[Bf-committers] 3D character engine

[Tom M]

> A brief overview of our requirements:
> -Able to render high quality human and non-human characters in real-time
While Blender GE might be able to do this, when the OGRE 3D
integration is complete that might be a better option.

> -Able to display a wide variety of emotional facial expressions (happy,
> surprised, sad, angry, etc.)
> -Characters should be full-body characters
> -Character is able to act out certain pre-set animations on command (yawn,
> dance, clap, etc)

That would all be based on the person who created your animation.

> -Able to lip synch with SAPI speech synthesis engines given real-time
> phoneme data from the speech engine

That would need to be programmed

> -Character should be in a 3D environment (a room or outdoors), but focus
> is on character not environment

Again doable, but likely better once the Ogre3D integration is completed.

> -User should be able to control camera to look around character, and zoom
> in and out

trivial

> -A fairly easy and documented way to create compatible characters for this
> engine to use

You might look at integrating the work done by the makehuman team
(parametric variation of a human body similar to poser).

> -Must be better or equal to quality of Zabaware's current character engine
> (Haptek)

Haptek is extremely good for the task you are doing.  Using the
Makehuman model and using lattice deforms inside of Blender could
fairly easily replicate faceputty.  Alternatively the makehuman
standalone can be used similarly to Poser.  However asside from the
body and body morphtargets there isn't content such as clothing and
props.  You cloud use either the standalone player or could update the
Blender Plugin system with or without the Ogre3D integration.  The
cost of developing Blender for that target purpose could possibly
exceed your licensing costs for the Haptek player.

> -Is this feasible?
A definite maybe :)  It really is dependent on more details.

> -Is the game engine a GNU GPL library license? We realize that these
> improvements to the engine will continue to be open-source and freely
> licensed, but we would not like to open our ai engine. Can we use the
> engine without it forcing us to open the rest of our software?

The engine itself is GPLed.  The Blend file you create to use with it
is your own and can be licensed as you see fit.  You should be able to
use the python API for communication between your closed source
software and the Blender GE.

Personally I would probably go the route of 1) use makehuman model and
morphs and Blender for allowing your users to do creating/modify of
the characters.  (If you use Blender then improving and changing
textures is easier, users can sculpt changes, or use lattice deforms;
if makehuman only then poser style slide based changes can be done) 2)
then export to Ogre3D format and have Ogre3D as your display engine.
Since Ogre3D is LGPL you can use their C API to communicate with your
AI and speech software.

> -How much work would be involved to meet these requirements in terms of
> man hours? Are there interested developers willing to take on this
> project? If so, what is your hourly rate?

It will probably take a lot more information to determine a man hours
estimate.  However others can speak with better informed opinion on
this aspect.

Tom M.
LetterRip