Realtime Architectural Rendering

[ikinman]

I have been asked by a writer for BusinessWeek to comment on a press release by HKS architects on their creating architectural visualizations using realtime gaming engines. Because both the press release and my comments are rather long for a forum post, I have uploaded the text files here:

 

http://www.animation.to/CGArchitect/RealtimeNotes.txt

 

I welcome anyone's input into both the press release as well as the comments I have made - historically my attitudes towards things such as this tend to be somewhat pessimistic, so if anyone has more positive experiences let me know. I would enjoy giving BusinessWeek some comments in addition to my own.

 

Thanks,

 

Ian

[Jeff Mottle]

I have been asked by a writer for BusinessWeek to comment on a press release by HKS architects on their creating architectural visualizations using realtime gaming engines. Because both the press release and my comments are rather long for a forum post, I have uploaded the text files here:

 

http://www.animation.to/CGArchitect/RealtimeNotes.txt

 

I welcome anyone's input into both the press release as well as the comments I have made - historically my attitudes towards things such as this tend to be somewhat pessimistic, so if anyone has more positive experiences let me know. I would enjoy giving BusinessWeek some comments in addition to my own.

 

Thanks,

 

Ian

 

He called me too yesterday.

[ikinman]

Jeff,

 

I'm curious of your comments. Did you respond yet?

 

-Ian

[Jeff Mottle]

Jeff,

 

I'm curious of your comments. Did you respond yet?

 

-Ian

 

I did the interview over the phone. Many of my comments were similar in nature to yours. There are very specific cases where this type of tool and implementation can be useful, but cost (both to the client and developer), timing required, and overall usefulness on a mass scale are not there in my opinion.

 

I've often made the comparison of the number of VR and realtime companies as being similar to the web 3D craze of the mid 90's. It's a fad whose time has not yet come. Like I said there are very specific cases where it is useful, but I think they are very few and far between.

 

The interviewer mentioned how they had used the technology for a stadium, which if the client required can be a really useful tool to ensure proper visibility lines in realtime, but that's a pretty large scale project, so the specialized tools can warrant the cost and time.

 

For 99% of projects, carefully selected images or a short film I think will always be much more effective and more cost effective than VR. In my opinion until VR and Realtime are a technology used by the masses, I don't see widespread adoption for arch viz.

[See3D.be]

I did the interview over the phone. Many of my comments were similar in nature to yours. There are very specific cases where this type of tool and implementation can be useful, but cost (both to the client and developer), timing required, and overall usefulness on a mass scale are not there in my opinion.

 

I've often made the comparison of the number of VR and realtime companies as being similar to the web 3D craze of the mid 90's. It's a fad whose time has not yet come. Like I said there are very specific cases where it is useful, but I think they are very few and far between.

 

The interviewer mentioned how they had used the technology for a stadium, which if the client required can be a really useful tool to ensure proper visibility lines in realtime, but that's a pretty large scale project, so the specialized tools can warrant the cost and time.

 

For 99% of projects, carefully selected images or a short film I think will always be much more effective and more cost effective than VR. In my opinion until VR and Realtime are a technology used by the masses, I don't see widespread adoption for arch viz.

 

With the above quote, I would say, wake up guys!

 

I still see that there are still many miss understanding of the usage, usability and cost of using Real-Time 3D. If you would say this before 2YK I would agree, but these days….

 

Just one example, you render out a short film, you are done, then the client says, ooo... I want yellow tiles instead of red… You can start over again with rendering. I will change that on the fly…

[ronll]

I think Jeff's point, and one I would agree with, is that it is not that we can't do it, it's that the market doesn't support it yet. I have been able to do real-time full-screen walk-thrus for several years now. But so far I have only sold it to one client. For whatever reason; lack of portability, motion sickness, the end users are afraid they might break it, whatever, it is a tough sell. Keep in mind, most of what we know about buildings in places we haven't been to, is learned from still photos in magazines. If you CAN sell it to your clients, more power to you.

[See3D.be]

Well about 15y ago I use still renders for my design presentations during my garden architect studies.

 

The teachers put me in a deep dungeon to make sure I was not able to do it ever again, but I did it again and again. .. With the comment people don’t want this because they are used to pencil – water color drawings….

 

And today I have the same feeling here again… but we will see when the game generation people start building homes… who can still sell still renders….

[Jeff Mottle]

With the above quote, I would say, wake up guys!

 

I still see that there are still many miss understanding of the usage, usability and cost of using Real-Time 3D. If you would say this before 2YK I would agree, but these days….

 

Just one example, you render out a short film, you are done, then the client says, ooo... I want yellow tiles instead of red… You can start over again with rendering. I will change that on the fly…

 

I see you're a Quest3D reseller, so I guess your opinion is somewhat biased, but to pick your brain, what are your thoughts on why VR and Realtime has not yet been widely adopted in the arch viz industry? I know Quest is one of the leaders in this type of work, and in my opinon, will be one of the companies that will still be around when and if it does become more mainstream.

 

I just don't think that the majority of projects warrant this type of application, even if it was as simple as pressing one button, the clients did not require fast video cards and have an aptitude to navgiating in a 3D world.

[See3D.be]

I will probably repeat me self now, but I guess like I experienced 15y ago. Where hand drawing illustrators thought that computer renders never would take over there job…

 

I agree its not easy to sell, but mostly that’s still based on the misunderstanding over real-time 3d.

 

One of them is already the video card, you don’t need a GF8900GTX photo realistic results… its also possible on a GF2.

 

User navigation, well not to hard to build another type of navigation then what you see in games.

 

Cost, I don’t see why a real-time application should be more expensive then a pre rendered movie.

[ikinman]

I actually use Lightscape on a fast modern machine, which, since the LS code was written ten years ago and is relatively simple, actually functions pretty much as realtime software these days. It's useful for previs and such. Ten years from now, one will essentially be able to do the same thing with today's rendering software. And ten years from now the situation will be roughly the same: you can get immediate results with realtime, or you can wait longer for processing something that is even better looking.

 

While the Quest3D engine is impressive, especially compared to realtime engines of several years back, it's certainly not as good as mainstream rendering software. You get more processing time, you get better results. It will always come down to what the client wants - immediacy or better quality. Interactivity or a narrative. From a business standpoint, I've never had clients that want that immediacy and interactivity. Especially interactivity - I've still got clients that have assistants print out my emails; there's no way they're going to handle a joystick and play a video game.

 

Another quick point is that my animation frames take about nine to twelve minutes each to render out - the same as they did when I started my company ten years ago. That seems to be a constant and acceptable processing time, blending quality and expediency. Obviously I'm using much faster computers than ten years ago - it's just that the software keeps doing more and better things. And I take advantage of it. I suspect that ten years from now (unless I'm on a beach somewhere) my animation frames will still be taking nine to twelve minutes a frame.

[See3D.be]

Well Im sure we can have endless discussions about this topic, I hope we can reevaluate the outcome in 10y from now.

 

I just hope that you guys don’t miss the train to the future…

 

But I will catch the train to the next future, Argumented Reality.

[Chris LeBlanc]

Ian,

 

Thanks for sharing. I have too many thoughts on this subject than time to write. This would be a great conversation over lunch sometime.

 

I do not think you are being too negative... someone needs to balance out the over enthuiasm of the press release, in which are several amusing points. Below are some of the high points.

 

"[Realtime] offers client a higher level of graphic photo realism [than prerendered 3D]."

 

Hilarious.

 

"It renders 30 frames per second instead of one per hour - that's 3,000 times faster than traditional animation methods.

 

I would like to see the animation that takes 1 hour per frame. Are they rendering on a x386?

 

"This makes it a first (and only) in the architectural industry."

 

Not really, a free copy of Sketchup can render at 30 frames per second. (and one can even choose to label its output as a higher level of photo realism in a press release :-)

 

 

Seriously, there are many uses for realtime 3D. For presentating data which is objective in nature, such as an automobile, a dna molecule, or even architectural massing and overall form, realtime argueably is viable choice.

 

But that is not the subject of this press release or discussion. As it relates to our industry, one of the primary uses for realtime seems to be in generating buzz. It is the kind of concept that can make an employee or department look good, and makes for good press. It is just unfortunate that HKS' press release couldn't simply read, "we have developed a team of talented 3D artists who can accomodate a wide range of presenation needs using a variety of tools" Which is probably much more accurate but would not likely get noticed by Businessweek.

 

Chris

[3DFactory]

This is actually an interesting thread. I've been involved in several hundreds real-time architectural visualisations, and is right now heading an outsourcing company in Bangladesh which is 100% focused on real-time interactive architectural models. Right now we have 50+ running projects and is hirering as quickly as we can. That should tell something about that there is a market.

But... I will have to agree with the negative voices saying that its still not something the customers really need. The funny thing though is that when the've been convinced of it's usefulness the first time they keep coming back for more.

Regarding the the time it takes to make such a solution, we often compare it to 2-3 ordinary renderings including the render time.

 

best

Claus

[Macpod]

There is a class at uni using Gary's mod for half life 2 as an architectural visualiser. Really cool stuff actually.

[Matt Sugden]

This is an interesting topic, while I have not read the press release, I have a great deal of experience with gaming technology having worked in that industry for over 9 years, before I started doing arch vis.

 

A close 'friend' of mine is a 3D graphics engine programmer (without wishing to sound boastful he is widely regarded as one of the very best ones around) and the conversations i have with him about the graphics capabilities of the next gen games is truly remarkable, HDR lighting, realtime GI, floating point lighting, sub surface scattering on shaders to name but a few.

 

Last time we talked, i felt really quite aggrieved that my renders were taking around 8-12 mins a frame and he was producing 30 frames a sec, and had the bonus of being completely flexible in terms of the access and control of the 3d environment.

 

As far as I can see, it won't be 10 years, the power is here now and is awesome, it's just a question of someone harnessing it and realizing it for arch viz.

 

I think the biggest issue with real time, is that is often gets bundled with web publishing, and as a result the emphasis is on small portable files.

 

I also know from my experience in games that optimization of files, both in terms of geometry and textures is extremely high on the priority list, which would certainly have an effect on the work flow of a lot of people in this industry, who throw thousands of polys into a model without giving it a moments thought, which of course they can do with offline rendering.

[womble]

Bewdy, the 3D 'Games' engine results sound brilliant, the quality of those games is increasing all the time and becoming very attractice. However I feel that the Architectural 3D industry could be harnassing these 3D engines to produce our walkthroughs and still images with much faster render times, rather than focusing on the somewhat 'gimicky' real-time walkthrough aspect.

 

A nicely directed and composited 3D animation has much more 'selling power' than letting a client loose in a keyboard controlled first person environment I feel.

[DavidR]

I don't see the point of using a game engine because of the time required to prepare assets, ie: baking textures, light, etc. This looks great, but you have zero flexibility, so if a client wants a change in a model, everything falls apart. To me, realtime for arch viz means that (GI) light+ shadows are calculated on the fly, which is only partially true in games.

[Matt Sugden]

DavidR, technology in (some) games now is so advanced that GI and shadows are being calculated on the fly, and not baked as suggested. As I mentioned, coders are now looking at expending shader capabilities to incorporate details such as subsurface scattering, for skin on characters mostly, not to mention hair simulation and fluid dynamics.

 

I appreciate the point that putting an architectural model into the hands of someone who doesn't know what they are doing does not necessarily sell the building, however, there is no reason why predefined camera paths can not be setup for dummies, but then also have the flexibility to explore the model if the user so wishes (most of the time they are paying for the model, so why not let them look at it). When all said and done, I find my clients love it when they get to see more and have the model manipulated in front of them, even if it's just a looking at the raw max viewport. And user interface and usability will improve.

 

I'm all for realtime, as waiting for animation renders, (despite the prettiness) to me is the biggest drawback of the job. i think someone made the point that render times aren't improving as such, only more is being done in the 'acceptable' 9min frame times. Which while all being well and good, I'd rather the industry started working towards faster render times, then the quality can come, but unfortunately it's biased the other way, and heavily reliant on hardware.

 

The games industry, especially console based technology works slightly differently, as they have a plateau of technology every few years, so by the end of a consoles life, typically the coding for it is so much more streamlined and efficient as they squeeze every last drop of performance from the boxes.

[Nic H]

Kind of off topic but has anyone played the new Crysis demo?

Its environments are amazing, and I was only using DX9!

Fully interactive environments, HDRI, realtime physics etc. Its amazing to play.

 

Do check it out iof you are interested in see the leading edge of gaming environments

[Macpod]

3dsmax 2008 will have better textured and lit viewports, the shadows will be in rea time as well. so in a way it is becoming real time.

 

Crysis doesn't look that great. I had everything on high and running DX9. Maybe if i had DX10 and was a ble to run it in very high it would be better

[Cubicspace]

Hi Everyone,

 

Chris LeBlanc (sorry to hear you'd left TS), Jeff how's it going - see you in Vegas,

 

An interesting thread - architects know that using game engines doesn't work, they weren't written for that task yet they're seen as a quick fix, which is the last thing they are. Aside from big investment in cost and time, the results are usually average. There's so many things wrong, and most weren't written for the artist. We went down that route at the start, then decided to invest and write our own system from scratch - 4 years later, rtre v1, another 4 years on and v2 is ready.

 

I'll be at AU this year to launch our long-awaited rtre version 2 - you'll find us on the Autodesk Media & Entertainment booth with our Global Publisher Turbo Squid. Tickets and hotel is booked, I'm looking forward to it.

 

So much has happened since launching rtre version 1 in 2002/3. We've rewritten pretty much everything and added a lot of new stuff. Our focus has been the artist again. Our new engine is according to external beta testers, amazing. The realism is hugely improved, as is the speed of rendering, and we know we can improve further still, regardless of scene and texture sizes. Lots of workflow improvements, lots of new interfaces and flexibility, a reworked helper system, big reductions in file sizes, and unlike most of our 'competitors' we're really taking advantage of the higher performance graphics cards which is what everyone wants - I mean how irritating is it to spend big on a new card and find no real improvement from your software - I know it annoys me every time.

 

Plus, our big news was being selected and commissioned by Europe's largest police force to write some specialist new tools on top of v2, for them to use everyday in the Royal Court of Justice in London, for the Inquest into Diana, Princess of Wales death, which is ongoing now. First they compared rtre with every major realtime rendering plug-in, before selecting us to develop the tools they needed. Much more is being planned by us for these and even more tools for a specialist market. Anyone interested can contact me direct.

 

So, much more at AU, hope to see everyone there. We're at QA now, only a few weeks to go, and I'm looking forward to Vegas!

 

It's been too long and I'm looking forward to seeing you all again.

 

Best, Steven

[3DFactory]

-> DavidR

 

I know exactly what you mean about things being a pain to change after baking (lightmaps), but we use a combination of methods with of course baking and vertex lighting being the biggest ones. But also lots with simple dynamic lighting (not GI). There is no need to bake everything. You can also do "compositing" in a real-time model.

Then the client gets the model for review before lighting (like a test-render), and is told that changes afterwards will cost extra.

 

It is a little different method to work with (both modeling and management) than the regular renderings, but not that much.

 

-> Bewdy

I'd love to use all the high-end shaders and cutting edge functionalities, but the problem right now is that arch-viz is most often shown in the offices, and not on the gaming computers with the up-to-date hardware. It's not a question wether it is possible or not (it is), but rather when the office computers are powerfull enough.

[Trick]

For me, the first reason I seldom use realtime technologies (and something I have not read here), is the fact that my clients do not want to give potential users the freedom to explore the world themselves. These are clients that certainly would pay for this technology without hesitation if they saw the use of it. These are the clients that visit large real-estate conventions where all the best presentations in the world are shown: so I guess they saw one or two of the best realtime presentations . With images and movies you can perfectly specify the look and feel of the project the way you and your client want to. With realtime technologies (just assuming for now that quality would be the same as in images and movies) the viewer has all the time to explore the project, and thereby there is no influence in how the viewer may experience this. This may be a good thing in the design process, but in 99% of the cases my end product is for a final project, and then the only thing that is important is seducing potential buyers/investors. Of course I can show architects, developers and other directly related people realtime conceptstages in course of working on a project (which may last some years in some cases), but this will never be shown to these buyers/investors...

 

Of course there will be a time that realtime technology has that much artificial intelligence that this seduction can also be arranged at any place and anytime in this realtime environment, but this time has certainly not come yet...

[Macpod]

What about using the speed of real time renders to produce choreographed sequences of footage?

 

I'm guessing you can hit a record button and just record a particular time line during the real time render.

 

that has to be better than waiting to render 2000 frames.

[Trick]

What about using the speed of real time renders to produce choreographed sequences of footage?

 

I'm guessing you can hit a record button and just record a particular time line during the real time render.

 

that has to be better than waiting to render 2000 frames.

 

I may be wrong here, but are we talking about realtime technology for the end-user or for the creator/developer ? And even if...I better wait for rendering laying back in a chair, then building a complete interactive 3d-environment in the same quality and then choose my camera paths. Of course I dream of creating environments in "Crysis"-style, but I don't want to leave this to my viewers: this would be the same as if the architect/developer also lets me design his building...