Facial animation L4D2

Hi, I am trying to make a model for left 4 dead 2 and i want to incorporate the survivors flex files that i got whenever I decompiled the models in crowbar. After having got the position of the eyes from qc_eyes and compiling it for use in hlmv, my model keeps looking like this.


Why is it doing that?

That’s because the decompiled VTA file doesn’t fit on your custom model. You need to re-create it from scratch, or give your model a head that already has a VTA file premade for it.

This. Source doesn’t use bones for facial animation like most games do. Instead, it uses morph targets saved in the VTA file. In other words, the VTA file stores the vertex locations in each “frame” – vertex animation instead of bone animation. The engine then handles this by blending cleanly between the different created flexes. Since this is vertex animation, you’re going to need to create your own VTA file by creating the different expressions listed here. If you still don’t understand why it’s not working – The vta file stores vertex animation which comes from a different model. Since the verts on the new head are different, it doesn’t know to move the eyebrows, mouth, or whatever, so it just moves random vertices that are listed or may simply not work at all.

Yeah I agree, you need to create your own facial flexes that fit that character. L4D originally uses the DMX format for storing facial flexes which have the advantage over vta that you dont need to do any qc scripting for the flex controls. However it really depends on what authoring package you are using since if your using 3ds max it doesnt support exporting to dmx. I suggest you take a look at the source files for the left for dead characters provided in the sdk_content folder for l4d2.


No but you do have a tool in SFM’s bin folder to convert fbx to dmx.

I’ve ran into a lot of issues using that thing so not worth me mentioning it.

Yeah, if I could only choose one person to ask about anything related to Source and its use of the DMX model format, it’d be cra0kalo.

The knowledge he has on the format is kinda scary, and the research and experiments he’s done on its usage in Source is terrifying.

No offense intended, Jojje.

Of course.
I’m nothing but a novice in the field of modeling.

Ok guys thanks for your help. It was driving me mad.