CS 655

1
CS 655

• Distributed Ray Tracing

2
Distributed Ray Tracing

• What is distributed ray tracing?
• Distributed ray tracing is not ray tracing on a
  distributed system.
• Distributed ray tracing is a ray tracing method
  based on randomly distributed oversampling to
  reduce aliasing artifacts in rendered images.
• Also called distribution ray tracing or
  stochastic ray tracing
• (The images in this presentation are from Andrew
  G. Zaferakis, Aneesh Naman, Allen Martin, Daqing
  Xue, Henrik Wann Jensen, and Jinho Lee)

3
Distributed Ray Tracing

• Idea Introduce noise into the ray tracer to
  minimize visual artifacts such as aliasing,
  perfect reflections, etc.
• Stocastically distribute rays over
• Space antialiases the image
• Reflection angle produces glossy reflections
• Transmission angle produces translucency
• Shadow ray produces soft shadows (penumbra)
• Lens area produces depth of field
• Time produces motion blur

4
Antialiasing

• The human eye samples using a Poisson disk
  distribution
• A finite number of photoreceptors
• Cones in the eye are distributed stochastically,
  but such that no two cones are closer than a
  certain distance

5
Antialiasing

• We can apply this approach to ray tracing
• Send out rays stochastically, but such that they
  maintain a Poisson disk distribution
• This reduces aliasing, but is expensive to
  compute
• However, we can approximate a Poisson disk
  distribution fairly quickly
• Idea begin with a regular grid
• Jitter the ray locations slightly within the grid

6
Uniform Sampling
Prone to aliasing in the image
7
Jittered Sampling
Antialiases the image
8
Jittered Sampling

• Approximates a Poisson disk distribution
• Not exact, but much cheaper
• Removes high frequencies
• Introduces noise in place of high frequencies
• Can still have problems
• May leave large areas uncovered by samples
• May have some areas with more samples than
  necessary

9
Improved Jittered Sampling

• To improve the image further, send out multiple
  jittered rays per pixel
• Make the number of sub-pixels easily changeable
  start small then increase as you are confident
  your ray-tracer works.

Similarly for all other pixels
10
Standard Ray Tracing

• No antialiasing

11
Distributed Ray Tracing

• With antialiasing

12
Standard Ray Tracing - detail

• No antialiasing

13
Distributed Ray Tracing - detail

• With antialiasing

14
Glossy Reflections

• We can use the concept of ray jittering to
  produce glossiness (blurred reflections)
• Tracing a ray based on perfect reflection angle
  produces sharp reflections

15
Glossy Reflections

• We can create a blurred reflection by sending out
  jittered rays about the reflection ray

N
R
I
Bound of jittered reflection rays
16
Glossy Reflections

• This has the effect of blurring the reflection
• The exact reflection vector is computed, then
  slightly jittered from the original direction
• The jittered ray may hit an entirely different
  object than the one hit by the true reflection
  ray
• This gives a smoothly blurred reflection

17
Reflections Standard Ray Tracer

• Perfect reflection

18
Reflections Distributed Ray Tracer

• 10 distributed rays

19
Reflections Distributed Ray Tracer

• 20 distributed rays

20
Reflections Distributed Ray Tracer

• 50 distributed rays

21
Standard Ray Tracing

• Perfect reflections

22
Glossy reflections

• 4 rays

23
Glossy reflections

• 16 rays

24
Glossy reflections

• 64 rays

25
Altering Jitter Amount

• h 10

26
Altering Jitter Amount

• h 45

27
Altering Jitter Amount

• h 400

28
Translucency

• We can also apply the concept of ray jittering to
  produce translucency (blurred transmissions)
• Tracing a ray based on perfect transmission angle
  produces sharp transparencies

N
I
T
29
Translucency
N
I

• Jittering the rays about the actual transmission
  angle produces a blurred effect

T
Bound of jittered reflection rays
30
Transmissions Standard Ray Tracer

• Perfect transmission

31
Transmissions Distributed Ray Tracer

• 10 transmission rays

32
Transmissions Distributed Ray Tracer

• 20 transmission rays

33
From Jia, Sun, and Xu
34
Altering Jitter Amount

• h 10

35
Altering Jitter Amount

• h 65

36
Soft Shadows

• We have been simulating lights with point light
  sources
• This produces hard shadows

Point light source -hard shadow
37
Soft Shadows

• Lights in the real world are not point light
  sources, thus the shadows are not sharp

Area light source
penumbra
umbra
Area light source -soft shadow
38
Soft Shadows

• The penumbra is the portion of the shadow
  resulting from partially obscured lights
• To simulate penumbra
• Send out multiple shadow rays from the
  intersection point to the area light source
• Jitter the rays according to the area light
  source
• The intensity of the surface point depends on the
  number of jittered rays that reach the light
  source

Area light source
Bounds of jittered shadow rays
Occluding object
Intersection point
39
Hard Shadows
40
Soft Shadows
41
Standard Ray Tracing

• Hard shadows

42
Ray Tracing with an Area Light Source

• Visible penumbra and umbra, but too distinct

43
Distributed Ray Tracing

• 10 rays per pixel

44
Distributed Ray Tracing

• 20 rays per pixel

45
Distributed Ray Tracing

• 50 rays per pixel

46
Depth of Field

• Idea the camera (or eye) should have a fixed
  focal length
• Objects at that distance should be in focus
• Objects closer or further away should not be in
  focus

Lens
Image Plane
Focal Plane
47
Lens Properties
D
VD
r
P
Vp
PD
Pf
C
F/n
Object in Scene
Lens
Image Plane
Focal Plane
48
Lens Properties

• F Focal length
• n aperture number
• F/n lens diameter
• Pf the focal point
• P distance from lens to focal point
• Vp distance from lens to image plane
• PD a point not on the focal plane
• D distance from lens to PD
• r radius of cone for object at distance D
• C circle of confusion

49
Simulating a lens

• Approach 1
• For an object located at PD, send out a group of
  jittered rays that lie within a cone of radius r
• How do we compute r?
• This simulates a lens without actually having
  one, but isnt very accurate

50
Simulating a lens

• Approach 2
• Find the focal point
• Send a ray from the center of the lens (eye
  point) through the screen and follow it a
  distance P
• Choose a jittered point on the lens
• Trace a ray from that jittered point through the
  focal point
• Return intensity information based on what that
  ray hits

51
Simulating a lens

• Step 1 determine the focal point by tracing a
  ray from the lens center through the pixel a
  distance P

52
Simulating a lens

• Step 2 Choose a jittered point on the lens

New point
53
Simulating a lens

• Step 3 Trace the ray from the new point through
  the focal point

54
Simulating a lens

• Step 4 Return the intensity information

55
Depth of Field Example
56
Depth of Field Example
From Alan Watt, 3D Computer Graphics
57
(No Transcript)
58
Motion Blur

• Distribute rays over time
• Static objects will not change
• Moving objects will be blurred, depending on
  their velocity
• How
• Jitter the rays with respect to time
• Determine object positions at each of the
  jittered time values
• Send each ray with the objects positioned
  appropriately
• Combine the rays back into one to get the motion
  blurred object

59
Motion Blur Example
60
Motion Blur Example (from Cook et. al.)
61
(No Transcript)
62
Summary of Distributed Ray Tracing

• For each ray do
• Jitter the spatial screen location of the ray
• Select a time for the ray and move the objects to
  that time
• Perform depth of field calculation
• Determine the focal point by sending a ray from
  the eye point (center of the lens) through the
  pixel, a distance P
• Determine a lens location by jittering the origin
  of the ray to a position on the lens
• Compute the intersection by sending the primary
  ray from the jittered location through the focal
  point
• Trace jittered reflection rays
• Trace jittered trasmission rays
• Trace jittered shadow rays

63
Summary of Distributed Ray Tracing
Shadow Ray
Reflected Ray
Transmitted Ray
64
Distributed Ray Tracing (Cook et.al.)