So, I am now animating character portraits for dialog in the game.
It is done in portraits as in classic games to reduce render time when experimenting, but it is yet unsure if that is how it will end up. The idea is to have full screen animation, but it might be simply too much to animate with body movement and gesturing.
It all comes down to render time:
1280x720 to 1920x1080 means more render time.
15 fps to 30fps means double render time.
Bigger picture means more detail and more render time.
Right now a single frame takes approx. 2 hours to render.
And now I have two i7 computers rendering 24/7.
Facial animation is done by having a pose library of mouth poses in Blender, and matching each pose to a phonetic sound in the speech file from the actor. Then the eyebrows, bliking and head movement is added from a filmed actor mimicing the spoken lines with facial emotion. It is really dirty, but it seems to work when put together.
Now I have only one pose for each phonetic sound, and due to the fact that I'm using bones only, it is not accurate. In cases where the pose does not match very well with the power of the speaking, I blend poses on a key frame to bring down the intensity a bit.
The video below explains how complex this really is:
It is easy to see that this is very expensive, and since I do all the visual work on the game I have to make a trade off with the quality.
Videos and pictures will come soon!
