“The Internet is wonderful for distributing and comparing information, but it doesn’t replicate the actual handling of items,” Silton explains. “While the Web can build product awareness, consumer knowledge, and to some extent affect user preference, product trials are still an important and often critical step before a purchase is made. There’s an amount of detail about how something looks, moves, and relates to a user’s context that Web developers can’t duplicate with existing imaging and VR technology.”
HyperActive was designed not only to give customers a more true-to-life interaction with products before purchase, but also to enable online stores to provide better support during and after the sale with real-time collaborative capabilities such as chat, white-boarding, laser pointing and measurements.
While images, grabbed or graphically synthesized, have shown products from different views and in different colors for some time now, they rarely function interactively and can’t let a potential buyer experience the product in context with the other objects–using the features of a CD player, or placing a couch in with other furniture. Images are utilized in Web design to add interest and to convey information, and can, of course, be static or dynamic in nature.
Dynamic images don’t require programming, but they do require design; meta information about the picture is placed into the image itself. Examples of dynamic image formats are FlashPix, QuickTime VR and VRML, all discussed in these pages as they’ve come on the scene. FlashPix supports multi-resolution formatting, allowing users to zoom in on key image details, apply certain operations (i.e. colorizing) to the image, and also allows audio data to be attached to the image file.
QuickTime VR supports 36O[degrees] viewing of images so that the user can look around from a single point of view in a scene, walk totally around an object, or rotate an object by 360[degrees]. This technology is often used to put the user behind the wheel of a new car or to create virtual tours of buildings and locations. However, QuickTime VR limits the user to navigating a series of still photos taken from different angles or directions. To navigate a virtual world, multiple data sets need to be selected and loaded, accomplished most often via a map-like navigation tool used in conjunction with the QuickTime VR viewer.
VRML: NOT REAL ENOUGH
Going even further towards a more real interactive experience, Virtual Reality Modeling Language (VRML) was developed to allow the creation, viewing and manipulation of 3-D objects in shared virtual worlds. It essentially adds the dimensions of depth and time to the viewing experience by describing objects and providing constructs for displaying them.
While, in concept, as we’ve discussed here, VRML can enable the creation of virtual shopping experiences, it has a few fundamental flaws which have kept it from being used for these purposes more extensively. First, since it was designed in research environments and developed as an encompassing standard, it is very big and somewhat fragile. This means that large files need to be downloaded to the user’s PC, and a large amount of CPU resources are needed to create those 3-D views.
In addition, VRML is based on the very rigid concept of a scene graph, a paradigm more suited to representing static worlds than dynamic, interactive ones desired in a VR experience. X3D, a new standard, proposes to be lighter in weight than VRML, but it still is based on the VRML concepts of scene generation and interaction. More importantly, objects rendered in VRML look, well, virtual. The graphics remain graphic-like-period.
But don’t we regularly address that problem by laying imaging-derived skins over the graphic object skeleton?
Not well enough, says Joshua Smith, Kaon’s CTO: “Actual photographs can be draped onto these models to create more realism, but it is extremely difficult to get the features from the model and the photos to match up. The net effect is that objects in VRML worlds look cartoonish.”
This is why Kaon’s HyperReality was designed to produce objects with photo realism and product authenticity-the 3-D geometry of the object is authentic, as opposed to just giving some views where scale may not be accurate.
2-D TO 3-D CONTENT
The process of creating content for HyperActive viewing begins with digital 2-D photography from a few different angles of an object to be modeled. These images are then converted into 3-D wire-frame models with Kaon’s HyperReality technology. 3-D CAD models can be utilized, but unless they already exist, the HyperReality process is much faster, on the order of hours as opposed to days, and much more photo-realistic in the end. This level of image credibility is achieved by draping the 2-D images onto the 3-D mesh. The unique aspect of the Kaon approach is that the mesh gets created from the same photos that are overlaid on it, creating an uncanny level of realism.
Other rendering engines use bump maps, multi-texturing, etc., based on the idea that you might make pictures look more realistic by using a realistic lighting model. However, this technique still does not look real unless you use high-end rendering engines that aren’t widely accessible; they rely on complex models and a good amount of rendering horsepower.
When I first visited the Kaon Web site using a 120 MHz laptop over a 56K connection, I was not only astounded by the realism provided in the rendered image, but more notably by the speed of the interaction possible. To see what I mean, visit the room planner at www.kaon.com. If you have ever used a 3-D home planning application, you will be amazed by its realism and ease of interactivity!
By incorporating the HyperReality content-generation process with the HyperActive programmable rendering technology, Kaon applications can be customized and animated using Kaon’s XML-compliant scripting language called Nimble. The resulting models can be posted to a client’s site, Kaon’s Web server, or an application service provider’s facility. Those models can be accessed and manipulated within a Web browser over a standard Internet connection.
Using commonly-available XML authoring tools, users can create Nimble scripts which incorporate ballistics, collision, and other advanced features. This leads to another important differentiation between Kaon’s HyperActive technology and other VR tools such as VRML and X3D: HyperActive defines simulations, not scenes. This means that everything in the scene has a lot of behavior attached to it. For instance, turning a couch upside-down could cause its skirt to flip over, just as in real life. This allows for the creation of very realistic and interactive objects and worlds.
Not only is the HyperActive technology extremely lightweight (about 40k for objects and 380k for the rendering engine, which only needs to be downloaded once), but it also uses very little CPU resources. In fact, it uses the direct renderer built into the Windows OS. No hardware 3-D accelerator card is needed, but any 3-D accelerator is automatically detected and used, including 3-D acceleration built onto the motherboards of almost all of the new PCs being sold today.
When a user visits a Web page that contains a HyperActive object, the HyperActive viewer is automatically downloaded and runs in the Browser. Any Web server can serve HyperActive content. However, if collaboration is used, an echo server is needed to bounce messages off of, and Kaon provides one for free. (This is a temporary measure until multi-casting is available on the Internet.) The echo server is not a streaming server, as the assets being manipulated on the client’s end are all in cache. The only information being sent are the commands to manipulate objects and mark up the scene. This is why HyperActive collaboration is so fast and represents an ideal technology for interactive gaming.
BEYOND SHOPPING: OTHER APPS
In addition to accelerating time-to-purchase for e-business retailers, Kaon’s HyperReality can also be used to improve product development, sales training, customer support, interactive gaming, and product documentation. Collaboration and development is facilitated with text and voice interaction. Using HyperActive technology, help-desk personnel can work with a customer to troubleshoot a problem with 3-D images and direct chat, thereby reducing the need for on-site service. 3-D games and entertainment can be easily created with amazing realism. Distance training and education can be conducted with multimedia content for deployment over the Internet,
How are e-businesses reacting to this new technology? Furniture.com has utilized Kaon’s technology to dramatically expand the ability of consumers to visualize their home decor options, with a new 3D-View service (www.furniture.com/ promotion/slipcover/slip.asp) that allows users to select a chair or sofa, instantly wrap it with a fabric pattern, then move, rotate, zoom and take custom measurements using HyperActive.
Carl Prindle, Furniture.com’s senior vice-president of product development, explains, “We want to provide consumers with the absolute best experience in decorating and home furnishing, and (these) enhancements to our Web site will ultimately enable shoppers to select and arrange new furnishings in a 3-D re-creation of their homes.”
So, when my wife and I are ready to purchase our next couch, I can imagine that we might be doing it from the comfort of our home by visiting an on-line virtual showroom, complete with other shoppers, the music of 101 violins playing ZZ Top’s “Sharp-Dressed Man”, and…oh no…a virtual, chatty salesperson who won’t take a break!