Transcript Curtis Sean E - McGill University

Computer-Generated Watercolor
Cassidy J. Curtis
Kurt W. Fleischer
Sean E. Anderson
David H. Salesin
Irwin Chiu Hau
Computer Science
McGill University
Winter 2004
Comp 767: Advanced Topics in Graphics
Overview
Introduction
 Properties of watercolor
 Computer-generated watercolor
 Applications
 Future work
 Conclusion

Introduction
What is watercolor painting?
 Computer generated watercolor as a
non-photorealistic rendering

Overview
Introduction
 Properties of watercolor
 Computer-generated watercolor
 Applications
 Future work
 Conclusion

Properties of Watercolor

Watercolor materials
 Watercolor
paper
 Pigment
 Binder
 Surfactant

Watercolor effects
Watercolor Paper
 Typically
not made of wood pulp
 But
from linen or cotton rags
pounded into small fibers

Extremely absorbent to liquids
 Filled
with sizing usually made of
cellulose

Source: misterart.com
Slows down the rate of water
absorption and diffusion
Pigment
• A pigment is a solid material in
the form of small, separate
particles (ranging from 0.05 to
0.5 microns)
• Pigments vary in density
Source: misterart.com
Binder and Surfactant
Binder
 Adsorption
 Enables
the pigment
to adhere to the paper
Surfactant
 Allows water to soak
into sized paper
Binder
Source: Jerry’s ARTARAMA
Properties of watercolor
Watercolor materials
 Watercolor effects

 Dry-brush
effects
 Edge darkening
 Intentional backruns
 Granulation and Separation
 Flow Patterns
 Glazing
Dry-brush Effects

Techniques



Effects


Source: Computer Generated Watercolor
Dry brush that is
almost dried
Applied at a proper
angle
Irregular gaps
Ragged edges
Edge Darkening

Techniques


Effect

Source: Computer Generated Watercolor
Wet-on-dry brushstroke
Darken edges
Intentional Backruns
 Occurs
when
A puddle of water spread
back into a damp region of
paint
 A wash brush dries unevenly
 The water tends to push
pigment along as it spreads

 Effect
Source:
Computer Generated Watercolor
Complex branching shapes
 Severely darkened edges

Granulation and Separation of Pigments
 Granulation
of pigments
Yields a kind of grainy textures
 Varies from pigment to pigment
 Strongest when paper is very
wet

 Separation
Refers to splitting of colors
 Occurs when denser pigments
settle earlier than lighter ones

Source:
Computer Generated Watercolor
of pigment
Flow Patterns
 In

wet-in-wet painting
wet surface allows the
brushstrokes to spread freely
 Effects

Source:
Computer Generated Watercolor
Soft, feathery shapes
Glazing
 Techniques
Adding very thin, pale layers,
or washes, of watercolor, one
over another
 Different pigments are not
mixed physically, but optically

 Effects
Source:
Computer Generated Watercolor
luminous
 glowing from within

Overview
Introduction
 Properties of watercolor
 Computer-generated watercolor
 Applications
 Future work
 Conclusion

Computer-Generated Watercolor
Real watercolor effects
Simulated watercolor effects
Source: Computer Generated Watercolor
Implementation
Paper generation
 Data structure
 Fluid simulation
 Optical compositing

Paper Generation

Use a simple model
texture is modeled as a height field h
and a fluid capacity field c
 Paper
h is pseudo-randomly generated , 0 < h < 1
c = h * (cmax – cmin ) + cmin
Example paper textures
Source: Computer Generated Watercolor
Data Structure

A complete painting consists of
 Ordered

set of washes over a sheet of paper
Each wash may contain
 Various
pigments in varying quantities over
different parts of the image

We store these quantities in
A
data structure called a ‘glaze’
Glaze

Each glaze is created


by running a fluid simulation
Inputs:
Properties of pigments, paper, watercolor
medium
 Wet-area mask


Once the glazes are computed

They are optically composited using the
Kubelka-Munk color model
The Fluid Simulation

Each wash simulated using a three-layer model
Source: Computer Generated Watercolor
The Fluid Simulation
Main loop
proc MainLoop
for each time step do:
MoveWater
MovePigment
TransferPigment
SimulateCapillaryFlow
end for
end proc
The Fluid Simulation
Cellular Automaton
 Definition from Mathworld.com

A
cellular automaton is a collection of "colored"
cells on a grid of specified shape that evolves
through a number of discrete time steps
according to a set of rules based on the states
of neighboring cells
Game of Life
Source: Mathworld.com
Move Water
proc MoveWater(M, u, v, p):
UpdateVelocities(M, u, v, p)
RelaxDivergence(M, u, v, p)
FlowOutward(M, p)
end proc
M : wet-area mask
u, v : velocity
p : water pressure
edge darkening
Move Pigment
Pigments move within the shallow-water layer as
specified by the velocity field u, v
 Pigment from each cell are distributed to its
neighbors at the rate of fluid movement out of
the cell

Transfer Pigment

Pigment adsorption and desorption
proc TransferPigment(g 1, . . . ,g n,d 1, . . . ,d n ):
for each pigment k do
g, d : pigment concentrations
for all cells (i, j) do
…
Source: Computer Generated Watercolor
Simulate Capillary Flow

Diffusing water through the capillary layer
proc SimulateCapillaryFlow(s, M ):
for each pigment k do
for all cells (i, j) do
…
s : water saturation
of the paper
dry-brush effects
backruns
Source: Computer Generated Watercolor
The Fluid Simulation
Main loop
initial velocity
initial wet-area mask
initial water saturation of the paper
initial water pressure
proc MainLoop(M, u, v, p, g 1, … , g n, d 1, … , d n, s ):
for each time step do:
initial pigment concentrations
MoveWater(M, u, v, p)
MovePigment(M, u, v, g 1, … , g n)
TransferPigment(g 1, … , g n, d 1, … , d n)
SimulateCapillaryFlow(M, s)
end for
end proc
Optical compositing

Rendering the pigmented layers
 Use
the Kubelka-Munk (KM) model to
perform the optical compositing of glazing
layers
Source: Computer Generated Watercolor
Kubelka-Munk (KM) Model

Comes from KM Theory

Tells us how to
 specify
the optical properties of pigments
 optically composite pigments
 optically composite layers
Overview
Introduction
 Properties of watercolor
 Computer-generated watercolor
 Applications
 Future work
 Conclusion

Applications
Interactive painting with watercolors
 Automatic image “watercolorization”
 Non-photorealistic rendering of 3D models

Interactive Painting

User creates
 Glazes

User adjusts
 Brush
sizes
 Pigments
 Wet-mask area
 Physical
parameters
Source: Computer Generated Watercolor
Automatic image “watercolorization”
“Automatically” convert a color image into
a watercolor illustration
 Is done in two steps

 Color
separation
 Brushstroke planning
Color Separation
Color Separation Process
Source: Computer Generated Watercolor
Brushstroke Planning

Painter control the concentration and the flow of
pigment in a wash
Too much pigment
Lack of pigment
Add a pigmented wash
Thins them by adding water
Brushstroke Planning
Source: Computer Generated Watercolor
Automatic image “watercolorization”
Original image
An automatic watercolorization
Source: Computer Generated Watercolor
Steps for Rendering
Source: Computer Generated Watercolor
Non-photorealistic rendering of
3D models
Given a 3D geometric scene, we
automatically generate mattes isolating
each object
 These mattes are used as input to the
watercolorization process
 The user specifies the pigment choices
and brushstroke planning

Non-Photorealistic Animation
3D Scene
Detail of one frame
Several frames from a non-photorealistic animation of moving clouds
Source: Computer Generated Watercolor
Overview
Introduction
 Properties of watercolor
 Computer-generated watercolor
 Applications
 Future work
 Conclusion

Future Work
Other effects
 Automatic rendering
 Generalization
 Animation Issues

Overview
Introduction
 Properties of watercolor
 Computer-generated watercolor
 Applications
 Future work
 Conclusion

Conclusion

That’s all about
Computer Generated Watercolor

Questions ???

Discussions ???
References

Cassidy J. Curtis, Sean E. Anderson,
Kurt W. Fleischer and David H. Salesin.
Computer-Generated Watercolor

Images


www.misterart.com
www.jerrysartarama.com