> Many standard shading algorithms exist for creating shaded reliefs, some can easily be found in applications like ArcGIS and QGIS but they are simply not expressive enough (see our paper for examples). Creating these reliefs is partially an art form rather than a simple illumination problem, so trying to encode the 'human-ness' into step-by-step algorithms is simply too difficult. This is where ML enters the picture.

> an art form

Which is why the linked Daniel Huffman's technique employed the convincing results of Blender.

The biggest thing that I'm seeing that Singh has over Huffman is the adjustable detail-preserving terrain smoothing. Detail-preserving blur algorithms don't generalize well. Application-specific NN solutions tend to excel at this problem, by having a model of what details are important vs incidental.

I would like to see a combination of the two, with the terrain pre-processed before being used in Blender.

This is a similar problem as automatic LOD computation for 3D game assets though, and that has been done for a long time without ML (but maybe ML may actually help there too, no idea).

I still find the 'human touch' argument a bit unconvincing for this type of problem.

ML is absolutely being used in automatic LoD computation for far better results, as well as in other geometry transformations (with Unreal Engine introducing the generic "ML Deformer", for example).

> Application-specific NN solutions tend to excel at this problem, by having a model of what details are important vs incidental

I'd say this is a context in which "the problem of Transparency" becomes evident: we would like to "see why", to come to learn a perfected deterministic algorithm from the implicit model that the NN developed.

The paper does exactly that: download a DEM height map and do shading from that. What they argue is that you can customize the amount of detail of the terrain to make it look hand-made (since an artist shading by hand wouldn't be able to interpret small variations and flat areas of the contour plot as well).

I was wondering about that too. The direction of a high quality shading is not uniform: https://i.imgur.com/Y8hIWAD.png (taken from Fig. 3 in the paper)

I don't think it is that ML is needed. I think the point is to use a very sophisticated/nuanced rendering system for topographical maps. Could be Blender is too big of dependency for what they have planned. This seemed like an above average use for ML of an HN submission to me.