Both methods very clearly describe the same end results, not effecting geometry, by simulating the lighting via normals. Both results yield the same non-edited geometry (the sphere is still smooth as well as the casted shadow)
In TG it's creating displacement, which is seen along the object, and in the shadows.
Which results in weird looking objects it seems, especially with light interaction. At which point does this become displacement, and not simulated bumpiness of light? The height map method says it's used to get the normals, for the lighting. Not using it for actual physical displacement.
You can clearly see in my examples, one has a hard time with TG's lighitng, and one is smooth more correct. I imagine because instead of simulating lighting, it's just displacement with actual object lighting. The other is proper displacement for TG to handle (smooth one). It's a ray traced TGO sphere.