Freeform Modelling (and Revit 8.1)

[stefan]

The wrote:
Main Improvements in Revit Building 8.1:
Considering that it is just a point release, Revit Building 8.1 packs a surprising number of new features. To start with, it enhances the Building Maker functionality for conceptual design mentioned in the previous section by enabling the import of NURBS surfaces from other applications in the form of DWG or SAT files. These can be turned into mass objects in Building Maker, after which you can create roofs, walls, floors, and curtain systems from selected faces (see Figure 5-a). A separate but related ability is the creation of non-vertical walls from massing components, created within Revit Building or imported from other applications (see Figure 5-b). These non-vertical walls behave like the regular walls: you can place doors and windows in them; they can be joined to other walls; they appear properly cut in plan views; they are correctly listed in schedules; and they can be exported along with their information to ODBC and other formats. Considering that Revit Building lacks a good set of freeform modeling tools, these new capabilities make it easier to bring conceptual massing models created in other applications into Revit Building and convert them into building models rather than having to start from scratch.
http://www.aecbytes.com/review/RevitBuilding8.htm

This would interest many people in the discussion around MaxonForm.

[TomWaltz]

Chazz wrote:

Djordje wrote:

Chazz wrote:
Perhaps more importantly, they have an astronomical user base all lined up to migrate.

Are you really sure of this?
Why is the ADT then given free, why didn't the astronomical user base migrate already?

Am I sure? Well, yes.

While I agree, some people are more likely to upgrade to Revit based on existing systems, that does not mean that they are "lined up to migrate." There's a big difference between the two.

A lot of people may show interest or be likely to want something. It's quite another for them to get out their check books and actually buy something. Ask any sales rep.

From the people using AutoCAD that I speak to, many have serious concerns with the licensing policies. The words "forced upgrades" are a recurring theme. I've personally seen three small (1 or 2-person) companies switch to Archicad because Autodesk ticked them off.

[Scott Davis]

Rashid wrote:
Revit currently looks as if run by talented programmers and marketers while ArchiCAD by talented Architects, notorious for being slow to adopt new methods. Ironically they were 15 years ahead of Revit.

The people making the decisions about Revit, and the direction to take it is almost entirely in the hands of Architects. The Revit Building Product Manager is an architect. The "Number 3" guy from the early days of Revit is an Architect, below the two founders, who were Russian mathmaticians from PTC. Most of the development team is architects. Phil Bernstein, the vice president of the Building Solutions Division of Autodesk, which Revit falls under, is an architect, and a professor at the Yale University of Architecture. So, yes, the Revit team has a team of talented programmers...because they must figure out how to make Revit do what all these architects want!

[Anonymous]

With relationships, you can easily check inconcistancies in a model, by just checking placeholders.
With the free modeling you should check in every change everything.
For large projects , it is better to do it all over again, if they changes are extensive.

[Rakela Raul]

relationships,

i can taste this using the 'accessories' ...i which i could select the exterior wall for example and have them be an 'accessory' of the 'slab'.

[Djordje]

Chazz wrote:
Am I sure? Well, yes. My assertion is simply that these users (and I'm talking only about architects and their ilk, not the billions of surveyors, mechanical engineers, etc using Autocad) are going to be much more predisposed to migrate to Revit than something called ArchiCAD (wait a minute, I've heard of that, it's from Russia, right?). That predisposition is all it takes. Having a single-vendor upgrade path is an incalculable advantage. Having the new product (Revit) maintain close ties to your old one -AutoCad- makes it that much more reassuring to the migratee.

Hm ... we can continue this in the Pub if you like as the point is very far off the software discussion, and cannot be called politically correct in the broadest sense of the word. Whether something is Russian, American, Serbian, or Martian is irrelevant - it works for you or not, that is the bottom line!

I still can't see why would someone - architectural practice, of course! - change to a completely different philosophy software just because it carries the same sticker as the previous one.

FYI - John Von Neumann was Hungarian, and they do have some of the best mathematicians and programmers in the world, as do Russians.
Plus, see Scott's post, the main Revit gurus are Russian! Horror! 😉

Bottom line - anyone who changes the software from flatland to intelligent 3D based on the same sticker simply did not do his homework.

[stefan]

Back to the 'relationships' concepts: I think they are very much at the core of what we, as a designer, are doing mentally, but it's not supported in most applications, Revit being the exception.

Everyone is free to try and play with it (30-day full demo or unlimited viewer, as long as you have some Windows running).

The concepts are valid and very architect-friendly: when you "offset" a wall, this is often more then just a simple intermediate operation on some geometry: you offset a wall, since that is a decision in the design. I'm glad that people are working on supporting that into the design process and not merely relying on the mind of the architect.

I'm not advocating that everything at every time is connected, but large parts are.

In fact, when I tried Triforma somewhere around 1995 or so, it also had these concepts: a floorplate is connected to a wall, a wall is connected to a roof. Sure, once you're into design documentation after a building permit, you are not even allowed to change positions of walls and roofs, but during the design process, these are valid operations. And even in Revit, you can disconnect elements or lock others at your own pace. It is a very valid approach which I, personally, would welcome into ArchiCAD... if it was not bolted on as an afterthought. So if the ArchiCAD engine is not capable of handling such things at all, then please leave them out untill the CORE can handle it.

[Petros Ioannou]

I just saw some pretty interesting videos regarding the london city hall
at
http://www.fosterandpartners.com/
it seems they use some generative components from triforma which seems to handle geometric relations between generator lines,arcs etc of the building.
IMHO it is more critical to make use of design rules in testing preliminary geometric form relations than searching for relations between slab-wall etc .

Petros

[Petros Ioannou]

just to get a better picture check on the same site :
Swiss Re Headquarters project
City of London, England 1997-2004
the last qtime video

Petros

[Rob]

Talking about relationship concepts I am not really keen on Revit's approach. However I like the idea of having relationships in the plan window where they would control vertical relationships like the top of the wall to ceiling or the edge beam underside of a slab when the story elevation got changed. Basically vertical offsets that are not visible on the plan giving you a bit confidence about 3D outcome (seeing just 2D dwg) so we could fully focus on the horizontal plane (basically what is a plan dwg about) . Horizontally (meaning XY plane), elements should be free and unrestricted and especially in the design phase where I prefer fiddling and messing around rather than following or introducing and switching on/off some set of rules. I can't imagine anyone who is able to define relationship between wall/slab/beam edges etc in early and medium design stages - it kills creativity and introduce restrictive rules which can't lead you to a decent and fully explored outcome. Well, that's my opinion.

[stefan]

Petros wrote:
I just saw some pretty interesting videos regarding the london city hall
at
http://www.fosterandpartners.com/
it seems they use some generative components from triforma which seems to handle geometric relations between generator lines,arcs etc of the building.
IMHO it is more critical to make use of design rules in testing preliminary geometric form relations than searching for relations between slab-wall etc .

Petros

I can say a few things about generative components:

- it's somewhere between textual programming and visual programming, with some amount of graphical control. You can generate "law curves" which can act as an interface to alter other properties. Sounds technical 😉

- you can put all kinds of relations between elements: graphical, mathematical... You are in control over what you put in.

- i have been to a workshop on using this software, but it was still pre-beta software (yet they are thinking of releasing it with the next commercial Microstation upgrade)

- it uses C# and AFAIK, it is bolted onto the Microstation Parasolid modeling kernel, although I assume that it might be portable to other modeling kernels. In fact, for most of the time, you are working in the Generative Components interface and only use Microstation to display things on screen. You build up a hieararchical network of relationships.

- Should I mention that Microstation doesn't run on the Mac and that you need Microsoft .NET Frameworks installed for generative components?

- A very powerfull feature is programming your own components (through code or through transforming a manual/visual graph into a component). This component could be used to fill in some underlaying structure, such as the faces on a surface (thus "generative" component).
Imagine making a custom parametric window, that gets populated all over the facade of a 20-story building, where each single window get's oriented and edited automatically.
You can argue that you could program a smart GDL object to do that, but I don't see a way to filling a double-curved surface with irregularily shaped modules being filled with some GDL element, which is mainly oriented on A, B and some Height...

- It doesn't stop at generating forms. These forms can be used as a way of extraction construction (production) information, such as folded out drawings to use with a laser cutter, automatically derived from the model and adapted when the model changes.

----------

Now comes the tricky thing: can we have a program that combines the best of these different technologies together?