Start earlier in Revit? Yes
Do it like your suggesting? No - not precisely

Adjust your expectations for the complexity or specificity of what you model to create your early design documents. For example, why do schematic layouts with 2D lines in drafting views when thin walls can look a lot like lines but they can host doors, windows, rooms, ceiling boundaries... Then you can switch those walls to more accurate wall types later easily and keep all the other related work intact. See my blog post for example images of that sequence.

Try to find ways to provide the logical transition between design stages using aspects of Revit that don't merely imitate what people have done in other software. If you're really comfortable with that and you have to start over to build the model, then why bother to make them change what they are doing. For example, I've run into folks that are quite comfortable with the purposeful transition from other tools to Revit later. They do so because intentionally modelling the early design again in Revit is their second look at the design, a chance to catch things they'd almost certainly miss if they didn't do that.

Keep looking at what your firm does, how they do it and keep trying to find a place for Revit to make that all better without driving everyone nuts in the process. Easy to write...beautiful when it happens.

Good luck!

Edit: When I worked at WATG my boss called our early stage approach to Revit use "The Horizontal Approach". We only use Revit modelling activites for plan centric documentation but include hand sketched and render elevations and such in Revit views, just turning off the model, keeping Levels for example. These scanned images can be resized to very closely match the scale of the view correctly. This way we could keep the artistry and emotions the sketches evoked but have the harsher reality of area number crunching side by side in plans and schedules etc.

It's not drastically different from what you wrote...that help?

Thanks for your response Steve. I have read about Jim Balding's horizontal approach before. The issue we have is that we have NURBs based curved and sloping walls that would need to be projected in plan to show the intent. We would need to take into account some of the 3d aspects of the building. It would work if everything was just extruded vertically, but that is not the case with a lot of our designs.

Also in our schematic drawings we show wall thicknesses and a lot of other information that typically gets drawn in DD due to the complexity of our designs. We do this so that we can start coordination earlier. A lot of this is modeled in Rhino or CATIA early on.

Okay, you work there, know your kinds of projects. You know what Revit can and can't do. You sure don't need my blessing. If you believe your project(s) could benefit from the approach, then be like the Nike ad, Just Do It!

Let us know how it goes.

In the past (2011), and just as an exercise to learn Revit, I modeled Drager House (by architect Franklin D. Israel) while at University. It was my first attempt at modelling something in Revit, and as I like challenges, I thought it would be cool to do it. That project contains lots of tilted walls and complex geometry (although no NURBS). It was not very intuitive to model, but the model had everything you wanted in a 3D environment.

I would not get discouraged by Revit's internal constraints of modeling. Using mass models that can easily host walls/roofs, and adapt to future changes is not a bad idea. I don't know how complex your building shape is (NURBS), but you also have Dynamo in Revit to help with complex geometry.

Can you upload an image of your project modeled in Rhino? This way we would be able to provide you with better advice.

There are images of a small project drawing set in Schematic Design. The project I will be preparing for is much larger than this, but with similar complexity. We do use Mass models in later stages once we develop and rationalize the surfaces, but we find them difficult to use early on because they don't take edits very well throughout the project and you end up recreating things. Its because the NURBs editing tools are not very evolved in Revit.

The problem for us is that due to the complexity of some of the surfaces, we deliver the model to the contractor instead of adding dimensions to the drawings. We find that in making the drawings look how we want them (which the higher ups here are very opinionated about) we end up having to compromise the accuracy of the model, and in fixing the model, the drawings are affected. So I am looking for a way to isolate the two for greater QA/QC control, while still being able to turn on the model and coordinate the two. For example, we may opt to use the model display for the enclosure but draft the interiors until further development, etc.

"Images removed"

Thanks.

Do you work for Frank Gehry by any chance?
I still find that it shouldn't be that difficult to model it in Revit. I would treat the NURBS surfaces as a series of different/independant mass families, which are composed by Lofts and move the vertices to make those non-orthogonal planes. Then I would use curtain walls with tilted mullions for the blades and another curtain wall with solid panels.

You could still use Revit for the interior design, and leave the complex geometry as an imported geometry from Rhino/CATIA into a generic model, and then insert that into your file. You would have to reload or recreate any changes in the original native file, re-import into the family and just see them in your Revit file (not able to change).

Honestly, I think it's worth trying to model inside Revit. You would have more control over the project. Good luck!

I may work at Gehry's office.... lol.

What you described is what we currently do. It's just that they are so used to customizing the look of every line, so most of the automation in Revit goes out the window. This is especially true since they are so used to the AutoCAD / CATIA combo. Its tough trying to establish a new workflow and have people accept change.

Plus then there is the extraction of master geometry, and the ton of scripts we had to do to accomplish that.

Revit now has ability to import Rhino and SAT files and apply category.
even better to use Dynamo so you can "edit family" along w/ naming control (via subcategory?) after import.
best if you can use API.
in short, Revit now can turn 3rd party stuff into native Revit geometry (called FreeForm if i remember correctly?)

It is called Direct Shape. Not that the name really matters to the functionality.