This is an update on the current state of creating 3d models using Gemini AI for creating pics and then using Bill Smiths/Tiling/ZoeDepth page on google Collab to create a depth map from that image.  Short Story:  There are improvements in the latest modules that produce depth maps.  Lots more detail that can be used for the initial displacement.  I use two stages of displacement.  The first stage of displacement is with the depth map so you can get a better less distorted 3d effect and then the second round of displacment is with the original pic for the details.  It is possible to get more depth detail by manipulating the amount of Depth Map displacement, but the side view gets more and more wonky.  Displacement is done in Blender.