CrazyTalk allows for pretty-much real time lip-synch and expressions on 2d images based on vectors as shown. Now, these vectors can pretty much be assigned to face features on 2d baked maps of avatars in 3d software, s.t. they appear to move their lips in-synch and emote based on their voice. Here is a simple example of how I spent 1 minute and 37 seconds configuring a CrazyTalk mesh by hand for one of my fave SL av's to make a quick weekend mini-production entry to a shopping confession contest. (But, I really think that this can be embedded directly onto SL avatar faces' baked texture maps in RT.)
And here's the final cut:
CrazyTalk allows for pretty-much real time lip-synch and expressions on 2d images based on vectors as shown. Now, these vectors can pretty much be assigned to face features on 2d baked maps of avatars in 3d software, s.t. they appear to move their lips in-synch and emote based on their voice. Here is a simple example of how I spent 1 minute and 37 seconds configuring a CrazyTalk mesh by hand for one of my fave SL av's to make a quick weekend mini-production entry to a shopping confession contest. (But, I really think that this can be embedded directly onto SL avatar faces' baked texture maps in RT.)
And here's the final cut: