Moving Objects. We need to move our objects in 3D space.

Transcription

1 Transformations

2 Moving Objects We need to move our objects in 3D space.

3 Moving Objects We need to move our objects in 3D space. An object/model (box, car, building, character,... ) is defined in one position (often centered around origo). Will be needed in another position in the scene.

4 Moving Objects We need to move our objects in 3D space. An object/model (box, car, building, character,... ) is defined in one position (often centered around origo). Will be needed in another position in the scene. Maybe in several places in one scene (town with houses and cars).

5 Moving Objects We need to move our objects in 3D space. An object/model (box, car, building, character,... ) is defined in one position (often centered around origo). Will be needed in another position in the scene. Maybe in several places in one scene (town with houses and cars). Maybe in different places in different scenes/frames (animation).

6 Moving Objects We need to move our objects in 3D space. An object/model (box, car, building, character,... ) is defined in one position (often centered around origo). Will be needed in another position in the scene. Maybe in several places in one scene (town with houses and cars). Maybe in different places in different scenes/frames (animation).

7 Moving Objects We need to move our objects in 3D space. An object/model (box, car, building, character,... ) is defined in one position (often centered around origo). Will be needed in another position in the scene. Maybe in several places in one scene (town with houses and cars). Maybe in different places in different scenes/frames (animation). Move model move triangles move points (vertices) f : R 3 R 3

8 Translation

9 Translation x x + p f y = y + q z z + r

10 Scaling

11 Scaling x u x f y = v y z w z

12 Rotation

13 Rotation Rotation around line through origin:

14 Rotation Rotation around line through origin: x? f y =? z?

15 Rotation Simpler case: Rotation around z-axis.

16 Rotation Simpler case: Rotation around z-axis. From formula for rotation in 2D (known from high school): x x cos φ y sin φ f y = x sin φ + y cos φ z z

17 Rotation Similar: Rotation around x-axis and y-axis. x x f y = y cos φ z sin φ z y sin φ + z cos φ f x z sin φ + x cos φ y = y z z cos φ x sin φ

18 Euler Theorem (Euler, 1775): any rotation with axis through origo can be created as three succesive rotations around the three coordinate axes. The angles of the three coordinate axis rotations are called Euler angles. Using Euler angles to specify generic rotations is often intuitive, but also has drawbacks. We will return to that later.

19 Matrices Move model move triangles move points (vertices) f : R 3 R 3

20 Matrices Move model move triangles move points (vertices) f : R 3 R 3 Any matrix induces a (linear) funktion f : R 3 R 3 : x x 1x + 2y + 3z f y = y = 4x + 5y + 6z z z 7x + 8y + 9z

21 Matrices Move model move triangles move points (vertices) f : R 3 R 3 Any matrix induces a (linear) funktion f : R 3 R 3 : x x 1x + 2y + 3z f y = y = 4x + 5y + 6z z z 7x + 8y + 9z Recall: Matrix multiplication is associative: A (B C) = (A B) C.

22 Matrices Move model move triangles move points (vertices) f : R 3 R 3 Any matrix induces a (linear) funktion f : R 3 R 3 : x x 1x + 2y + 3z f y = y = 4x + 5y + 6z z z 7x + 8y + 9z Recall: Matrix multiplication is associative: A (B C) = (A B) C. Hence: x x A (B (C (E (F y )))) = ((((A B) C) E) F ) y z z

23 Matrices Move model move triangles move points (vertices) f : R 3 R 3 Any matrix induces a (linear) funktion f : R 3 R 3 : x x 1x + 2y + 3z f y = y = 4x + 5y + 6z z z 7x + 8y + 9z Recall: Matrix multiplication is associative: A (B C) = (A B) C. Hence: x x A (B (C (E (F y )))) = ((((A B) C) E) F ) y z z Saves calculations: 3D object = many triangles = many points. All points go through the same sequence of transformations (moves). Calculate the matrix product once.

24 Matrices Move model move triangles move points (vertices) f : R 3 R 3 Any matrix induces a (linear) funktion f : R 3 R 3 : x x 1x + 2y + 3z f y = y = 4x + 5y + 6z z z 7x + 8y + 9z Recall: Matrix multiplication is associative: A (B C) = (A B) C. Hence: x x A (B (C (E (F y )))) = ((((A B) C) E) F ) y z z Saves calculations: 3D object = many triangles = many points. All points go through the same sequence of transformations (moves). Calculate the matrix product once. Question: can all our needed transformations be expressed as matrices?

25 Transformations as Matrices

26 Transformations as Matrices Scaling x s 1 x s x f y = s 2 y = 0 s 2 0 y z s 3 z 0 0 s 3 z

27 Transformations as Matrices Scaling x s 1 x s x f y = s 2 y = 0 s 2 0 y z s 3 z 0 0 s 3 z Rotation angle φ around the z-axis x x cos φ y sin φ cos φ sin φ 0 x f y = x sin φ + y cos φ = sin φ cos φ 0 y z z z

28 Transformations as Matrices Scaling x s 1 x s x f y = s 2 y = 0 s 2 0 y z s 3 z 0 0 s 3 z Rotation angle φ around the z-axis x x cos φ y sin φ cos φ sin φ 0 x f y = x sin φ + y cos φ = sin φ cos φ 0 y z z z Translation? x x + x 0??? x f y = y + y 0 =??? y z z + z 0??? z

29 Transformations as Matrices Scaling x s 1 x s x f y = s 2 y = 0 s 2 0 y z s 3 z 0 0 s 3 z Rotation angle φ around the z-axis x x cos φ y sin φ cos φ sin φ 0 x f y = x sin φ + y cos φ = sin φ cos φ 0 y z z z Translation? x x + x 0??? x f y = y + y 0 =??? y z z + z 0??? z No. Translation is not linear: f ( x 1 + x 2 ) f ( x 1 ) + f ( x 2 ).

30 Homogeneous Coordinates Go to 4D: x x y y z z 1

31 Homogeneous Coordinates Go to 4D: And back: x x y y z z 1 x y x/w z y/w z/w w

32 Homogeneous Coordinates Translations (in 3D) can now be expressed as matrix multiplication: x 0 x x + x y z 0 y z = y + y 0 z + z

33 Homogeneous Coordinates Translations (in 3D) can now be expressed as matrix multiplication: x 0 x x + x y z 0 y z = y + y 0 z + z All 3x3 matrices are still available (incl. skaling and rotation): x 1x + 2y + 3z y z = 4x + 5y + 6z 7x + 8y + 9z

34 Projection Projection to screen: f : R 3 R 2.

35 Projection Projection to screen: f : R 3 R 2. Prespective projection:

36 Projection Projection to screen: f : R 3 R 2. Prespective projection: Expressed as 4x4 matrix multiplication (d = near): x x y z = y xd/z z yd/z d 0 0 1/d 0 1 z/d

37 Transformations in OpenGL OpenGL uses 4x4-matrices/homogeneous coordinates internally. Matrices are normally created by more intuitive commands: gltranslatef(dx,dy,dz) glscalef(sx,sy,sz) glrotatef(angle,ax,ay,az)

38 Transformations in OpenGL OpenGL uses 4x4-matrices/homogeneous coordinates internally. Matrices are normally created by more intuitive commands: gltranslatef(dx,dy,dz) glscalef(sx,sy,sz) glrotatef(angle,ax,ay,az) Each command generates the corresponding matrix, and right-multiplies it on the current matrix. So last transformaton specified in code is first applied to vertices. Cf. the math notation f (g(h(x))) (where h is applied first to x, then g, then f ).

39 Transformations in OpenGL OpenGL uses 4x4-matrices/homogeneous coordinates internally. Matrices are normally created by more intuitive commands: gltranslatef(dx,dy,dz) glscalef(sx,sy,sz) glrotatef(angle,ax,ay,az) Each command generates the corresponding matrix, and right-multiplies it on the current matrix. So last transformaton specified in code is first applied to vertices. Cf. the math notation f (g(h(x))) (where h is applied first to x, then g, then f ). There is a current matrix for model-view transformations, for projections, and for textures. Each has a stack.

40 Matrix Stack

41 The Trick

42 Example Program