Meeting another person is one of the most amazing experiences you can have in Virtual Reality. It is quite unlike communicating through any other medium except a real life face-to-face conversation. Because the other person is life size and shares a virtual space with you, body language works in a way that cannot be done on a flat screen. This course will enable you to create realistic social interactions in VR. You will learn about both the psychology of social interaction and the practical skills to implement it in Unity3D. We will take you through the basics of 3D character animation and how to create body language. You will learn about how to make characters that can respond to players’ speech and body language. You will also learn about avatars: the virtual representation of other players, and agents: computer controlled NPC characters and how to implement both of them. As many people have said before us, social is the future of VR. This course will help you become part of the future of Virtual Reality social experiences.
Introduction to Skeletal Animation
Loading...
审阅
4.2（33 个评分）
- 5 stars20 ratings
- 4 stars6 ratings
- 3 stars3 ratings
- 2 stars2 ratings
- 1 star2 ratings
从本节课中
Body Animation in VR
In this module you will learn about the role of the body in social interaction and also about how to animate and control virtual characters' body movements.
教学方
Dr Sylvia Xueni Pan
Lecturer, Department of Computing
Dr Marco Gillies
Senior Lecturer
[MUSIC]
The human body is a single complex object, but
one whose individual parts move in different ways.
My hands and head are all part of my body and tightly linked, but
I can move them all pretty independently.
Like most 3D graphics objects, human bodies are represented as a polygon mesh.
But unlike other object it isn't enough to simply move the whole thing around with
a single transform.
We need to move our limbs.
We could create a compound object with separate objects for our hands, lower arm,
and upper arm, and this is what the early computer games and
computer graphics did, but it isn't very realistic.
My lower arm and
upper arm are not separate objects, even though they move relative to each other.
They form a single smooth surface.
So now a days we do something a bit more clever.
While the body is complex, and
it's outward appearance has many features, the majority of the movement
actually comes down to the movement of the skeleton.
The skeleton consists of a number of rigid bones
linked together by rotational joints.
When we move, our joints rotate.
And so rotate our bones relative to each other.
In 3D character animation,
we have a single, probably very complex mesh to represent the body.
But we also separately represent the skeleton.
The skeleton is a series of rigid links with bones,
with rotational transforms, the joints.
We animate the character by rotating those joints.
These rotations are then mapped onto the mesh using a process called smooth
skinning.
Which is done automatically in the GPU.
This results in a realistically moving mesh.
But we don't have to think of the individual movements of all the vertices,
we only have to animate the joints.
This makes our animation data relatively simple.
It's just the number of key frames to the rotations of the joint.
Setting up this kind of skeleton animation can be quite complex as we need to create
a realistic mesh and also a skeleton, which consists of a number of transforms.
These transforms are hierarchical.
That means that each transform has transforms attached as children and
these children themselves have children.
So my shoulder has my upper arm as a child, which has my lower arm,
which has my hand, my hand is very complex with a child for each finger.
This skeleton needs to be attached to the mesh, which means that we need to
say which bone each vertex is attached to, a process called rigging.
To achieve smooth movement, vertices can be attached to several bones.
This whole process is pretty hard work.
Most animation packages will partially automate it, but
it rarely works perfectly.
Luckily, as usual, the asset store comes to our rescue and
contains a number of excellent pre-rigged characters.
Skeletal animation works in a process called forward kinematics or FK.
I rotate a joint, for example, my shoulder.
That moves my whole arm.
I then rotate my elbow, which moves my lower arm and
my wrist which moves my hand.
The position of my hand is worked out by starting at the top joint,
my shoulder, and applying its rotation and then going down
each joint in turn applying that rotation, until we reach the hand itself.
FK is just handled automatically by the way transforms work.
And you really don't have to think about it at all.
The only reason I mentioned it is because we need to know about its opposite,
inverse kinematics, or IK.
Forward kinematics can be quite complex to animate with.
Often we just want the hand to be in a particular position,
maybe to pick something up.
Doing that with FFK is very fiddly.
I will take my shoulder to where I think it should be.
Then my elbow.
Then my wrist.
It's kind of there but not quite or it doesn't look right.
So it has to go back to tweak my shoulder and it all continues.
That's why inverse kinematics was invented.
IK is an algorithm which takes the position where you want the hand, and
does a load of calculations to work out what
the rotations of all the joints should be, to get your hand in the right position.
It's very useful for animation, and it's included as standard in
most 3D animation packages, like Myer, 3D Studio, Max or Blender.
But it's also useful for doing real time interaction.
For example, getting characters to pick up objects.
Or even doing a fist bump with you.
It isn't just for arms either.
We'll see in a future video how we can use it to get cowards to look at things.
Skeletal animation is a very powerful and flexible technique for
human body animation.
And we'll see in future videos that it underlies all the more complex things that
we'll want to do with body language.
[MUSIC]
