I posted in the animations support thread about this, but I want the BGE community feed back for it.
Why isn’t Mocap more popular for BGE? Or if it is used often, people don’t talk about it. Does it not work with BGE? I thought Mocap was just making animations, which could just be played like any other animation.
Are there any well known resources around to use for learning how to use the motion capture in blender/with kinect? I looked on the google and all the videos and such I could find are ether 3 years old, or for a different program (as if there is any life besides the blender life ).
It might be easier though, to do an elaborate fight scene or something, then just correct it. Rather than trying to do it all by hand at least.
I am hesitant to the tango because I want to start using it sooner rather than later, plus I already have a kinect and wouldn’t need to spend the monies on a tango tablet/dev kit. Are you able to share any info on how to use said kinect with blender for mocap?
Motionbuilder is the software of choice for working with motion capture however, once you get good data, you should be able to plug it right in and create transitions from one action to another. I think the reason people don’t use it much with Blender is because it’s for pretty high end game development. The cost of good motion capture data can be pretty extreme. Blender has such good keyframe animation tools that you really could do a lot of the work yourself and save a bunch of money.
Oh… I wanted to mention, Motionbuilder works so well with that kind of data because you can layer animation. So you get mocap with a key on every frame, but then you can add a layer, place anchor keys and adjust anything on the mocap and it looks quite natural. If Blender had a way to add keyframe layers like that it could be really useful when working with mocap.
What we discuss on the kickstarter forum is twofold:
(1) using mocap data for keyframed renders into Blender to achieve normal animation and VFX, that can be achieved through mapping the BVH to armatures of the models in Blender, I saw the tut from Tomwalks, he referenced Blender docs on BVH and I found none on the wiki. Do you have a good reference on how BVH works, not just the few tuts out there who only seem to have a test case nature to me?
(2) using the mocap data real time. The equipment is planned to deliver a real time BVH stream in ASCII through Wi-fi and there will be an SDK etc (eg for Unity). But I thought how about using the BGE to interact and use the BVH as input data into the BGE for both simulating the armature it drives and the interaction in the BGE scene (eg a human hand joystick). But all the documents on the BGE I found were on it internally - do you have a pointer on how the BGE could accept external data in a real time stream (eg. through a data socket).
I will search the forums and if I do not find anything I will go to the devs, but before I make a fool out of myself, I thought I ask around a bit.
The camera you see is connected to another computer and not the one you see, so the data is actually being streamed over LAN. This led to the patch I wrote that enabled unlimited action layers :RocknRoll:
I have been researching MoCap for the last weeks. I do choreography and in the past have created a 1 hour animated film so I want to put the two together. I am also currently negotiating with major studios and acts to licence hit music for the dance based music videos. I am writing this from memory so apologies for errors… There are systems that cost $100,000 which is outside my budget. Phasespace have a system at about £25,000 which needs a large room and many cameras. www.motionshadow.com have an inertial sensor system that you strap on and doesn’t use cameras and so can be used anywhere, cheaper at $12,500 but not cheap. www.ipisoft.com have software that uses two kinnect or ASUS sensors or multipl sony webcam type cameras plus their software which totals less than $1,000.
I have seen demo video and have seen example MoCap files, but I haven’t yet used any of these systems. My plan is to buy the cheapest and if not good enough cry as I write the cheque for one of the others. I do not recommend and have no connection with any of the suppliers.
If you uad something like the perception nueron,
And you could get the data into python, I can do the rest.
Just have cubes marked invisable, and set there rotation based on the perceptive neuron rotations, then you armature just needs copy rotation to each cube/joint.
).
