Stacho, Thanks for the clarification on the 3d modeling. That has helped a lot with my understanding of how the output is generated. Even in its current form, it is an exceptionally simple way of generating an initial model for AutoCAD. I look forward to any improvements to the 3d modelling in the future. The suggestion by Bruce to introduce additional dimension points close to the ends of the scraps has reduced the lack of alignment to a few cm, which is prefectly acceptable for our purposes. Estimating distance underwater, at depth in poor visibility is not very precise! Also, adding the points for the same single station to both scraps has improved things. I will try the single scrap method, but actually really like the ability to use different scale plans of the dry cave sections in the same .th2. Do you have proposals about how the proposed 3d algorithm will work in the future? It seems, to me at least, that the ability to introduce cross-sectional shape on LRUD data would be even more useful than the use of elevation scraps for passage height. I understand this is a subject that may get very complicated, very quickly. I'm a big fan on AutoCAD and am therefore somewhat biased. I have a few suggestions for the dxf export that may be worth considering for future releases : 1. Export of a closed 3d polyline of the assumed (rectangular) cave cross-section at each station on another layer. This x,y,z data is already generated for the vertices of the surfaces and would simplify over-drawing of known cross-sections and 'working inside the cave' in autocad. 2. Export of a 3d ellipse or the current flat floor/arch ceiling 3d polyline at each cross-section on a separate layer. Again, this x,y,z data is already calculated. The latest autocad versions have a very powerful 'LOFT' command that can seamlessly connect together complicated cross-sections along paths. 3. Maybe the export of cross-sections drawn in Xtherion, exported as correctly scaled and aligned 3d polylines in the dxf would be a first step. The complex triangulation of the surface between two cross-sections could be handled by autocad/3dsmax in the first instance. Excellent software and support. Thank you for your efforts, Steve Clark