GIS notes

gdalbuildvrt:
- use the raster menu, not the processing toolbox as it doesn't allow custom options!
- use either -vrtnodata 255 or -addalpha
Note that if you use addalpha you can't convert the result to a tiff with jpeg compression

gdalwarp:
- if using JPEG compression, use -ot=Byte
- use tiled output
- generate a world file
- for better JPEG compression, use -co PHOTOMETRIC=YCBCR 

How to use this?:
--config GDAL_TIFF_INTERNAL_MASK YES
Do we have to use it with gdal_translate and this before using gdalwarp?: -b 1 -b 2 -b 3 -mask 4

Converting asc output by 12d, qgis claims it is float64, although all the values are round to the nearest mm, which doesn't match result converted to float geotiff.
use these options to set the nodata value:
-a_nodata -9999.0 for float64
-a_nodata 0 byte or uint16 or uint32 (0.0 accuracy)
float32 doesn't require it!?
float32 file size is only a fraction less than float64, and numbers seem to be the same.  What?!