This is a late addition to the interview roster. Joshua was kind enough to offer his time to answer some of the same questions I’m asking everyone, plus some that are unique to his company, Kaon Interactive, which does many of those interactive 3D product brochures (among other things) you see on the Web. As such, he’s had to solve a lot of Web-3D-related problems for the last 5 years or more. Please enjoy.
Joshua Smith is co-founder, CTO, and Alpha Geek of Kaon and has served in lead development and engineering roles with the company since its inception. He is responsible for the technologies that serve as the infrastructure to the company’s Kaon Advisor platform, the Meson platform, and Kaon PDF products. Prior to joining Kaon, Joshua served for six years as Software Architect for advanced simulation systems developed by BBN/Loral Corporation. He holds a Bachelor of Science degree in Computer Science from Worcester Polytechnic Institute.
These questions were asked and answered via email, around the 4th week of September, 2006.
AB: Joshua, what’s your vision of "Web3D?" How do you differentiate it from the common "2D web" experience and existing "networked 3D" applications, like Google Earth or Second Life?
JS: I think the natural place for 3D on the web is in "rich media" applications. Most people have seen really fancy Flash pieces that mix traditional user interface elements with video and vector graphics. Add a good 3D rendering engine and you’re there. Frankly, I think if Flash included a 3D rendering engine, all of us 3D viewer companies could take our toys and go home. But I suspect Macromedia/Adobe feel so burned by the Shockwave 3D and Atmosphere flops that it’s going to be a while before that happens (plus, everyone inside Adobe who knows anything about 3D is now on the Acrobat team). I emphatically DO NOT think it is about replacing the whole browser with a 3D interface and stuffing the 2D web onto unreadable perspective-distorted planes in that space. I think that’s just plain dumb.
I think Google Earth is an insanely cool application with an amazingly usable UI. But it is nothing like the web, so I don’t see what it has to do with Web 3D. I haven’t used Second Life, but I’m familiar with the genre, and while I suppose you use a web browser to get there, it isn’t really the web either. While those are networked applications, and the web is a networked application, I think that’s where the similarity ends. The web is not really about the network. It’s about the content you can reach via that network. It’s persistent. It’s searchable. It is not transient like a game or a chat room. The web is a means for individuals and companies to convey information in a compelling way. Interactive 3D models help make rich media on the web much more compelling.
AB: Personal computers have been powerful enough for decent 3D graphics for 5, maybe 10 years. Yet, for most participants on the world wide web, 3D is not a typical daily feature, despite some very successful specialized uses of 3D. Why do think that is?
JS: Content creation is the big issue. For example, you wouldn’t think of selling something on ebay without including a picture. If getting a 3D model of the thing you were selling was as easy as taking a picture, then 3D would already be everywhere. But it isn’t that easy. It’s extraordinarily expensive and time consuming to produce 3D models of real things. So 3D is only used where there is money to spend, which means professional marketing applications. Over the past couple of years, we’ve managed to lower the cost of getting professional quality 3D models by an order of magnitude or two, which has led directly to a big uptick in the number of products we’re able to put onto the web in 3D for each customer. But a big part of the cost reduction is due to economies of scale, which means we can only lower prices dramatically if we increase the number of products we model for a customer. That leads to a lot more 3D content on a given site, but doesn’t do much to increase the number of sites that have 3D.
The other problem is the lack of a ubiquitous 3D platform. If a content developer wants to put slick vector graphics into their site, they are very comfortable with the expectation that the end user will already have Flash installed. And since Flash added video, I’m seeing a ton more video on the web. Windows Media, Quicktime, and Real Player each could guarantee you about a 40% chance that the person viewing your site would be able to see the video. That wasn’t good enough. Flash bumped your odds to 85% which made all the difference. It doesn’t even matter than the video compression in Flash is terrible. It’s available — that’s what matters. The same is true for 3D. We’ve been quite successful because our 3D technology requires Java 1.1 which "just works" (as the Shout 3D guys used to say) on >90% of browsers. But using unsigned Java applets means we cannot use any of that 3D horsepower you referred to in the question. We do it all with software rendering.
We might be on the precipice of a big change in these two roadblocks. Strata has a product called "Foto 3D" that promises to make content creation a lot easier for the masses. And we’ve licensed them our technology (embodied in their "Strata Live 3D" product) to let that content run on the ubiquitous Java platform, so perhaps we’ll start to see more "down market" use of 3D.
AB: Kaon Interactive has made a business out of the "3D product pitch," where a product is depicted in fairly realistic 3D graphics inside a browser window, which lets the potential customer rotate and, in some cases, interact with the products being presented. Breiefly, how is it done? In what ways do your end-users and corporate clients inform these windowed 3D experiences, both from a design and perhaps a technological perspective?
JS: The vast majority of the work we do is delivered in our "Meson Platform" Java applet, which includes a powerful programming language, all the 2D stuff you need to create rich media, and a 3D rendering engine. It’s important to note that the product models almost always include interactive behavior. Just spinning an object around without any behavior is only marginally more interesting than a collection of stills, and is almost never worth the cost of creating the 3D content.
What we’ve figured out is that the interactive 3D experience builds an emotional connection between the consumer and the product. In sales & marketing, it’s all about that emotional connection. When you go to buy a car, the sales guy tries really hard to get you to sit in the drivers seat. He is trying to get you to bond with the vehicle on an emotional level. To us geeks, this may sound profoundly silly, but it’s absolutely true, and it’s the reason marketing with interactive 3D product models works so well. Playing with a product in 3D, seeing how it works, zooming in on the features that interest you, and having it all be as realistic as a photograph… that’s an emotional experience you don’t get looking at a picture on a web page. Here’s an interesting stat: when Sun put our 3D models of their servers onto their site, more than 40% of the visitors spent more than three minutes looking at the models. More than a third downloaded the 3D-enabled PDF version so they could continue to play with the model off line. Customer satisfaction with their web experience jumped from 63% to 90%. It’s all about building that emotional connection between customers and products.
For years, we would replace the product image on a web page with a 3D interactive experience. But over the past year or two, we’ve been working a lot on integrating the 3D experience with the rest of the marketing story for a product. We’ve developed sophisticated UI templates that let us tell the story in beautifully rendered text, and tie it to the 3D interactive product model so there is continuity of experience. 3D in a window on a web page is cool, but it doesn’t address the whole problem of how a manufacturer can tell the product story in a consistent way through diverse sales channels. You really need to bundle the whole product experience together, and an integrated 2D/3D presentation is the only way to do it. Some sites now use our applet to show their whole product catalog with most products in interactive 3D, some as zoom-able 2D, and all accompanied by features, benefits, and the product story. For the customer, there is a huge benefit because there is no applet start-up delay as they go from product to product. It’s all just one seamless experience, similar to an all-Flash website. (Having a rich client interface also lets us do cool things like client-side catalog search, that gives you results as you type, instead of having to wait for the server to do the query.)
AB: In what other ways would you like to deliver 3D content to end users that is still perhaps on the horizon? How do you see people interacting with 3D marketing materials in 5 or 10 years?
JS: At Kaon, like a lot of the 3D industry, we’ve been about 5 years ahead of the market all along. What consumers see as cutting edge today, we’ve really been doing since the late 1990s. So my prediction is simply that the things early adopters are doing with 3D in sales & marketing applications today will be commonplace in 5 years.
The most interesting parts of our business these days are actually not on the web at all. The real action is in putting a consistent marketing message, anchored by interactive 3D products, into all the sales channels. That means not only the web, but on mobile networks, in PDF, at trade shows, on sales force laptops, and in retail. And by retail, I mean inside stores. We have a couple pilot projects going right now with "big box" retailers, to put high definition 3D interactive product models on big HD touch displays. Floor & shelf space are hugely expensive for these retailers, so being able to extend the line of products available in a store by using 3D technology is really a no-brainer. I think that in just a couple of years, most appliance sales are going to be done using interactive 3D displays in stores. The products are just too big, and the margins are just too low, to justify using up all that floor space. There will be just a couple physical products that you can use to get a feel for how well the brand is built, and then you’ll use high definition 3D to browse the thousands of variations available.
AB: So what aspects of the main browsers (Firefox, Opera, IE) or web protocols do you think need to change to make Web 3D more common?
JS: Right now the only ubiquitous platform that can do 3D is Java, so I would LOVE to see the browser makers work with Sun to get the JVM to start up faster. We’ve done a ton of work to make sure our applets are tiny, our models are highly compressed, and everything streams well. But none of that matters for the first 5 seconds while you wait for Java to start up. (Thankfully, that only happens on the first Java-enabled page you hit, not on every page.)
I suppose in principle it would be nice if they all implemented a standard 3D rendering environment so 3D content on web pages could "just work" without Java or any plug-in. But there are huge problems with that idea in practice. It takes 5-7 years for a browser change to ripple through all the users and become ubiquitous. So anything they do today will only start to be useful in sales & marketing applications sometime mid-next-decade. The only candidate 3D format for them to use is X3D, which is really just VRML with an XML encoding, which is not a rich media format. It’s strictly a 3D format, so the best you could do would be to put a 3D window into an HMTL page. Everyone in the W3C seems to think that integrating content types together is somebody else’s problem. Plus, X3D is way behind on basic Web 3D enablers like streaming textures, so if the browsers all added X3D today, what we would have in 5 years is 10-year-old technology. Not compelling.
I think Flash is a good analogy here. Google has shown that you can do some really neat things using just the capabilities built into the browser, but the very best AJAX application is on par with only the most basic Flash applications. In 5 years, people will be creating AJAX applications that are very compelling, but one can only imagine what Flash will be capable of by then. Open standards and commodity browsers were a great thing to get the web up and running, but they’re not competing effectively today with proprietary solutions from Adobe, Apple, and Microsoft.
AB: What aspects of the current crop of standards, X3D (VRML), COLLADA, etc.. are useful to you in your business? Where do you want to see these standards go in the future?
JS: I think COLLADA’s focus is exactly right: being the glue between the tools. It is very user-focused, solving a real problem that everyone who creates 3D content has. And at the same time, it will lower the cost of entry for software vendors who want to contribute to the 3D content creation toolset, but cannot afford to write plug-ins for 20-some vendor’s 3D tools. I also love that it is centrally controlled by a team of professionals, not a bunch of volunteers: the specification is really, really nice. But COLLADA is still very rough around the edges: google sketchup is starting to flood the market with not-quite-valid COLLADA files (which could lead to a repeat of the whole "it’s wrong, but it works in IE, so it’s right" HTML debacle); installing a COLLADA plug-in to 3DS MAX was about as easy as installing linux without a distro (that is, insanely hard); and the format itself is so tangled with cross references and unstructured tables that the hype about using XSL to convert COLLADA files to other XML representations is clearly coming from people who know ABSOLUTELY NOTHING about XSL.
X3D/VRML is just a scene graph interchange spec, which is about as useful for interchange as an OpenGL debug trace. However, VRML97 is absolutely ubiquitous as an export format from CAD, so until COLLADA gets the kinks out, and gets adopted by the CAD vendors, I don’t see any way to avoid it for now.
U3D is a truly awful format, foisted on us by some guys at Intel who’s religion about progressive meshes has blocked their ability to recognize what a tremendously ill-conceived idea streaming geometry is. However, it’s supported by Adobe Reader, and nothing else is (yet), and Adobe Reader is ubiquitous, so we will continue to use U3D as a presentation format because our customers LOVE having interactive 3D in PDFs.
I’ve read a little about FBX, but I don’t know anyone who is actually using it. If it comes down to COLLADA vs. FBX for data exchange, I think COLLADA is probably preferable. It’s XML, rather than binary, which makes implementing it a lot easier. And COLLADA definitely has a lot more momentum than FBX right now, with the Google connection. And, of course, you can’t ignore the fact that FBX is so tied to Autodesk that other vendors are going to be hesitant to embrace it in the long run.
For the future, I would like to see the standards bodies focus on interchange, not presentation. XML is a huge success; SVG was a flop. TIFF is a huge success; JPEG2000 was a flop. VRML as a 3D data exchange mechanism between CAD and visualization tools was a huge success; VRML as a browser plug-in was a flop. There’s an undeniable trend here. ONLY a truly open standard can effectively address the problem of data interchange between vendors. So they should do that. But open standards are a lousy way to get content to end users, who expect things to "just work". Even HTML, the most successful presentation standard ever, has taken nearly a decade longer than PDF did to get 2D pages onto users’ screens reliably and consistently. (Of course, there are successful quasi-standards for presentation like PDF/E, U3D, and H.264 which are more like anti-trust work-arounds than open standards.)
Right now it looks like COLLADA has a better foundation for data interchange than X3D, which is all tied up with presentation stuff that doesn’t belong in an interchange standard (SAI, Scene Graphs, Sensors, etc.). (Yes, I know about the "interchange profile" of X3D; calling a subset of your standard an "interchange profile" does not make it useful for interchange.) But COLLADA hasn’t yet taken ownership of issues like "archival data" and the whole CAD problem, which the X3D group has at least acknowledged are important. Time will tell which of these gets more industry support in the end, which is what makes or breaks a standard, but my money is on COLLADA right now.
AB: Thanks for your time and for sharing your thoughts with us.