I didn't make the REALLY cut! The blogetery file is mine. I'll update that site one day, I have more to show.
The real trick is no trick, it's an understanding of TG2 from experience and the forum/docs and an exact idea of what you want to depict. Depends on the requirement. Does the customer (this sounds like you're doing a job for someone) care that your moon looks just like one of the landing sites photographed up close? Will you look up to any hills, or just one angle toward the ground? Getting close to photo-reality, call it 90%, is straightforward (and time-consuming). To achieve an exact match is quite difficult I find. What we find from the hi-res orbital imagery of the LRO satellite is that many areas of the moon don't look like the Apollo or Surveyor or Luna/Lunokhod sites, that the morphology of the moon's surface is rich and variegated and has specific weirdnesses. Making a generic lunar surface with little reference to reality is not too difficult.
The REALLY file is maybe at the 80% level, I can easily look at it and tell it's CG and TG, but still better than NASA's Frassanito surfaces (which excel more in the depiction of craft/astronauts). The stones are not textured enough, there are what I like to call "octave marks", i.e. ridges that result in low octave Perlin noise (ok for Martian dunes), no craters, but overall it's ok, and reasonably convincing esp. the roughness of parts of the scene. I hope this is not too harsh for the artist. The blogetery file, which also falls short, is an attempt to recreate some of the features seen at the Apollo 17 landing site, including the mountain texturing (it's too pronounced with other inaccuracies). It is a current subject of attempts at improvement on my part.
Suggest you avoid using DEM files if you don't have to match an exact set of large features. The grayscale heightmaps need to be worked, they do not yield good up-close views in my experience, maybe somebody else has had better success with those.
Hope this helps a bit,
Bill