Related Posts with Thumbnails

« Exploded Axons | Main | Not a creature was stirring, not even a mouse »

December 14, 2009


Feed You can follow this conversation by subscribing to the comment feed for this post.

I'm curious from Autodesk's point view how they approach these sort of lectures, as there under lies not just algorithms for building masses Archtecturally, what about structurally or even MEP?

If algorithms are implmented Architecturally within Revit in the future (which I am all for as I've seen what Inventor can do with complex forms) I think things have to be thought about for the other disiplines such structural and MEP.

I am sure Autodesk are thinking about all the disiplines, but it would be good to hear that! As what happens if tools are implemented where building forms can be changed easily from an Architectural point of view, but when moved further down the chain to Structural and MEP the link breaks?

I'd imagine structural and MEP algorithms would algorithms would be very different beasts in comparison to Architectural algorithms.

Allan, I have to confess, since Erik and I both come from architectural backgrounds, the blog tends to "skew" towards the black-rimmed glasses set ;-) But, I assure you, staff from the MEP and Structure design and dev teams attend these lectures. Also, in general, we are striving towards more generalized, pattern-based UI solutions in Revit. Will this translate to generative algorithms for MEP and structural systems? I don't know... I will put it back to you, what are some potential scenarios where you could see this idea being applied to engineering?

Hi Tom,

I'd point you to the Key Note of AU this year by Carl Bass, there was a structure shown that appeared to be driven by algorithms, can't remember the name of it, but it very briefly appeared on the large screens at the opening address, it was a large complex roof based around the idea of a traditional chinese hat, parameters appeared to be added that drove the geometry of the roof and effectively 'pulled' the roof trusses structure with it. This was the first time I've seen not only the Architectural form being pulled but the actual 'structure' as well.

With the experience I've had, traditionally, roof's (non-planer faces) have always been the most time consuming thing to work out, and amend as they are never liner. If Architectural 'forms' can host structural elements, and when these forms are adjusted they drive the structure would be pretty spectacular. And from what Carl Bass showed at AU, it's already happening. I think it was done in Inventor, which I am starting to look at. Ideally, I would love to see Revit being able to do this in the long run, but of course, I am thinking structurally as I can see huge benefits to this.

Of course, when you have a series of drawings linked to this geometry all annotated and dimensioned up, will Revit be able to handle changes to complex forms without losing the dimensional setting out you've spent time adding to your drawings? When forms twist, rotate or change plane, dimensions are often deleted by Revit as it's not finding the same point you previously dimensioned too.

Of course this is well down the line! Get the hard bit looked at first!! It's an interesting debate for sure. As I say, most of the time things get looked at Architecturally without looking beyond that, so it's good that Autodesk are thinking about all parties involved in delivering a structure.

I think Bentley are using generative components, Inventor uses iCopy, maybe Revit will have something like that in the future! Lot's of possibles.



I have a small worked example of how I could see the conceptually massing tools being used for structural purposes if you'd be interested in seeing it.

Very very basic and it's a very different workflow from how we do things now. but it seems stable and straight forward. There are a few things I would highlight to you. But you'd get the idea.

By saying that algorithmic architecture is the future is like saying that the mobile phone will become the standard of communication in the next 10 years. It seems that autodesk has been living under a rock for the last decade. The fact that autodesk doesn't know how this design computation concept will apply to Revit is very concerning.

The good news is that if Autodesk wants to keep up with industry trends and technology they should look into applications like grasshopper, an algorithmic editor, which is used world widely on many projects. Also generative components from Bentley is a prime example on how to integrate design computation and BIM seamlessly.

We have seen how severals stadiums in the last olympic games have been design trough the use of software packages like Generative Components. The fact that Revit can't be used on none of these types of buildings should be enough reason of concern of Revit lifetime expectancy.

James, Do not misinterpret our intellectual curiosity for ignorance. We are, and have been, well aware of these trends and related products for some time. The point of this blog is to generate two-way discussions with our users. So, if we ask questions, it is to stimulate the conversation.

Also, I think when Autodesk mentions algorithnmic architecture as "the future," they mean as the future for the mainstream or majority of architects. Right now, eventhough these developments have been around for a while, they have been used by a very small minority.
The true usefulness of an idea is tested when it is applicable to a broader audience where it has the potential to make a difference.

The same argument can be made about BIM and 3d modeling in general. Did Revit invent BIM? No! ArchiCAD has been around for 20 years, but in those 20 years what did they accomplish? Did the industry embrace the technology? Without the technology being embraced, what's the impact in how buildings are designed every day? Revit came along and brough BIM to the masses. THAT has value!

Well, that is exactly my point. Revit is catered towards the average Joe, which by all means, isn't bad at all. At the same token, anyone that wants to do something beyond the white picket fence will have to resort to other tools. Revit seems to be the only product at Autodesk that always is the last in line to any piece of technology advancement. When I asked a few ago why Revit doesn't have any up to date massing tools, their answer was that the average Joe doesn't need it. A few years later, when Revit was the only tool that couldn't do a simple loft, they finally implemented it. Same goes for the rendering engine.

So as an Autodesk customer for many years, it's not strange that a few of us may have the impression that the policy on Revit is to have a wait-and-see approach. And if it's absolutely necessary, only then certain new features and technology will be implemented.

I applaud the fact that Revit brought BIM to the masses, but at the same time, it wouldn't hurt to develop tools that go beyond the current needs of the masses. To think ahead of time. SketchUp has brought 3d sketching to the masses too, and at the same token, tools like Paracloud can be used within a free copy of SkecthUp. Something that us Revit users can only dream of.

I see where you are coming from James. Revit definitely seems to be catering towards what the majority of architects need. That makes economic sense in my opinion. Why would they invest money on technology that not many people use right now? I agree that they should invest in it in advance and roll it out when there is enough evidence that the industry is ready for wide adoption.

There will always be technology on the fringe that is bringing new ideas and workflows to the forefront. The difference is that those companies have less to lose. They are small startups that HAVE to innovate in order to stay afloat. They are in a very different business that is customer driven. Autodesk has reached a point of market share and acceptance where they are trying to shape the market as well as try to listen to their biggest customers. They have to do this because they have investors to answer to. Bentley also has investors, but they are not the market leader so they have to be ultra responsive to customer needs, just like ArchiCAD, and just like Revit before the buyout.

So, if you notice, AutoCAD and Inventor get new modeling tools and all the new improvements first. Why? The programming budget for AutoCAD is much larger than Revit’s because AutoCAD’s market share is so much larger. Autodesk is more confident that it can get its money back from the investment. Inventor competes against other products that are well established, so they have to compete feature by feature in that market. With Revit the competition is behind in market share by a factor of 10. So Autodesk’s strategy is to release the new technology in the packages that will definitely yield an ROI, like AutoCAD, and then roll out the technology into the other products slowly once it has paid for itself.

Revit will get these advanced features; it will just take more time. To be perfectly honest I would rather that they fix the current issues of interoperability between disciplines, and lack of site modeling tools etc., rather than expanding into esoteric “grasshopper” like exercises. Let’s get tools that help us pay the bills first, and once we have a solid foundation to build from, we can venture out and innovate some more. You have to remember that to the majority of architects, engineers and contractors, this BIM stuff is still not fully understood and very new. Also, for the most part the impact of all of this has not been tested by the courts.

From a business point of view, I can see the reasons why Revit and all other Autodesk products have gone through certain developments. However, I think the sequence of development for Revit should have been different. Like I said before, the main issue with Revit is its very poor modeling tools. Whether you are developing software for manufacturers, animators, industrial designers, the most essential item of the software are the modeling tools. And once you have laid downs the basics, then you can spend time developing other related features. Revit seems to be developed the other way around. Revit can schedule pretty much everything you throw at him, but doesn’t have the ability to create (complex) geometry. If you can’t create it, you can’t schedule it.

You have mentioned the lack of proper site tools. I can understand that Autodesk has other priorities as Revit is an application to design and document buildings. However, if Revit had decent generic modeling tools (nurbs, splines, subdivision surface modeling tools) then you can easily create an in place topo surface, roads, curbs, using these tools. In the meantime Autodesk can develop a more sophisticated automated site tool. Same goes for stairs and railing. If you can’t create a railing out of the box, a generic 3d spline would have been the answer.

It’s obvious that Autodesk isn’t able to develop every single feature for Revit. Grasshopper hasn’t been developed by Mcneel either. But because Revit isn’t open enough, tools like Grasshopper, Paraclouds, T-splines or a 3d Connexion mouse can’t be used in Revit. This puts Revit in a very awkward position. And with its limited import and export options, makes the situation even worse.

In Autodesk defense, they have inherited an application that wasn’t build by from the ground up. So they can’t be blamed for every flaw in Revit. At the same token, If they want to prioritize on features being implemented in Revit, I would advise them to enhance the modeling tools first. This makes or breaks a software.

The difference between the tools you mentioned and Revit is that they are not purpose built building modelers. They are more generic modelers, and as such have very generic modeling tools.

Revit has always addressed the modeling tools in a discipline specific way in order to make modeling accessible to the non-experts. Also in Revit it is not enough to just introduce new modeling tools, because we have to consider the interaction with other elements. In Revit you can constrain elements to one another, and almost everything is parametric. Because of this, when introducing new tools we have to make sure that when constraining one element to another the system can handle the interactions expected. This is why constraints, interactions and the modeling tools are limited. Revit has to keep thinking about these relationships "interactively,” and in real time as you work on your model. None of the tools you speak of have to do this. Their forms are not tied to each other or do any calculation until you run a script or perform a modeling operation.

In the end you can still model complex shapes with the new modeling tools in Revit. Granted it is limited to the Mass elements, but I am sure they will expand it to other family types soon. You can always model in Rhino or any other software to get the shapes you want. Even AutoCAD has advanced Nurbs, etc now. With the Suite you could use Revit and AutoCAD for not much more money. These files can then be linked into your Revit model to accomplish the documentation of almost any shape.

Keep in mind that most architects I have asked in California and online are not using the advanced modeling tools in Revit 2010 for the curvy shapes. Blobs are actually out of fashion now, as they have been done to death. Granted, I am interested in parametric form generation with scripting, and I agree that we should have this capability soon. I am more interested in more intelligent parametrics, like building in intelligence for code compliance, accessibility, and real time clash detection with all disciplines. If we had that the industry would really benefit for the better.

By the way, one of the major projects that the factory has been undertaking in the last few releases is expanding the API in order to make Revit more open to 3rd party developers. They have also been working under the hood to optimize the code, modernize the graphic engine (DirectX / 3D), as well as the user interface with the ribbon. One of the often overlooked benefits of the ribbon implementation is that it opens the door for 3rd party developers to customize the user interface.

So you see you cannot compare Revit and Swiss army knife modelers. They exist for very different purposes. The more tools you put in one software, the harder it is to learn. Let’s keep Revit simple to use, powerful, parametric, bi-directional and focused.

Your explanation is very clear and I can see why Revit is what it is now. But here is the million dollar question. Do architects really want all of those constraints? Don’t get me wrong, it’s great to have a model that updates automatically, but at the same token, you lose sometimes a full day getting walls to join correctly. I’m not sure if people are willing to give up the ease of modeling with nurbs for constraints in Revit that can sometimes be very frustrating. After all, the reason why architects love SketchUP is because of its generic character and the ease of use.

We tend to talk about architects in the bay area, while we should be looking at architecture in general. For example, unlike the United States, architecture in Europe is over 1000 years old, so the reproduction of ancient architecture is not done. Architecture there is like any other industry where you need to innovate to be ahead of the game. A client in California may commission you to design a Venetian style hotel, while a client in Zurich wants a monolith structure with smooth transitions between the building components. It’s quite difficult to translate that into Revit logic. Walls, roofs and floors are all one component.

So, as you can see, the needs there are little bit different than in California as the needs in Dubai or in Asia. I personally don’t know many firms that use Revit in Europe, however, Maya and Rhino are pretty much standard there. The firms I spoke to see the potential in Revit and are willing to implement when Revit becomes more spline of nurbs based. The new modeling tools is very good first step. I just hope that Autodesk doesn’t get discourage by the current user’s base that don’t need these tools. People are not going to use an application if the application doesn’t have the tools they need. So the wishlist from current users aren’t a true representation of what is really needed in general.Besides, it’s not Autodesk place to question whether an architectural style is trendy or not. They should provide the tools and let the architect decide what style best fits him or his practice. Archicad for that matter, pretty much missed the bandwagon, when they used the blob-is-old-fashion statement as an argument not to develop and implement better modeling tools. These same tools could have been used on complex detailed Greek columns. If blobs are old fashion, aren’t brick colonials old fashion too?

You mentioned a few option to import geometry from Autocad or Rhino, which would work if the geometry can be edited within Revit. This is an ability that all generic 3d modelers and parametric modelers like Inventor have. Without the ability to edit within Revit or link, you will end up with unintelligent dead geometry which kills the BIM process. It’s also very difficult to explain to people why they should model the geometry in a non architectural software and then import it in to a purpose built application for buildings, because the software can’t create the geometry.

At the end of the day, we all want our designs to be built and therefore I hope Revit, which is limited to e certain type of architecture, can expand and become a solution for everyone, the same way Autocad was for a long time.

Well I thank you for your reply. Here is my response:

As far as working with other 3d modeling programs, you can link 3d forms into an in-place mass in Revit 2010. This was not available before but now it can be done. It should make working with Rhino and AutoCAD, etc., easier.

Revit is not limited to a certain type of architecture. It is up to you what you achieve with the tool at hand, and Revit after all is just a tool. Granted some forms are harder than others to model, but they are not impossible. Also, I would question whether blob like forms are actually more advanced in any way. The mathematics for these forms was mastered back in the age of the enlightenment. This is old news to mathematicians today. We are just so behind in architecture that we actually think these forms are innovative because we have not been able to build curvy shapes inexpensively before. The computer is enabling us to build any form without much effort and pretty soon creativity will be unlimited.

That begs the question: Should we do this just because we can? What is the meaning of these blobs? How well do they reflect the society we live in? You mention that a client in Europe is somehow more sophisticated than a client in the USA because they might want a more modern look. I question whether clients in Europe want modern architecture over traditional architecture, or whether this means they are more sophisticated. It might be the case that people respect the profession of architecture more in Europe, because as you said there is a culture of architecture there, so they let them get away with more whimsical ideas. Here in the US clients are not tied up with the respect of the profession and its history. Clients respect the quality of a design based on how well it meets the budget, schedule and the program. Design has value, but it does not come before the other considerations. We architects must remember that architecture is built for a client as an investment, and that investment must yield a return.

The clients that I am in contact with, along with surveys and reports produced by CURT (Construction Users Round Table), which is a consortium of large owners that commission lots of large projects, all say the same things. The most important considerations for them are budget, schedule and program. Sustainability is very important as well. I would argue that more fluid modern forms are appropriate for achieving more sustainable solutions that are not so rigid. In that respect I am all for a different aesthetic. I am not in favor of the aesthetic for its own sake in some sort of masturbatory form making exercise.

So it comes down to the language, and the grammar of architecture that concerned post modernism. What is a wall, a door, and a roof? In reality these do not exist, but are abstractions we invent in order to understand our sorroundings. Walls and other "building elements" are really assemblies of materials. So if as you mention new aesthetics are about the redefinition of the language of architecture into more fluid forms we have to make sure that new language retains structure, because if it gets too fluid, it will be unreadable. The public already cannot understand most modern and current day architecture. We cannot design just to impress other architects. Our work has the responsibility to engage the public.

Also remember that Aesthetics are merely a visual language. We can still communicate a modern message with a well established grammar, just like the words I am writing now. The structure has been around for a long time, but it is still readable today. The only reason architects want to invent these “new” forms, which are really not so new after all, is to appear to innovate and create something different. I think we should concentrate on the message of our work, and not so much on the window dressing.

So let’s define what a wall, window and door means for our industry, so that we may move forward. Our industry has actually lost productivity in the last century because it failed to embrace technology. We seem to want to reinvent our language every decade for some reason. Let’s focus on the real problems in the industry so that we may design better performing projects that make more money for our clients. Let’s integrate and collaborate with builders so that we may end the adversarial relationship that plagues our industry, and gives a lot of our profits to lawyers. This enabling of communication among disciplines is the most important transformation we need to embark on. We need a stable language in order for this communication to become reality.

The comments to this entry are closed.


  • Subscribe