I am fairly new to the world of GIS and also using QGIS. I normally use a greyscale height map to create a 3D terrain in external 3D software. I have a location where I only have access to a RGB raster LiDAR map, and I wondered if there's a way to convert it to a greyscale height map that I can use to displace my surface in my 3D software.

See attached images of the raster I have and the sort of greyscale image I would like to achieve.

** Update ** I found a better source for the LiDAR info I needed, so this is now not an issue for me.

my RGB raster

desired greyscale output

  • Sorry, not happening. This is an image of elevations dropped on top of a shaded relief. This does not look like heights but, rather a ground filtered point cloud turned into a DEM, not DSM or heights. Heights are whats left after the ground is normalized out and a DSM is everything (non-filtered elevations). Whereas, a DEM is the ground elevations only (with above ground measurements filtered out). That aside, this image is for visualization only, not analysis. In other words, it a picture not data. Follow recommendations of requesting the derived lidar data products, not raw point cloud. Commented Jul 10 at 19:52
  • Hi, thanks for your input. Trying now to track down DTM (which was what this one was labelled as) Commented Jul 10 at 19:54
  • 1
    A DTM is a digital terrain model, which aims to be a model of the ground surface, and a DSM is a digital surface model, which shows things like the tops of trees and buildings. If this is UK Environment Agency data then the derived surfaces should be available...
    – Spacedman
    Commented Jul 10 at 21:15
  • Thanks - I normally use DEFRA LiDAR but this location is Wales and I'm just figuring out how to access that data as it's a different platform - there's not as much coverage as England but I have now found enough at (2m accuracy) to cover my site - in the elevation greyscale I was looking for. Commented Jul 10 at 22:25

1 Answer 1


I'd class this as very hard to do. It's been coloured by height (there's a brown blob at the top, a different green in the valley) and then shaded by a hill-shading algorithm which considers an artificial light source which appears to be coming from the NW. Deconvoluting these two processes is tricky and can't be done "locally" since a dark pixel has no colour information yet could be a shaded pixel at the top of a mountain or the bottom of a valley.

A possible approach might be to feed a learning algorithm with a bunch of known DEMs and the resulting hillshade, such that it learns how to create a DEM from a hillshade... But this might require a lot of training and be very error-prone anyway. I doubt its been done already (but I'll search in a bit and update) and it might be easier to get the source data.

  • OK, thanks for your input Commented Jul 10 at 19:53

Not the answer you're looking for? Browse other questions tagged or ask your own question.