• Advertisement
Sign in to follow this  

AI Character behaving according to real human acting

This topic is 1394 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

Hi guys,

 

How can I make AI character act exactly like a human actor?

 

Is there is a way to record the real human acting such as face expression and body movement and get AI character to act exactly the same?

Share this post


Link to post
Share on other sites
Advertisement

SOE/LA recorded live dancers for the dances for the Entertainer class in Star Wars Galaxies, so it can be done.

SOEmote does something similar to what you're asking, I think.

 

Can you do it easily, without spending a fortune on recording equipment? How good are you with Poser, I suppose.

Share this post


Link to post
Share on other sites

You can buy fairly cheap 3D cameras, the USB type. The hardware I used isn't available anymore. The company was bought by Microsoft and the technology incorporated into kinect.

 

These return a 2d array of distances, so you can use it to record facial expressions. 

 

Turning these frames into something you can render quickly is a different issue, you would have to do some serious coding.

 

Another approach would be to get a standard head mesh, there are some free ones on the net somewhere. You can get mouth shapes from them and use some kind of text to speech system to animate the mouth then mix in emotions based on observation of the actors.

 

Have a look at this to give you an idea of what I mean

 

http://stainlessbeer.weebly.com/facial-animation.html

Share this post


Link to post
Share on other sites

I think some sort of this technique was used in some movies such as Avatar, do you know what techniques they used?

[attachment=20681:Avatar3.jpg]

Share this post


Link to post
Share on other sites

If you want a dynamic mesh, 4d photogrammetry / videogrammetry

 

 

Otherwise, you create a facial rig using a huge number of bones and blend-shapes, and then use an algorithm that can look at the dynamic mesh produced from above and automagically choose the best weights for your bones/blends that will recreate the same mesh pose.

Edited by Hodgman

Share this post


Link to post
Share on other sites
Sign in to follow this  

  • Advertisement