Potential of OpenGL

Hi guys, I am a pretty new member on this community. I need to create an interactive talking head (using AI), which can have a variety of facial expressions to interact with the user. I would need something that can be integrated with Java and would work on Windows. Is Open GL my solution? I have the AI part ready with me. So all I need is the front end and connectivity with my backend program.

Yes OpenGL can render such things and there are Java bindings but your biggest problem will be to get the data for that head.

You need some kind of modelling tool like Blender to design and export the animated head and then write code (or use a library) to import that data in your application. Do you have modelling skills? If not, this is a tough task and you maybe want to check out existing modells and probably buy one.

[QUOTE=Betrayal;1265025]Yes OpenGL can render such things and there are Java bindings but your biggest problem will be to get the data for that head.

You need some kind of modelling tool like Blender to design and export the animated head and then write code (or use a library) to import that data in your application. Do you have modelling skills? If not, this is a tough task and you maybe want to check out existing modells and probably buy one.[/QUOTE]

How difficult do you think it would be to learn to use Blender (considering that I have literally no skills/knowledge about designing)

Difficult to say, i’m new into Blender for myself and just know the very basics. An animated head sounds complicated to me and probably needs a good amount of time to learn that Tool.

The difficulty isn’t in learning to use software, it’s in learning to model. In particular, learning to model humans requires some knowledge of anatomy.

Find an existing model or hire a modeller to create one for you.

The biggest issue is likely to be that basic animation techniques such as skeletal animation aren’t particularly suited to lip-synch, and the data formats for more advanced techniques tend to be complex and/or proprietary. So you’ll probably end up having to export a separate version of the mesh for each expression and interpolating between them.