cpsc 641: computer graphics image formation jinxiang chai

CPSC 641: Computer GraphicsImage Formation

Jinxiang Chai

Are They Images?

Outline

• Color representation• Image representation• Pin-hole Camera• Projection matrix• Plenoptic function

Color Representation

• Why do we use RGB to encode pixel color?

• Can we use RGB to represent all colors?

• What are other color representations?

Human Vision

Model of human vision

Human Vision

Model of human vision

Vision components:

• Incoming light

• Human eye

Electromagnetic Spectrum

Visible light frequencies range between:– Red: 4.3X1014 hertz (700nm)– Violet: 7.5X1014 hertz (400nm)

Visible Light

The human eye can see “visible” light in the frequency between 400nm-700nm

Visible Light


400nm 700nm

Visible Light


400nm 700nm

- Not strict boundary

- Some colors are absent (brown, pink)

Spectral Energy Distribution

Three different types of lights

Spectral Energy DistributionThe six spectra below look the same purple to

normal color-vision people

Color Representation?

Why not all ranges of light spectrum are perceived?

So how to represent color?

- unique

- compact

- work for as many visible lights as possible

400nm 700nm

Human Vision

Photoreceptor cells in the retina:

- Rods

- Cones

Light Detection: Rods and Cones

Rods: -120 million rods in retina -1000X more light sensitive than Cones - Discriminate B/W brightness in low illumination - Short wave-length sensitive

Cons: - 6-7 million Cones in the retina - Responsible for high-resolution vision - Discriminate Colors - Three types of color sensors (64% red, 32%, 2% blue) - Sensitive to any combination of three colors

Tristimulus of Color Theory

Spectral-response functions of each of the three types of cones



Can we use them to match any spectral color?



Color matching function based on RGB - any spectral color can be represented as a linear combination of

these primary colors

Tristimulus Color Theory

So, color is psychological- Representing color as a linear combination of red, green, and

blue is related to cones, not physics

- Most people have the same cones, but there are some people who don’t – the sky might not look blue to them (although they will call it “blue” nonetheless)

- But many people (mostly men) are colorblind, missing 1,2 or 3 cones (can buy cheaper TVs)

Additive and Subtractive Color

RGB color model CMY color model

Complementary color models: R=1-C; G = 1-M; B=1-Y;

White: [1 1 1]T

Green: [0 1 0];

White: [0 0 0]T

Green: [1 0 1];

RGB Color Space

RGB cube– Easy for devices– Can represent all the colors?– But not perceptual– Where is brightness, hue and saturation?

red

green

blue

Tristimulus

• Since 3 different cones, the space of colors is 3-dimensional.

• We need a way to describe color within this 3 dimensional space.

• We want something that will let us describe any visible color with additive combination…

The CIE XYZ system

• CIE – Comission Internationale de l’Eclairage- International Commission on Illumination- Sets international standards related to light

• Defined the XYZ color system as an international standard in 1931

• X, Y, and Z are three Primary colors. - imaginary colors - all visible colors can be defined as an additive combination of

these three colors. - defines the 3 dimensional color space

Chromaticity Diagram

• Project the X+Y+Z=1 slice along the Z-axis

• Chromaticity is given by the x, y coordinates

CIE Perceptual Space

Which colors can RGB monitor displays?

Monitor/Print/Scanner Gamut

HSV Color Model

Perceptually appropriate:- Hue: the color type (0-360 deg)

- Saturation: the intensity of the color (0-100%)

- Brightness: the brightness of color (0-100%)

Nonlinear transform between the HSV and RGB space

Outline


Image Representation

An image is a 2D rectilinear array of Pixels

- A width X height array where each entry of the array stores a single pixel


A pixel stores color information

Luminance pixels - gray-scale images (intensity images) - 0-1.0 or 0-255 - 8 bits per pixel

Red, green, blue pixels (RGB) - Color images - Each channel: 0-1.0 or 0-255 - 24 bits per pixel


An image is a 2D rectilinear array of Pixels

- A width X height array where each entry of the array stores a single pixel

- Each pixel stores color information

(255,255,255)

Outline

• Color representation• Image representation• Pin-hole Camera• Projection matrix• Plenoptic Function

How Do We See the World?

Let’s design a camera: idea 1: put a piece of film in front of camera

Do we get a reasonable picture?

Pin-hole Camera

• Add a barrier to block off most of the rays– This reduces blurring– The opening known as the aperture– How does this transform the image?

Camera Obscura

• The first camera– Known to Aristotle– Depth of the room is the focal length– Pencil of rays – all rays through a point

Camera Obscura

How does the aperture size affect the image?

Shrinking the Aperture

• Why not make the aperture as small as possible?– Less light gets through– Diffraction effects…

Shrink the Aperture: Diffraction

A diffuse circular disc appears!

Shrink the Aperture

The Reason of Lenses

Adding A Lens

• A lens focuses light onto the film– There is a specific distance at which objects are “in

focus”• other points project to a “circle of confusion” in the

image– Changing the shape of the lens changes this distance

“circle of confusion”

Changing Lenses

28 mm 50 mm

210 mm70 mm

Outline


Projection Matrix

• What’s the geometric relationship between 3D objects and 2D images?

Modeling Projection: 3D->2D

The coordinate system– We will use the pin-hole model as an approximation– Put the optical center (Center Of Projection) at the origin– Put the image plane (Projection Plane) in front of the COP– The camera looks down the negative z axis


Projection equations– Compute intersection with PP of ray from (x,y,z) to

COP– Derived using similar triangles (on board)


Projection equations– Compute intersection with PP of ray from (x,y,z) to

COP– Derived using similar triangles (on board)

– We get the projection by throwing out the last coordinate:

Homogeneous Coordinates

Is this a linear transformation?– no—division by z is nonlinear


Is this a linear transformation?

Trick: add one more coordinate:

homogeneous image coordinates

homogeneous scene coordinates

– no—division by z is nonlinear


Is this a linear transformation?

Trick: add one more coordinate:

homogeneous image coordinates

homogeneous scene coordinates

Converting from homogeneous coordinates

– no—division by z is nonlinear

Perspective Projection

divide by third coordinate

Projection is a matrix multiply using homogeneous coordinates:

Perspective Projection

divide by third coordinate

This is known as perspective projection– The matrix is the projection matrix– Can also formulate as a 4x4

Projection is a matrix multiply using homogeneous coordinates:

divide by fourth coordinate

Perspective Effects

Distant object becomes small

The distortion of items when viewed at an angle (spatial foreshortening)

Parallel ProjectionSpecial case of perspective projection

– Distance from the COP to the PP is infinite

– Also called “parallel projection”– What’s the projection matrix?

Image World

Weak-perspective Projection

Scaled orthographic projection

- object size is small as compared to the average distance from the camera z0 (e.g.σz < z0/20)

- d/z ≈ d/z0 (constant)

Weak-perspective Projection

Scaled orthographic projection

- object size is small as compared to the average distance from the camera z0 (e.g.σz < z0/20)

- d/z ≈ d/z0 (constant)

Projection matrix: λ

λ

d z0

Spherical Projection

What if PP is spherical with center at COP?

In spherical coordinates, projection is trivial:

Note: it doesn’t depend on focal length d!

View Transformation

From world coordinate to camera coordinate

P

Viewport Transformation

x

y

u

vu0, v0

From projection coordinate to image coordinate

Viewport Transformation

x

y

u

vu0, v0

u0

v0

100

-sy0

sx 0u

v

1

x

y

1

From projection coordinate to image coordinate

Putting It Together

From world coordinate to image coordinate

u0

v0

100

-sy0

sx 0u

v

1

Perspective projection

View transformation

Viewport projection

Putting It Together

From world coordinate to image coordinate

u0

v0

100

-sy0

sx 0u

v

1

Perspective projection

View transformation

Viewport projection

Image resolution, aspect ratio

Focal length The relative position & orientation between camera and objects

Camera Parameters

Totally 11 parameters,

u0

v0

100

-sy0

sx 0u

v

1

Camera Parameters

Totally 11 parameters,

u0

v0

100

-sy0

sx 0u

v

1

Intrinsic camera parameters

extrinsic camera parameters

How about this image?

Outline


Plenoptic Function

What is the set of all things that we can ever see?

- The Plenoptic Function (Adelson & Bergen)

Plenoptic Function

What is the set of all things that we can ever see?

- The Plenoptic Function (Adelson & Bergen)

Let’s start with a stationary person and try to parameterize everything that he can see…

Plenoptic Function

Any ray seen from a single view point can be parameterized by (θ,φ).

Color Image

is intensity of light – Seen from a single view point (θ,φ)– At a single time t– As a function of wavelength λ

P(θ,φ,λ)

Dynamic Scene

is intensity of light – Seen from a single view point (θ,φ)– Over time t– As a function of wavelength λ

P(θ,φ,λ,t)

Moving around A Static Scene

is intensity of light – Seen from an arbitrary view point (θ,φ)– At an arbitrary location (x,y,z)– At a single time t– As a function of wavelength λ

P(x,y,z,θ,φ,λ)

Moving around A Dynamic Scene

is intensity of light – Seen from an arbitrary view point (θ,φ)– At an arbitrary location (x,y,z)– Over time t– As a function of wavelength λ

P(x,y,z,θ,φ,λ,t)

Plenoptic Function

Can reconstruct every possible view, at every moment, from every position, at every wavelength

Contains every photograph, every movie, everything that anyone has ever seen! it completely captures our visual reality!

An image is a 2D sample of plenoptic function!

P(x,y,z,θ,φ,λ,t)

How to “Capture” Orthographic Images

Rebinning rays forms orthographic images

Multi-perspective ImagesRebinning rays forms multiperspective

images

Multi-perspective ImagesRebinning rays forms multiperspective

images

……

Multi-perspective Images

Outline


They Are All Images

Next lecture

Image sampling theory

Fourier Analysis

cpsc 641: computer graphics image formation jinxiang chai

Documents