Introduction to Computer Graphics
GAMES101, Lingqi Yan, UC Santa Barbara
Lecture 10:
Geometry 1 (Introduction)
http://www.cs.ucsb.edu/~lingqi/teaching/games101.html
Announcements
• Homework 3
- The framework has been updated
- Together with an “FAQ” section in the BBS
• New TAs
- Peng Yu ( ) BUAA [email protected]
GAMES101 2 Lingqi Yan, UC Santa Barbara
Last Lectures
• Shading 1 & 2
- Blinn-Phong reflectance model
- Shading models / frequencies
- Graphics Pipeline
- Texture mapping
• Shading 3
- Barycentric coordinates
- Texture antialiasing (MIPMAP)
- Applications of textures
GAMES101 3 Lingqi Yan, UC Santa Barbara
Last Lectures
today
Today
• Applications of textures
• Introduction to geometry (2nd part of this course!)
- Examples of geometry
- Various representations of geometry
GAMES101 5 Lingqi Yan, UC Santa Barbara
Applications of Textures
Many, Many Uses for Texturing
In modern GPUs, texture = memory + range query (filtering)
• General method to bring data to fragment calculations
Many applications
• Environment lighting
• Store microgeometry
• Procedural textures
• Solid modeling
• Volume rendering
•…
GAMES101 7 Lingqi Yan, UC Santa Barbara
Environment Map
[Blinn & Newell 1976]
Light from the environment Rendering with the environment
GAMES101 8 Lingqi Yan, UC Santa Barbara
Environmental Lighting
Environment map (left) used to render realistic lighting
GAMES101 9 Lingqi Yan, UC Santa Barbara
Spherical Environment Map
Hand with Reflecting Sphere. M. C. Escher, 1935. lithograph Light Probes, Paul Debevec
GAMES101 10 Lingqi Yan, UC Santa Barbara
Spherical Map — Problem
Prone to distortion (top and bottom parts)!
GAMES101 11 Lingqi Yan, UC Santa Barbara
Cube Map
(u, v) = (1, 1)
x=y=z
right face has
x > |y| and
x > |z|
(u, v) = (0, 0)
x = –y = –z
A vector maps to cube point along that direction.
The cube is textured with 6 square texture maps.
GAMES101 12 Lingqi Yan, UC Santa Barbara
[Emil Persson]
Much less distortion!
CS184/284A Need dir->face computation
Ren Ng
In an earlier paper 121, the author described
This effect might be desirable for some the effect of aliasing on images made with color
applications but undesirable for others. A scale texture mapping. The same problems can arise with
Textures can affect shading!
invariant perturbation, D', must scale at the same this new form. That is, undesirable artifacts can
rate as N. An obvious choice for this is enter the image in regions where the texture
pattern maps into a small screen region. The
D' = a D INI/IDI solution applied to color textures was to average
the texture pattern over the region corresponding
50 ID’1 = a INI to each picture element in the final image. The
• Textures doesn’t have to only represent colors
bump texture definition function, however, does
where a is independent of scales in P. The value not have a linear relationship to the intensity of
of a is then the tangent of the effective rotation the final image. If the bump texture is averaged
angle. the effect will be to smooth out the bumps rather
- What
tan+' =ifID'l/lNl
it stores
= a the height / normal?
than average the intensities. The correct
solution to this problem would be to compute the
intensities at some high sub-pixel resolution and
choice is a- generalization
This can be defined in various ways. One simple average them. Simply filtering the bump function
Bump / normal from themapping
simple, flat can,
- however,
-. reduce the more offensive artifacts
unit square patch o f aliasing. Figure 10 shows the result of such
- Fake the detailed geometry
an operation.
FigureFigure 8 Drawn
A- Hand Bump Funtions
Hand Drawn Functio
Relative height to the height -> normal -> shading
underlying
Before surface
:
After
GAMES101 14 Lingqi Yan, UC Santa Barbara
Figure 10 - Filtering Bump Texture
Bump Mapping
Adding surface detail without adding more triangles
• Perturb surface normal per pixel
(for shading computations only)
• “Height shift” per texel defined by a texture
• How to modify normal vector?
n=?
Bump texture p
b(u,v)
GAMES101 15 Lingqi Yan, UC Santa Barbara
How to perturb the normal (in flatland)
• Original surface normal n(p) = (0, 1)
• Derivative at p is dp = c * [h(p+1) - h(p)]
• Perturbed normal is then n(p) = (-dp, 1).normalized()
GAMES101 16 Lingqi Yan, UC Santa Barbara
How to perturb the normal (in 3D)
• Original surface normal n(p) = (0, 0, 1)
• Derivatives at p are
- dp/du = c1 * [h(u+1) - h(u)]
- dp/dv = c2 * [h(v+1) - h(v)]
• Perturbed normal is n = (-dp/du, -dp/dv, 1).normalized()
• Note that this is in local coordinate!
More will be elaborated in FAQ of HW3
GAMES101 17 Lingqi Yan, UC Santa Barbara
Textures can affect shading!
• Displacement mapping — a more advanced approach
- Uses the same texture as in bumping mapping
- Actually moves the vertices
Bump / Normal mapping Displacement mapping
GAMES101 18 Lingqi Yan, UC Santa Barbara
3D Procedural Noise + Solid Modeling
Perlin noise, Ken Perlin
Ken Perlin
GAMES101 19 Lingqi Yan, UC Santa Barbara
Provide Precomputed Shading
Autodesk
Simple Ambient occlusion With ambient
shading texture map occlusion
GAMES101 20 Lingqi Yan, UC Santa Barbara
3D Textures and Volume Rendering
Marc Levoy
GAMES101 21 Lingqi Yan, UC Santa Barbara
Today
• Shading 3
- Applications of textures
• Introduction to geometry
- Examples of geometry
- Various representations of geometry
GAMES101 22 Lingqi Yan, UC Santa Barbara
Examples of Geometry
GAMES101 23 Lingqi Yan, UC Santa Barbara
Examples of Geometry
NetCarShow.com
GAMES101 24 Lingqi Yan, UC Santa Barbara
Examples of Geometry
GAMES101 25 Lingqi Yan, UC Santa Barbara
Examples of Geometry
GAMES101 26 Lingqi Yan, UC Santa Barbara
Examples of Geometry
GAMES101 27 Lingqi Yan, UC Santa Barbara
Examples of Geometry
GAMES101 28 Lingqi Yan, UC Santa Barbara
Examples of Geometry
GAMES101 29 Lingqi Yan, UC Santa Barbara
Examples of Geometry
I will call it
“Trump Virus”
COVID-19 forever.
GAMES101 30 Lingqi Yan, UC Santa Barbara
Examples of Geometry
Adriana Franco, National Geographic
GAMES101 31 Lingqi Yan, UC Santa Barbara
Many Ways to Represent Geometry
Implicit
• algebraic surface
• level sets
• distance functions
• ...
Explicit
• point cloud
• polygon mesh
• subdivision, NURBS
• ...
Each choice best suited to a different task/type of geometry
GAMES101 32 Lingqi Yan, UC Santa Barbara
“Implicit” Representations of Geometry
Based on classifying points
• Points satisfy some specified relationship
E.g. sphere: all points in 3D, where x2+y2+z2 = 1
f(x,y)
More generally, f(x,y,z) = 0 +1
f=0
-1
GAMES101 33 Lingqi Yan, UC Santa Barbara
Implicit Surface – Sampling Can Be Hard
p
2 2
f (x, y, z) = (2 x2 + 2
y ) +z 1
What points lie on f(x,y,z) = 0? y
z x
Some tasks are hard with implicit representations
Implicit Surface – Inside/Outside Tests Easy
2 2 2
f (x, y, z) = x + y + z 1
y
Is (3/4, 1/2, 1/4) inside?
Just plug it in:
f(x,y,z) = –1/8 < 0 ( 3/4, 1/2, 1/4 )
Yes, inside.
z x
Implicit representations make some tasks easy
“Explicit” Representations of Geometry
All points are given directly or via parameter mapping
Generally:
Explicit Surface – Sampling Is Easy
f (u, v) = ((2 + cos u) cos v, (2 + cos u) sin v, sin u)
What points lie on this surface? y
Just plug in (u,v) values!
z x
Explicit representations make some tasks easy
Explicit Surface – Inside/Outside Test Hard
f (u, v) = (cos u sin v, sin u sin v, cos v)
y
Is (3/4, 1/2, 1/4) inside?
( 3/4, 1/2, 1/4 )
z x
Some tasks are hard with explicit representations
No “Best” Representation – Geometry is Hard!
“I hate meshes.
I cannot believe how hard this is.
Geometry is hard.”
— David Baraff
Senior Research Scientist
Pixar Animation Studios
Slide cribbed from Keenan Crane, cribbed from Jeff Erickson.
Best Representation
Depends on the Task!
More Implicit Representations in
Computer Graphics
Many Implicit Representations in Graphics
Algebraic surfaces
Constructive solid geometry
Level set methods
Fractals
...
GAMES101 42 Lingqi Yan, UC Santa Barbara
Algebraic Surfaces (Implicit)
Surface is zero set of a polynomial in x, y, z
More complex shapes?
GAMES101 43 Lingqi Yan, UC Santa Barbara
Constructive Solid Geometry (Implicit)
Combine implicit geometry via Boolean operations
Union
Intersection Difference
Boolean expressions:
GAMES101 44 Lingqi Yan, UC Santa Barbara
Distance Functions (Implicit)
Instead of Booleans, gradually blend surfaces together using
Distance functions:
giving minimum distance (could be signed distance)
from anywhere to object
Distance Functions (Implicit)
An Example: Blending (linear interp.) a moving boundary
Blending Distance Functions (Implicit)
Can blend any two distance functions d1, d2:
GAMES101 47 Lingqi Yan, UC Santa Barbara
Scene of Pure Distance Functions
See https://iquilezles.org/www/articles/raymarchingdf/raymarchingdf.htm
Level Set Methods (Also implicit)
Closed-form equations are hard to describe complex shapes
Alternative: store a grid of values approximating function
-.55 -.45 -.35 -.30 -.25
-.30 -.25 -.20 -.10 -.10
-.20 -.15 -.10 .10 .15
-.05 .10 .05 .25 .35
.15 .20 .25 .55 .60
Surface is found where interpolated values equal zero
Provides much more explicit control over shape (like a
texture)
GAMES101 49 Lingqi Yan, UC Santa Barbara
Level Sets from Medical Data (CT, MRI, etc.)
Level sets encode, e.g., constant tissue density
GAMES101 50 Lingqi Yan, UC Santa Barbara
Level Sets in Physical Simulation
Level set encodes distance to air-liquid boundary
See http://physbam.stanford.edu
Fractals (Implicit)
Exhibit self-similarity, detail at all scales
“Language” for describing natural phenomena
Hard to control shape!
GAMES101 52 Lingqi Yan, UC Santa Barbara
Implicit Representations - Pros & Cons
Pros:
• compact description (e.g., a function)
• certain queries easy (inside object, distance to surface)
• good for ray-to-surface intersection (more later)
• for simple shapes, exact description / no sampling error
• easy to handle changes in topology (e.g., fluid)
Cons:
• difficult to model complex shapes
GAMES101 53 Lingqi Yan, UC Santa Barbara
Thank you!
(And thank Prof. Ravi Ramamoorthi and Prof. Ren Ng for many of the slides!)
