Del foro de Xnormal. Sí, directional map == Vector displacement map. I use it for the DX10 spiked ball example, so por favor take a look todo that example for more información.
You probably get no detail output because you ned todo setup the Cage or because you arent using the matchuv característica.
Some advices:
1. If your High Poly model is just your lowpoly Mesh subdivided and sculpted, use the matchuv característica todo get better results. When you subdivide the lowpoly model, UVS are subdivided todo so, at the final, the lowpoly model and the High Poly model share the UV space (a los known as Dual parameterization). The god thing about this is that you wont ned todo setup a Cage as each point in the High Poly Mesh Will have a matching point on the lowpoly Mesh.
So, in resume, start your High Poly Mesh subdividing the lowpoly Mesh and, then, Sculpt it. In that bien youll get the best results.
The other alternative is todo re-topo but, then, youll ned todo setup the Cage and the vdm(Vector displacement map) wont be as accurate as the matchuv due todo Cage distorsións (like hapens in a cylinder Cage).
2. The output vdm file Will include the XYZ normalized Vector component in the RGB channels. The Alpha Channel Will contain the height scalar. So, youre forced todo use an image format with Alpha Channel support (like the tif, OpenEXR, tga).
3. Caution if you output the vdm with a high-dynamic-range format (like tif FP 32 or OpenEXR). Xnormal always normalizes the height component into a [0,1] range so take that into considerartion. Internally, the value is signed (because the point can be in one face side or the opposite one), then unbiased into the [0,1] range and written into the output file.
Ill probably Will change that for the next reléase allowing todo output HDR values directly.