Realtime Architectural Rendering

[adbanks]

I don't think realtime is there yet for the quality that most clients want. The act of 'visualising' a building is trying to create a realistic representation of what it will look like; in realtime, this just isn't possible, which is why it isn't used as much. So for 'Architectural Visualisation', I think realtime is a 'no' for now.

 

In my eyes realtime is best used not for walking round a building, with miniscule res textures and bad AA. It is best used in a different context, for visualising things in a different way. It could be used to explain projects, or layouts of buildings, or land planning (like Landsim3D for instance). Buttons to click on, giving information about a particular room or building, giving the user information that can be called up on demand.

 

The only problem I find is that you need an engine that has a good UI that enables you to do what you want with the model, but more importantly doesn't require any programming knowledge. As far as I know, this is a bit lacking in softwares at the moment.

 

Andy Banks

[JFinkell]

Late entry here...

 

Ian I found your review interesting as well as the commentaries here.

 

However, I do think See3D.be has some good points noting the mis-understaning of the usage, usability and cost.

 

I think the ability to sell or not sell VR comes back to managing expectations, and clearly differntiating between what can be achieved with real-time interactive navigation versus photo-realism animations and thus who target audience is. Today these are still hard to reconcile and achieve cost effectively at the same time. That is quickly changing, though we are still someway's away. Look at the images themselves in the Press Release. They are certainly credible, but some might argue they have a quality that goes back several years and lacks a certain amount of photo-realism. Gaming technology by it's nature reduces certain complexities in modeling geometry and textures for improved navigation performance.

 

But is that so bad? It all depends on who the technology is being "sold" to. Ian, your article and many of comments here talk a lot about the client driving the navigation, but what if it were the contractor or sub-contratcor and they were able to navigate to specific sections of the building and isolate an interference of some sort and work through the associated issues with other trades, certainly their need for phot-realism is not as great, and in fact the "clarity" of the less photo-realistic quality may be an advantage.

 

My point being that though the initial "wow" effect on the part of the client is great, and they were the initial target's of this technology, they are ultimately more interested in the photo-realistic look. Therefore perhaps the bigger sell of this type of technology today is not necessarily to the client directly but to the GC's and Facilitiy Manager's, which already have tools along these lines at their disposal like NavisWorks, Walkinside, and more along the line of achieving more photo-realism, VirTools, etc.

[beestee]

I view these realtime solutions as a design tool, and solutions such as the VRay realtime rendering or even some clever use of MAX 2008's new viewport shading should be sufficient for design decisions. It truly does depend on the client though.

 

Forward thinking is kinda what turned me on to fryrender(which I havn't had much time to explore myself) because they have approached the development in a way that seems future-proof. They are developing a top notch unbiased rendering engine with the real-time aspect kept in mind throughout the development. From what I have read, it seems it will be as simple as changing the render core from standard to real-time and the engine will take care of the technicalities...to the user as simple as sending a standard animation would be right now. Of course, when this becomes more than words it could be a different story, but I certainly hope not because really, it makes perfect sense and the timing could not be better.

[Tamir]

Architectural visualization can be simplified to either construction or seduction.

 

Construction is a term I use to describe the process design. The architect, during design, would inevitably reach a point where rendering can be used for making educated design decisions, be it material selections or lighting schemes, light studies and such.

 

Seduction refers to the fact that; whoever commissioned the design in the first place, as well as those who have a say about it (municipalities and government agencies) need to fall in love with it; and for that to happen… well… Everything is fair in love and war!

 

For most of us in the architectural visualization community, seduction is everything. And the question is… Can real-time technologies help?

 

There is no doubt that at present, of-line CPU based rendering engines, surpass the quality of GPU based engines. But the real question is… Is that a theoretical thing or a practical thing?

 

Let me explain.

 

A high end rendering engine, that is capable of producing photo-real imagery, also requires tremendous amounts of time and expertise. As a result the vast majority of architectural visualizations – even though made with engines capable of producing photo-realism – look as 3rd or 2nd rate computer generated imagery at best. Furthermore; even those images that cannot be distinguished from a photo, usually display a kind of banality that make them look as if a snap shot was taken by a person with a bad pocket camera.

 

By contrast, those images that make us go “WOW” are the ones that are not necessarily photo-real, but rather the ones that contain a visual quality and aesthetics which transcends photo-realism! We can tell that they are made by a computer, but it does not matter; we love it! We Are Seduced!

 

The creation of these kinds of images requires much more than a technical understanding of a particular rendering engine (as most of us know). It requires artistic interpretation and sensitivity as well as understanding of composition and theatrical lighting. These kinds of images are quite difficult to produce using existing CPU based engines, since by nature they require considerable amounts of trial and error.

 

This is where GPU based engines can make the difference, those engines can be harnessed to give artists true theatrical tools (VS the physics based tools inherent in most CPU based engines) and as such they can make the job of seduction a much faster and easier one. This of course does not refer to being able to navigate a 3D environment game style, but rather for the creation of off-line, highly sophisticated, and thought after seductive presentations.

 

In other words; GPU based engines which can be used for the creation of off-line content is what the architectural visualization community needs most, not the ability to navigate 3D environment in game like manner and game like quality.