I will be working with my friends on our final year project. Our game will be FPS and I have to draw some animations for FPS view and other enemy character that can be programmed easily to make a good game.

I would like to know how game characters are made that are movable? What kind of software and engines are used for these characters?

  • \$\begingroup\$ Is this a duplicate? If so I'll vote to close, but otherwise it's probably a good idea for this site to answer this basic question once. \$\endgroup\$ – jhocking Jun 8 '14 at 17:16
  • 1
    \$\begingroup\$ i searched about it first, then asked this question :-) \$\endgroup\$ – Ahmed Jun 8 '14 at 17:39
  • \$\begingroup\$ Do you mean 2D or 3D characters? Characters that have a skeleton or that are amorphous blobs? Are you looking for stand-alone results or integration with a physics engine? This could be a good question, but at the moment, a complete answer would fill multiple books. \$\endgroup\$ – Anko Jun 10 '14 at 9:43

Game characters are usually animated using a technique called skeletal animation: enter image description here (Image source: Valve Software)

Each 3d model has an invisible bone structure (the red and teal lines in the image above). Each polygon of the model is connected to a bone. When you define a motion sequence, you define it as a sequence of rotations of the bones around their connections to other bones. When a bone moves, the polygons connected to it move with it. This allows to define motion sequences without having to change the position of every single polygon. When multiple characters share the same bone structure, they can also share the same animations, so you don't have to recreate each animation for each character (although you might want to make some animations different for characters of different genders, personality types and levels of physical fitness).

Most 3d modeling software supports skeletal animation and have export-formats which are readable by widely used 3d engines. Please do not ask for product recommendations here, because they are off-topic.

However, letting an artist create animations from scratch doesn't always lead to the best results. It's hard to recreate all the subtle nuances of human body-language from memory, so the results often look robotic and unnatural. That's why larger game studios which have the resources for it use a process called Motion Capturing.

Human actors are hired to enact the actions of the game characters. The actors wear special suits during the performance which have highly visible marks attached to them. When they are recorded by a video camera during the enactment, the movement of these marks can be mapped to the bones of the character 3d models. That way the movements of the actors can then be applied to any 3d model which uses the same bone structure. This method usually leads to much more naturally-looking results and when there is a large number of animations to create, it can also be a huge time-saver. enter image description here (Image source: Toptear Games)

However, the cost for the equipment needed for motion capturing will likely far exceed your budget, so you will have to resort to defining skeletal animations manually in a 3d modeling program.

  • 1
    \$\begingroup\$ Good answer. But maybe you could mention that there are, in fact, low cost motion capturing techniques, utilizing computer cameras (webcams, Kinects, or PS Eyes). There are, also, services that sell reusable animations for common skeletal mesh setups. \$\endgroup\$ – T. C. Jun 8 '14 at 16:33
  • 1
    \$\begingroup\$ @TheodorosChatzigiannakis I am not very knowledgeable in either area. Maybe you would like to elaborate in an own answer? \$\endgroup\$ – Philipp Jun 8 '14 at 17:07
  • \$\begingroup\$ "letting an artist create animations from scratch doesn't always lead to the best results". Claiming that motion capture is inherently better than manual animation is like claiming that photography is inherently better than canvas painting. Both manual animation and mo-cap are different artistic techniques used for different purposes, and competent artists in each technique will provide excellent results. If your motion artists can't do good manual animation, then maybe you should get some better artists... \$\endgroup\$ – Panda Pajama Jun 9 '14 at 3:19
  • 2
    \$\begingroup\$ I've worked with a bunch of top notch 3D animators, and most of them prefer animating things manually over using mo-cap, as they have much larger control over the expressions in the animations they make, and they have much more freedom over what the characters can do compared to what a human body can do. However, when replicating a specific human movement is critical (for example, with AAA sports games featuring real players), mo-cap is quite necessary. \$\endgroup\$ – Panda Pajama Jun 9 '14 at 3:26
  • \$\begingroup\$ You could add that software recommendations can be asked in the software recommendations StackExchange. \$\endgroup\$ – Ramchandra Apte Jun 9 '14 at 14:18

Actually, the answer by Philipp is quite good, but misses an important point.

Some games (Quake/Q2 being the most notable examples) didn't use any skeletal animation at all - they instead resorted to vertex animation (ie. animating the position of vertices in the mesh). It's actually a lot simpler to support in code, although for big projects this usually don't hold true (but still, many games were made that way*.

For specifics, see e.g.


Also, this was the default way of defining animations in Flash, 3D Studio and 3DSmax in the days of old (although it stopped being that way after the implementation of skeletal animations for them). I think that for simple games, frame-based interpolation of vertices is still the easiest (in terms of both code and modelling effort) way to get started.

  • \$\begingroup\$ Indeed, but most engines (I realize this is a generic question) support skeletal animation natively, and at least in my opinion, they are quite simpler to do, after setting up the skeleton correctly \$\endgroup\$ – Kroltan Jun 8 '14 at 17:01
  • 1
    \$\begingroup\$ This isn't so much an either/or as much as a historical note. Games used to do this, but they don't anymore. As long as we're making historical notes, before game engines supported vertex animation they did hierarchy animation (think N64/PS1) \$\endgroup\$ – jhocking Jun 8 '14 at 17:20
  • \$\begingroup\$ Also describing this as "keyframe-based" is very misleading since skeletal animation uses keyframes too. I'll edit it to "vertex animation". \$\endgroup\$ – jhocking Jun 8 '14 at 17:21
  • \$\begingroup\$ @jhocking /agree, esp. with hierarchy animation; and yup, vertex animation is actually the term here. And yes, new games don't do it anymore. \$\endgroup\$ – user40973 Jun 8 '14 at 18:11
  • \$\begingroup\$ @Kroltan As I said, for simple games without actual skeletal animation engine it can be the best option. I've hand-coded both vertex animation and skeletal animation for my physics project in OpenGL couple of years ago - and believe me, I was able to code vertex anims in about one hour, while skeletals took me couple of days. If you use a dedicated engine (e.g. Ogre) then skeletals are the obvious way to go. \$\endgroup\$ – user40973 Jun 8 '14 at 18:13

Philipp's answer is correct, but it's worth adding that there are (relatively) low cost markerless motion capturing solutions that utilize sensors commonly found at home, such as an array of webcams, Microsoft Kinect sensors or Sony PS Eyes. They usually don't output production-quality animations (you can expect mediocre precision and jitter in most cases), but you'll have the timing right and if you're willing to do some cleanup and some fine-tuning, it'll save you a lot of time compared to manually animating the skeleton (or the mesh directly).

Additionally, there's Mixamo, a service that sells ready-made animations for common skeletal meshes. If I recall correctly, they give an amount of animations for free, for evaluation purposes.

  • 1
    \$\begingroup\$ I would like to add that a University MOCAP Lab here has actually replaced all their expensive (a few year old) equipment with a markerless kinect based solution, which gave better results. Of course markerless cannot compete with top-tear hardware. But a Kinect is pretty damn good (especially if we take price into account) for these kinds of things. \$\endgroup\$ – Roy T. Jun 8 '14 at 20:25

Not the answer you're looking for? Browse other questions tagged or ask your own question.