http://www.makehuman.org/

create a model based on photos of yourself:

http://www.facegen.com

Great stuff, Erwin! I’m going to have to play with this, and see if I can make a “virtual me”. That is, if I can somehow come up with the $300USD required to buy the full version.

I like to add some more info on Blender here:

One very special option in Blender is it’s game-engine. It uses a real-time physics engine. I think this is a great option to create a user-interface for a chatbot or even strong-AI. I can see how the ‘virtual sensors’ in my model later on are going to be hooked up to such a model. Think of it as ‘virtual robotics’. This will be much cheaper then to build a real robot, and virtual animated characters already move much better then hardware-bots.

Then think of the possibilities when combining AI and a game engine together with augmented reality.

http://www.blender.org/features-gallery/features/ (scroll down for info on the game-engine)

video: http://www.blender.org/features-gallery/feature-videos/?video=game_engine_one

yeah, great, I just told you about our 126 synonyms, and you are simply introduce another one: ‘virtual robots’, i.e. a robot without a physical body, and still having all its sensors.

tx for the links!

oops, we already had ‘virtual robot’ in our list, but not the ‘computerized avatar’ which I found through the google search on virtual robot .

I don’t market myself as an avatar building company but over the past 4-5 years I have produced quite a few of them.
My working process makes use of the following software:
facegen : to produce a starting point model from reference images.
3ds Max : to model, rig and animate the avatar.
Mudbox: to add finer detail if needed.
Photoshop: to produce the textures.
Combustion: to post produce the avatar
flash: to deliver the avatar as an interactive element.

Thanks Adrian! Good to have animation experts on board!

Ladies and Gentlemen, may I present to you my basic lip sync animation program, running on a website hosted by my private webserver, since I’m still writing it. This video is its public debut…

‘
.
.

Special Preview: http://www.youtube.com/watch?v=3MCFWcS3LnA

‘
.
.

My guess as for why dozens of views were received on the first day on Youtube was that people enjoy original designs. Simply a clean, small avatar—uploading it to YouTube has blown it up about four hundred percent, hence the fuzziness.

Special Thanks

To You For Watching.

Don’t usually post this stuff here, I just post it on AiDreams. Anyhow here’s a couple of past attempts.

The first one is Pseudo 3D with pre rendered frames :  https://www.youtube.com/watch?v=MHBxeYnUMb0

This could possibly be migrated to web pages using Javascript.

This one is true 3D, made using Daz Studio models and animated in WPF in C# : https://www.youtube.com/watch?v=H3p5la_0Ar8

The latter one I like the most because it has a lot more options.

Currently I am working on things like blinking, eye movement and noise to make her look a little less static. Oh yeah and I have hair on the newer ones I am working on.

The latter model appears optimized for fast rendering.  Render speed may slow down once hair and a body is introduced to the scene. In Makehuman, referenced on the top of this thread (by Irwin back in 2011) I noticed that since then, there has been a crew cut hair style added to the face texture.  If you add this to your model’s face texture, it may optimize the level of detail available to fast rendering.

There’s no optimisation on the mesh or anything - it’s a straight export from Daz Studio. Adding hair does not really affect the speed at all. That last model was in some way optimised by myself in that it only animates part of the mesh - the mouth and teeth.

The thing that affects speed is mainly the number of polygons. This has an affect on how many 3D points can be calculated over time when something is morphed. That model has far too many polygons to work smoothly on my i5 Haswell when you start introducing more morph-able parts - so I went back a generation in the model and used a reduced poly model.

The model I am using now has about 17K polygons and again I only animate part of that model. What you cannot see in that old model is that there is a full body under the head, I’ve actually got a full figure in there. I’ve added hair to this and it has no real affect on the smoothness of rendering.

I’m also considering just using the head if clothes do slow things down. I have built some tools that allow me to reorder, remove and de-fragment groups in OBJ models so that helps.

Glad to hear it is animating so efficienty.  My comments were based on the number of polygons being reduced by leaving the hair and body out of the rendering. I’m sure you must of heard the 3D rendering discussion about hair slowing down rendering, since each folicle may need to be rendered separately.  But, I also consider all the new techniques available.

My guess, is there may be some type of state of the art rendering technique going on, perhaps removing, from the render process, the maximum number of polygons automatically. In any case, it always makes for an interesting discussion.

17K polygons?  Yes, whenver it comes to making A.I. more convincing, it seems there are always tens of thousands of something needed.  Building your own tools, sure is a way to learn a lot.

I am looking for animated characters for web. The problem of the current options is that they will not work on all the platforms. I would like something that could work on any browser, including mobile (android, IOS).
Options like “Site Pal” require Flash (not supported by android). I would also like to avoid options that require installing software on the device.

However, at the moment I am finding difficult to find HTML5 alternatives including lip sync and text to speech.
I found this option: http://chungswebsite.blogspot.fr/search/label/chatbot_chung , which fulfills my technical requirements but it is not advanced enough to be applied in a professional environment.

Is anyone aware of any HTML5 alternative?

Thanks!

> http://en.wikipedia.org/wiki/WebGL

What is required is WebGL; and as far as I know, there is no stock solution available, affordable or otherwise.  In fact, I’m currently in Bangalore canvassing developers on this very issue (see photo).

I posted a summary of all options known to me on Quora not long ago:
- Are there any software tools or APIs that will allow you to create a 3D talking head that will generate correct lip movements when given a text stream

Examples of currently available research avatar systems include:
- embots.dfki.de/EMBR
- marc.limsi.fr
- vhtoolkit.ict.usc.edu

However, I recently inquired about the USC Virtual Human Toolkit, and it starts at $10,000 for any non-academic user.