, int((lry - uly) / geotran[5] + 0.5))

I think this calculates offsets to read from, and the length in x,y to read.

• My doubt is why is 0.001 and 0.5 added to resolution.
• Shouldn't they calculate incorrect offsets and sizes for low pixel resolution images, and especially in geodetic coordinates.

Further edits:

1. I think I misunderstood basic BODMAS rules, this is a stupid question, and they don't bring significant changes to the calculated values.
2. Still, I'd love to know some edge cases for which they're there.