See my answers below:
- Mainly visualization for early design stages?
Yes. At the beginning of a project during concept/early schematic design a civil engineer usually has not been brought on board yet, or if they have, they are awaiting a set of drawings/models from the architect to start their work, so the onus is on the designer to kick it off.
- Do you need proper cut & fill information?
Always. Ideally to be able to quantify different cuts and fills independently to make changes. There is always a limit to how much grading can be performed on a site and having this information early on before coordinating with a civil engineer (even rough ballpark figures) are extremely valuable. Having the information early on prevents the risk of producing a design that is prohibitive and i never want to pay an expensive consultant to tell us something impossible - i prefer to already know its possible and have the engineer sort out the details.
- What is your typical input data? A raster DEM, elevation points & breaklines from surveyors, some other format?
Typical input data comes from a civil engineer or surveyor. Which means topolines that need to be converted to a mesh. The TIN surfaces are proxygraphics in ACAD so they typically don’t import well to Rhino. My current workflow is importing the DWG into Revit, create toposurface from the 3D contour lines on a select layer, and then, export to Rhino. I am aware this is possible with Rhino as well.
In some cases, prior to a survey, or in conjunction with a survey, the architect will also have a pointcloud. Not all surveyors use scanning workflows. Most don’t. But at my last firm we would recieve as-built models that came with pointclouds in advance of a survey. I’ve used these pointclouds to establish topography but it is time consuming because you have to use external software to eliminate buildings, trees, etc. from the cloud data before attempting to mesh it. Ultimately, however, the survey is a legal document for a set of drawings and even if the pointcloud is higher in detail than the survey, the survey takes precedence. That is of course, if the two filetypes are from different parties.
- How will you use the topo surface downstream? Does it go to Revit/Archicad or other BIM software, or stays in Rhino?
The toposurface is critical throughout the whole project. Some firms work exclusively in Revit, other firms work in Rhino, ACAD, etc. in the early phase for design approval, and then seitch to Revit during the documentation phase. Depending on the design firms workflow native rhino geometry may be used as a linked file within the Revit file during later phases, so its not like it has to be re-built. Whatever the case, a firm will do those prelim calcs in one software, either Rhino or Revit. Revit provides calcs for grading but the manipulation of graded surfaces is clunky and very limited. For things that are flat, it usually works well, but when you’re dealing with 3D curvatures it becomes extremely unforgiving. It is currently on their trello roadmap to improve, who knows when…
- Do you care about automated construction (essentially a 2.5D TIN model without overhangs)
Not sure what you mean.
- What is the ‘killer feature’ you’d like to have for terrain modelling and can’t find in any existing software?
My thought is that terrain design is a volumetric study, not a study through brep (boundary representation modeling). I’m actually pretty stunned how well Dendro fulfilled my needs. Architects need to be swift and adjusting contour lines manually and dealing with surfaces that need to be re-built and re-computed is a pain, and my best guess is that voxelization combined brep is the best path here. I want to be able to push and pull on surfaces and solids and get immediate visual and numerical feedback on small design moves that add up to the larger story. So far dendro has done this but I haven’t got to the part of calculating the cuts and fills yet. I’m guessing there may be some difficulty there.
As for killer features, I think its all about being to combine feature sets into unique workflows. LANDS Design plugin has some interesting terrain tools for Rhino. FOREGround plugin for Revit has some interesting features as well. I would love to see voxel features like Dendro or even LiveBoolean in Zbrush (quick and immediate feedback theough exploration) be integrated with those Brep types (slower, but capable of producing better drawings, consistent feature analysis, documentation), but the mathematics is completely different in the approach. I wonder how parallel processing using CUDA cores on Nvidia GPUs can also speed things up.