The Web becomes social. Web 2.0 delivers the technology to get desktop-like mimic into the Web browser. We're on the way to better user experience. New services allow more flexible communication and a new kind of relation to other people.
If you have a look at Second Life e.g. you may be disappointed about the limited presentation. But, for me it's a first step to a "Web 3.0" that is based on 3D technology, with highly dynamic interaction and full-blown multimedia, whereas video content is central. Wouldn't it be nice to get the possibilities of todays TV production combined with the possibilities of a 3D-based Web?
Today this is a bit of science fiction. But, there are a lot of things today we couldn't imagine 10 years back. Maybe we only have to wait another decade to get all this in a manageable tool set.
Why I'm writing about all this? There are already parts of this vision we can realize today. A first step to my vision of a "Web 3.0" is to take certain 3D tools, that allow to create digital humans for Web site integration. This is a small step in interaction design, but it allows to intensify the communication with your visitors. For short, humanizing Web sites with the help of 3D technology.
Today there are tools in the 3D arena that allow to create photo-realistic digital humans. Using such a technology is cheaper and more flexible than to contract a photographer in the long run. Even with the free tools you can get today you can create photo-realistic presentations.
So, what does it mean to humanize a Web site? For short: add some humans to your Web site and the acceptance of its content increases continuously. I don't wanna talk about the marketing aspects here or content quality in general. A certain quality assumed, I take this measurable fact for granted and wanna have a look at the technical aspects now.
First of all some examples to get an impression of what can be reached today. The following images show a basic female 3D model in different occurrences.
The woman on the left shows a grandma from Western Europe. In the middle we have an Asian business woman. On the right you can see a young black girl.
All women are based on the same 3D model. They differ in
- Hair model
- Hair material
- Clothes material
Here's a simple description whats behind the images: The 3D technology delivers meshes, that define the shape, and textures that give an object a surface. So, our digital human examples use a female mesh and combine this with different textures. Additionally, we have morph channels that define how the mesh can be influenced in different body parts. So, you get e.g. morphs to create an Asian or African head, or full body morphs to create a nice fat lady or a teenage girl.
Todays models deliver so many morphs, with so much detail, that it is even possible to impersonate celebrities. The resolution of textures allows to create close-up views that can't be differed from real photos.
But, the most important thing with this is the quality in simulation of light. The quality of photo-realistic presentations depends on the quality of light you can simulate in your renderings. After three years of self-study light is still a challenge to me, but the pros can create such results ;-).
Although, the free tools are not the best for animation, you can let your models even walk on the cat walk. If you invest some money you can add lip-sync animations. This allows you to let a model speak with a voice. You can use your own recordings or invest some more money for a distribution license for a digital voice. There are voices available that differ in language and gender.
So, still images and even animations can be created with todays free tools. What can we do with all this in context to a humanized Web site?
You can create your own character for identification and recognition. This is very interesting, because there are a lot more ways to do this than with real humans. The images above show women that can be used in conservative contexts, like banking or insurance. But, it is also possible to create teens that support a more creative site, or fantasy and science-fiction characters for edutainment sites.
After you have designed such a character, also called an Avatar, with
- clothes, hairstyle, maybe makeup
you can visualize scenes with it. Still images e.g. can help to visualize a context of a text block on a page. Your character can have different poses and facial expressions to emphasize emotional aspects of your product. You can even use speech balloons to create comic-like conversations with the visitor. With this you can e.g. express all the critics the visitor feels about your product presentation. If the character design follows the disposition of your product's target group this can create solidarity. Next step would be to let the character refute the arguments ;-).
Lip-sync animations help to intensify the emotional tie. This face-to-face communication, the visitor already knows from the real world, let her feel a real conversation. This is of course a one-way conversation, because the visitor can't talk back to your character.
For this we have so called Chatbots, that add the possibility to interpret written text of the visitor. But, feasible implementations are very expensive, because it is still not trivial to analyze human language. There are OpenSource implementations you may have a look at to experience the limitations of such systems yourself ;-).
If we have a look at the complete site your character can be a companion to the visitor. It can present e.g. a site tour and place emphasis on certain pages. Best of all, it helps the visitor to recognize your site. Humans are predestinated to recognize faces on a subsecond timescale.
Finally, some software links for self-study.
Modeling Digital Humans
The most important application here is Poser (PC, Mac). It's free counterpart is DAZ Studio [D|S] (PC, Mac). There is a model starter pack from DAZ you can get for free, too. It can be used in both tools (PC, Mac). Although, DAZ was a pure content provider for Poser characters in the past, today there's also a 3D software portfolio. They started to develop D|S in the timeframe of Poser 5, which had quality problems over years. At that time the functionality of Poser and DAZ was almost the same. Today, I recommend to use Poser because of flexibility. There are some interesting D|S plugins in the DAZ shop you can buy, but for me D|S keeps a marketing tool to sell content.
Most interesting with DAZ is their Platinum Club offer. There's no offer that's cheaper to get Poser stuff. Other sources for Poser add-ons are Renderosity, RuntimeDNA and PoserPros (now also part of the DAZ universe). Content Paradise combines the offer of e-frontier, the producer of Poser, RuntimeDNA, Renderosity, and shops of smaller vendors, like PhilC.
There are also free offers at Renderosity and DAZ (although a bit out of date). The arising star here is ShareCG. Besides this there is a ton of other smaller vendor site with free offers, like my wife's Poser site with free textures, tutorials, etc.
DAZ also offers a product called Mimic. It can analyze audio files and creates a special pose file that is adapted to a previously chosen character. Finally, you can assign it to the corresponding model in Poser or D|S.
With the Poser 7 license you get the Talk Designer that produces similar results. If you plan to use it, have a look at the Expansion Kit, that delivers support for every important Poser character on the planet.
Payable digital voices for commercial use, this includes use on Web sites or inside products, are offered by Cepstral. For me the Natural Voices from AT&T sound a little bit clearer, but they may be beyond price for small projects.
There's also an OpenSource voice processor implementation in Java, called Mary. It combines different research results from the last decade. Some voice combinations can't be used in commercial contexts (e.g. voices from the MBROLA project). To be honest: you can work with it, but the quality of those voices is obviously worser than that from the Cepstral voices.