Sign in to follow this  
  • entries
    2
  • comments
    8
  • views
    300

The inelegant process of implementing blend shape motion capture facial animation into Leadwerks

brndx

436 views

blog-0502497001488979347.png

In an effort to further boost production values and the general quality of the cinematics in Enshrouded World I implemented facial animation for the characters and as the title suggests rather inelegantly.

 

 

 

 

Below is one way of going about implementing facial animation but it's most likely not the best.

 

 

 

- Sift through thousands of frames of mocap facial animation and export the model every five frames.

- Load and release the vertex data of each of the hundreds of models needed for one cutscene to reduce RAM usage.

- Set the positions of the vertices in the base face to the positions of the vertices in the next model (This requires interpolation and continues to the next model every 83 milliseconds).

 

 

As you would expect the first step there was very tedious and despite my efforts there are still issues with the framerate whenever the facial animation is being rendered.



5 Comments


Recommended Comments

Jeez, going through all that animation must have taken forever, did you automate that process or do that manually? Also have you got any video of this animating? I'd be interested to see how well it works.

I can see why there's framerate issues, sound's like a lot of processing to do on the CPU.

Share this comment


Link to comment

It was done manually unfortunately. You can see an example of the animation here

and

here

.

I wasn't really confident enough in my ability to write a shader for it so I had to just use the surface class in the api and rely entirely on the cpu.

Share this comment


Link to comment

Hey I have no idea what about the game should be, but yes I find the videos cool, can imagine hours work cost to make this, great atmosphere, keep going on

Share this comment


Link to comment

What I've seen is that for games, they use a FACIAL RIG. It all composed of bones. You would need then 2 variant of your model, one with the face rig and another without it. (Also seen old games, had a model only of the head that was parented to the neckbone of the body, because they could not handle all the bones per model) Here is a video showing the bone setup (I find this quite complex but Leadwerks can handle up to 256 bones per model).

 

It would be nice that we could take a tool like FACE RIG (

) and map it to a generic model... It would save tons of time. Opps! Checked again and they give a way to use your own model with it, but it doesnt output... too bad.

Share this comment


Link to comment

What I've seen is that for games, they use a FACIAL RIG. It all composed of bones. You would need then 2 variant of your model, one with the face rig and another without it. (Also seen old games, had a model only of the head that was parented to the neckbone of the body, because they could not handle all the bones per model) Here is a video showing the bone setup (I find this quite complex but Leadwerks can handle up to 256 bones per model).

 

It would be nice that we could take a tool like FACE RIG (

) and map it to a generic model... It would save tons of time. Opps! Checked again and they give a way to use your own model with it, but it doesnt output... too bad.

 

The software I used to capture the motion data (http://facewaretech.com/products/software/realtime-live/) is very similar to facerig.

Share this comment


Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Create Your Account

Sign in

Already have an account? Sign in here.

Sign In Now