First Person Animation

To follow up the article about character design I thought I would give a brief overview of how we go about rigging and creating animation for our characters.

Rigging

We have two separate rigs for our in-view (read: first-person view) purposes: The actual high res arms with weapon models and then the lower res body for interaction animations that require us to see Renko’s legs. This also allows us to exclude the legs when we do not use them, freeing up memory for more polygons, textures, and bones in the arms and weapons you see all the time.

In addition to these rigs (which are built in Maya) we have various controls on the arms in the engine to allow us to override animation or add things to animation programmatically. This is great for things like camera shakes or weapon recoil which are tedious to do by hand, and this also allows us to add more variety without disrupting our animation budget.

Animation

First-person, in-view animation is an interesting topic because you would never actually see your arms the way you do in this view. Your arms, anatomically speaking, would have to be bolted to your face to actually see what you see in the game, which generally precludes capturing the animation from a motion actor in our Mo-Cap studio. Nonetheless, when we animate the in-view weapons we still try to make it feel natural, and we use the framing of the player view to guide this process.

Animations fall into two broad categories: Weapon animation and cinematic animation. Weapon animation consists of animation required for player movement and action, things like firing, reloading, jumping, and basically moving around. Cinematic animation covers most other environmental interactions, like being knocked to the floor after an explosion and then getting back up. Both need to be animated in different ways, especially if the legs are involved.

Weapon animation is pretty straight forward. We try to immerse the player without being distracting; weapon animations are all about player control. We work a lot with the creative and game-play leads to make sure every animation not only looks good but does not get in the way of player control. We are constantly tuning the timing of how long things take to match the pace of game-play. I cannot count how many times we adjusted every single one of the weapon animations over the course of the project.

Interestingly, every weapon animation takes place faster that most humans could actually perform those tasks. Reloading in particular is crazy fast. I did some research on AK-74 reloads (this is the weapon closest to the fictional machinegun we created), and found that quickest experts take around 1.5 to 2 seconds total to reload, assuming they had emptied the magazine. In the game almost every weapon has a base reload time of 1 to 1.5 seconds and with a perk you can get that down to 0.5 to 0.75 seconds!

Cinematic animation is completely different. Responsiveness is not what we are concerned about here. Whenever these animations are made we generally take control away from the player for a short period of time, like knocking them to the floor from an explosion. We are trying to create an immersive, visceral experience for the player. They are generally more complicated than the weapon animations for several reasons: In an average weapon animation we only have to worry about the weapon and player hands, but in a typical cinematic animation we frequently have to animate those and the players legs AND any interactions with the environment or creatures.

For example, we at one time had a sequence where a helicopter blew up a bridge the player was on with some soldiers and a truck. The player gets knocked off his feet by the explosion. He then slides to the edge of the broken bridge and dangles off the edge, watching the soldiers and the truck plummet to the bottom of the gorge before pulling himself up. This involves many separate elements:

1. The first-person arms
2. The player legs dangling
3. Soldiers
4. The bridge
5. A helicopter
6. A truck
7. An exploding bridge

With a rig and animation for each element, this requires a very good plan for how the whole sequence should work.

The animator (Mark Champigny, in this case) would take all these elements and arrange them, preparing them for hand-keying. Here is a typical workflow for this type of scene:

1. Take all the elements and put them in one big Maya scene.
2. Block in the initial explosion animation, how big it will be, and how everything will be destroyed.
3. Block in the first-person arms animation based on decisions made while blocking in the explosion.
4. Firm up the interaction of the first person arms with the bridge pieces. (like what piece of wood or metal our hero hangs onto while dangling)
5. Start blocking in the animations for the soldiers and truck based on what the player can see.
6. Add in the player’s legs animation for when he is dangling.
7. Now that all the major decisions are made the animator will push all the animations to completion.

Initial Scene set-up, including the world geometry and truck rig:

Explosion from the players camera view:

Hanging on to the edge from the player view, this shows all the rigs in view, bridge, soldier, inview arms and player legs:

Same shot from a perspective view camera, note how far apart the player legs and arms are; it looks ridiculous in this view but from the player cam you can’t even tell:

After that we export it and work with a designer to make sure it is implemented successfully and that the motion lines up properly. There you have it, a very rough overview of how we accomplish first-person animations for Singularity. Hope you enjoy the game!

David Gulisano
Senior Animator
Raven Software