Blender-Compatible Motion Capture Library
Motion capture is a special animation technique that allows human actors to define the movements of an animated character. Special cameras track the actor’s movements and can then transfer them onto a CG character. The result is a simple yet realistic method of animating. Unfortunately, however, motion capture systems are very expensive, far too expensive for the hobbyist. That’s where this great resource comes in!
Great ressource, thanks!!
Amazing stuff!
good one, thanks for pointing it out! :) and thanks olivS!!
Very interesting article!
Id interested in seeing is some good tools for cleaning up the IPO curves, as well as tools for applying motioncapture armature to another armature since the rig of the BVH is rarely the same as one your using (this is known as re targeting)
of course constraints could be used, for example, if you wont to import 3 BVH files and join together as 1 animation, you could add 3 constraints to each bone and then key each one to blend between each armature influence but how easy is this?
It would be really worth refining this workflow.
@PlantPerson: well, more an information than a tutorial, to be honest…
@Ideasman42: hey Cam, I hoped you’d be around as there’s something that you could help with: “Edit: Import As Empties looks broken/buggy at the moment, with the script bundled in Blender 2.44.” perhaps you would be the man who could address this comment, prove it wrong or help to fix it? ;-) Or perhaps there’s something I missed about its usage… In any case i’d be glad to update the article ;-)
I had problems importing BVH as well. Importing as armature seems to work fine, importing as empties generates a python error:
Traceback (most recent call last):
file “”, line 735, in load_bvh_ui
file “”, line 357, in bvh_node_dict2objects
AttributeError: ‘module’ object has no attribute ‘IpoKeys’
There’s a program called Endorphin where you can use things such as behaviors, motion capture and physics for animation.
It was used in movies such as Posideon and if you remember that Pepsi ad where a guy flips all the way down the skyscraper like one of those glue toys!
I kind of have a love-hate relationship with motion capture. On one hand, the saved man hours in animating can make the system pay for itself a hundred times over for a significant production. Some examples of films that used only motion-capture data are “Final Fantasy: The Plot Without;” “Polar Express;” and the upcoming “Beowulf”
Notice anything that those have in common? Yeah, the results are extremely stiff and low quality in the animation side (but really good quality in the artwork- go figure)
Motion capture is good for base animation that can be used for reference and guides. It shines in scenes where you need to have a CG character interact with a live actor- assuming of course that you go back and seriously tweak the animation on it. Think gollum or Davy Jones; both examples of Mocap done right. (actors need something to interact with, which the stand-in does very well) So Mocap has its place, and can yield really good results if done correctly.
At Siggraph it was a running joke, because there were dozens of companies there trying to sell mocap systems. Most yielded equally stiff results. Sadly, I think that’s the direction the industry is going. I think part of the problem is that the actors in the suits tend to forget the very basic principles of animation: Exaggerate everything… Especially if you are aiming for realistic.
1. Design mocap system and display it at Siggraph.
2. ???
3. Profit!!!
@ToastBusters,
Agreed. for the most part. Mocap generally produces fairly accurate representations of a person’s movements. However people are not, for instance, super heroes. They don’t fly, jump from building to building, die (on command), or get blown away. For this reason among others, it will always be necessary for a talented animator to create drama from stock or custom motion capture data. Even regular life imitation, when it’s not superheroes or space marines, benefits from adding drama and exaggerating motion and expression.
Mocap, of its own accord, is incapable of producing quality results. At the very least, it’s dependent on the actor. It’s standard practice to put a mocap suit on just about anyone and have them stand on the mat while the cameras record their movements. The problem is that real-life is not a good substitute for art, but the quality of the data will rise with the quality of the acting. Strong, dramatic acting in a motion capture studio will produce better results, better data, but this still cannot stand on its own in most contexts.
For that reason I tend to think of the benefit of mocap as high, but limited. Motion capture alone will never be a good replacement for good animation. In my estimation, it almost puts cel animation on its head. In cel animation, the lead animator creates the keyframes and other animators filled out the tweens. In good mocap practice (IMHO) the data gives the tweens, and the animator can emphasise the “key” motions.
Mocap data provides a great filler for the drama and motion exaggerations that should be added to all animations. Subtlety of motion that produces realism is difficult to duplicate, and slow to animate, so mocap is a great way to get this subtlety and realism and save some real time on a big project. However the results will never be great if a talented animator doesn’t step in and do something with the data, and mold it to fit the mood and action of a scene.
Just my 2 cents.
Least-squares applied locally to the data to rebuild the splines would do the trick. Possibly with some backtracking to average overlapped results.
“You still may not be able to do your own motion capture, but mocap becomes affordable, this is the next-best thing. Don’t miss it!”
There is a typo on that that first sentence which is in the second paragraph. it should be “but until becomes affordable, this is the next-best thing”
@Apollos, fixed the bug with empties, will be in 2.45 release. I mainly included empty import because the first BVH importer I wrote could ONLY import empties, its kind of odd functionality and Id wonder if anyone would seriously use it.
@Ideasman42: I think that working with Empties can be useful if you already have a fully rigged character (like Ludwig or Mancandy, for example): then you can use the empties to control the existing armature. Anyone has tried that?
@Ideasman42, good to know thanks much, not just for the empties fix but for everything you guys do. Looking forward to 2.5!
@olivS, I haven’t tried rigging the empties yet. I’ve noticed a couple of tutorials here and there though. I’ll probably check it out in 2.5. At some point it would be great to produce a video tutorial on the process. I hope to do this eventually.
nice one link, thanks
Amazing.