How do developers get animations incredibly fluid? Like Platinum Games for example, in Nier Automata the animations can be breathtaking at times. Is it only motion capture or do they use that in tandem with something else? Like let's say that a character has centipede like tendrils protruding from his back and most of his moves involve flips and feints of the tendrils in tandem with hand to hand combat. How does that work?
Let me try to answer your question in a slightly roundabout way. Let’s talk about motion capture for a moment. A lot of gamers seem to think that, for video games, animators will hire actors to wear the motion capture suits, record some data, and then just plug that data into the game and everything is great. This is not how it works. Mocap is only the beginning, not the entire process.
If you’ve read my [Animation Primer], you know that animation data is generally represented as a sequence of specific positions over time. Each of these sets of positions is called a frame. The character is placed in these positions in order. The speed at which the character changes position to match the frame is called the frame rate. However, frame rate isn’t just used for playback. Computers run things in discrete time steps, after all, and computers capture motion data. When we record motion capture, we record it at a certain number of frames per second too. Most motion capture data is recorded at much higher frame rates than we expect games to display - typically around 120 to 160 frames per second. Just consider - if we have 120 to 160 frames per second of animation data, how do we choose which frames to display for the game that displays at 60 frames per second?
We could do it the basic way - just show every other frame we recorded. 120 fps becomes 60 fps this way, and everything is played back at the same speed it was captured. But that’s assuming that the data we motion captured is enough, and assuming that real life motion is what we’re after. If we’re talking about a game like Nier Automata, for example, where the main character is an android with greater capabilities than a human actor who’s goal isn’t to actually hurt anyone during motion capture settings, things change. So who decides what frames of animation get put into the actual final product? This is where the animator comes in.
A lot of dealing with motion capture data is “cleaning it up” - chopping off the bits that lead to and from the reference pose, making it run to the target frame rate, and choosing frame data that looks right for the game experience. These changes can be subtle or they might be large. Sometimes you want a 1:1 rate with the original mocap data, and sometimes you want to slow something down, or speed something up so that it reads better or gives a different impression of the motion. There are even times where animators will adjust certain parts to be faster while others are slower. Look at the differences in the above animations - they might seem small at first glance, but the edited frame has a lot more emphasis of the action to the viewer. The raw mocap looks more like the attacker is simply walking up from behind and pushing the victim. While this might work in real life, it doesn’t read as well in a video game. In the edited animation, the attacker has some anticipation built in before the stab to make it look better. Look at the added pull back and then stab forward with the left hand and how the attacker grabs and maintains his hold on the victim’s mouth in the edited animation. See how the overall time is about the same, but some parts were sped up in order to make room for the added anticipation movements?
Motion capture suits also don’t have the extra bells and whistles of in-game costumes. Any extra bits like tendrils, robot arms, wings, tail, or clothing has to be animated separately. That typically means either some sort of simulation (usually via physics or some sort of procedural system), or hand-animating the extra bits. Some stuff lends itself more easily to procedural motion, like clothing. Physical features like tentacles, antennae, wings, or tails tend to be hand-animated, and that’s all about the animator’s skill at making legible motion.
The real answer to your question is that fluid-looking animation isn’t so much a question of the frame rate as it is a testament to animator skill. Animation looks fluid because the animators have carefully selected the right frames to string together to best convey the motion you’re seeing. This is the major difference between really good-looking animations and passable “ok” animations. It’s not a question of how many frames of animation there are, it’s what the animator does with those frames.
Got a burning question you want answered?
- Short questions: Ask a Game Dev on Twitter
- Long questions: Ask a Game Dev on Tumblr
- Frequent questions: The FAQ