CSE 167:
Introduction to Computer Graphics
Lecture #18: Volume Rendering
Jürgen P. Schulze, Ph.D.
University of California, San Diego
Fall Quarter 2013
Announcements
► Cross-check all scores on Ted
► Two more blog deadlines: Sunday and Wednesday
► Final Project Presentations in CSE 1202, Thursday 3-6pm
► Bring computer with VGA adapter
► contact instructor if this is not an option for you
► CAPE reminder
2
Lecture Overview
► Volume Rendering
► Overview
► Transfer Functions
► Rendering
3
What is Volume Rendering?
► Volume Rendering is a set of techniques used to display
a 2D projection of a volume data set.
► User specifies viewpoint, rendering algorithm and
transfer function
► Transfer function defines mapping of color and opacity to
voxels
► Wide spectrum of application domains
4
Weather
► Clouds
► http://www.youtube.com/watch?v=7obZdsEoGGA
5
Lecture Overview
► Volume Rendering
► Overview
► Transfer Functions
► Rendering
6
Transfer Functions
► Most volume data sets consist
of a single data value per voxel,
for instance the measured
material density.
► This data value can be
interpreted as:
► Luminance and rendered as a gray
value on a scale from black to
white
► An index into a color look-up
table
► Another look-up table maps
opacity to data values.
7
From GPU Gems
Opacity Transfer Function
► Modifying the mapping of
data value to opacity
exposes different parts of
the volume.
8
From Shin et al. 2004: Images a-f
show decreasing opacity.
Volume Filtering
► Applying a filter to the
volume data set can
improve image quality
► Filtering operation defined
by filter kernel
► Filter kernels on right:
► (a) blur filter
► (b) sharpen filter
► (c) edge filter
► In 3D, filter kernels typically
use a 6-, 18- or 26-voxel
neighborhood
9
Derived Voxel Data
► Applying a filter on the volume data set generates a new,
derived volume data set.
► Derived volume data can be stored with original volume in
a multi-channel volume data set.
► Example:
► Channel 1: original density data
► Channel 2: gradient magnitude
10
2D Transfer Functions
► A 2D transfer function can map RGBA values separately
to every combination of density data and gradient
magnitude
11
2D Transfer function and its parameters (Wan et al. 2009)
2D Transfer Functions
► Example: Rectangular 2D transfer function editor
12
Images by Gordon Kindleman and Joe Kniss
Lecture Overview
► Volume Rendering
► Overview
► Transfer Functions
► Rendering
13
Volume Rendering Techniques
► Iso-surface
► Cross-sections
► Direct volume rendering (DVR)
► Slicing with 2D textures
► Translucent textures with image plane-aligned 3D textures
► MIP
► Spatial constraints
► Region of interest (cubic, spherical, slab)
► Clipping plane
14
Volume Rendering Techniques
15
Images from: A Simple and Flexible Volume Rendering Framework for Graphics-
Hardware-based Raycasting, Stegmaier et al. 2005
Maximum
Intensity
Projection
Iso-Surface
Transparent Iso-
Surfaces
Raycasting
(DVR)
Raycasting and
Iso-Surface
Volume Rendering Outline
Data Set
3D Rendering
Rendering done in real-time on
commodity graphics hardware
Classification
16
Ray Casting
► Software Solution
Image Plane
Data Set
Numerical Integration
Resampling
High Computational Load
Eye
17
Ray Casting
► Software Solution
Image Plane
Data Set
Numerical Integration
Resampling
High Computational Load
Eye
18
Plane Compositing
Proxy geometry (Polygonal Slices)
19
Compositing
► Maximum Intensity Projection
No emission/absorption
Simply compute maximum value along a ray
Emission/Absorption
Maximum Intensity Projection
20
2D Textures
Decompostition into axis-aligned slices
Draw the volume as a stack of 2D textures
Bilinear Interpolation in Hardware
3 copies of the data set in memory
21
2D Textures: Drawbacks
Sampling rate is inconsistent
d
d´ ≠
d
Emission/absorption slightly incorrect
Super-sampling on-the-fly impossible
22
3D Textures
R G B A
R
G
B
For each fragment:
interpolate the
texture coordinates
(barycentric)
Texture-Lookup:
interpolate the
texture color
(trilinear)
23
3D Textures
Slices parallel to the image plane
3D Texture: Volumetric Texture Object
Trilinear Interpolation in Hardware
One large texture block in memory
24
Comparison of 2D with 3D Texturing
25
Left: 2D textures, right: 3D textures
[Lewiner2006]
Resampling via 3D Textures
Sampling rate is constant
d
d
Supersampling by increasing the
number of slices
26
Shadows
27
Volume rendering with shadows
(from GPU Gems)
Implementation: Loading a 3D Texture
►
// init the 3D texture
►
glEnable(GL_TEXTURE_3D_EXT);
►
glGenTextures(1, &tex_glid);
►
glBindTexture(GL_TEXTURE_3D_EXT, tex_glid);
►
// texture environment setup
►
glTexParameteri( GL_TEXTURE_3D_EXT, GL_TEXTURE_MIN_FILTER, GL_LINEAR );
►
glTexParameteri( GL_TEXTURE_3D_EXT, GL_TEXTURE_MAG_FILTER, GL_LINEAR );
►
glTexParameteri( GL_TEXTURE_3D_EXT, GL_TEXTURE_WRAP_R, GL_CLAMP_TO_EDGE );
►
glTexParameteri( GL_TEXTURE_3D_EXT, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE );
►
glTexParameteri( GL_TEXTURE_3D_EXT, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE );
►
// load the texture image
►
glTexImage3DEXT(GL_TEXTURE_3D_EXT, // target
►
0, // level
►
GL_RGBA, // color storage
►
(int) tex_ni(), // width
►
(int) tex_nj(), // height
►
(int) tex_nk(), // depth
►
0, // border
►
GL_COLOR_INDEX, // format
►
GL_FLOAT, // type
►
_texture ); // allocated texture buffer
►
glPixelTransferi(GL_MAP_COLOR, GL_FALSE);
28
Demo: DeskVox
► DeskVox was created at IVL/Calit2
► http://ivl.calit2.net/wiki/index.php/VOX_and_Virvo
29
Videos
► Human head, rendered with 3D texture:
► http://www.youtube.com/watch?v=94_Zs_6AmQw
► GigaVoxels:
► http://www.youtube.com/watch?v=HScYuRhgEJw
30
References
► Volume rendering tutorial with source code
► http://graphicsrunner.blogspot.com/2009_01_01_archive.html
► Simian volume rendering software
► http://www.cs.utah.edu/~jmk/simian/
31
Lecture Overview
► Deferred Rendering Techniques
► Deferred Shading
► Screen Space Ambient Occlusion
► Bloom
► Glow
32
Deferred Rendering
► Opposite to Forward Rendering, which is the way we
have rendered with OpenGL so far
► Deferred rendering describes post-processing algorithms
► Requires two-pass rendering
► First pass:
► Scene is rendered as usual by projecting 3D primitives to 2D screen
space.
► Additionally, an off-screen buffer (G-buffer) is populated with
additional information about the geometry elements at every pixel
□ Examples: normals, diffuse shading color, position, texture coordinates
► Second pass:
► An algorithm, typically implemented as a shader, processes the G-
buffer to generate the final image in the back buffer
33
Lecture Overview
► Deferred Rendering Techniques
► Deferred Shading
► Screen Space Ambient Occlusion
► Bloom
► Glow
► The Future of Computer Graphics
34
Deferred Shading
► Postpones shading calculations for a fragment until its
visibility is completely determined
► Only fragments that really contribute to the image are
shaded
► Algorithm:
► Fill a set of buffers with common data, such as diffuse
texture, normals, material properties
► For the lighting just render the light extents and fetch data
from these buffers for the lighting computation
► Advantages:
► Decouples lighting from geometry
► Several lights can be applied with a single draw call:
more than 1000 light sources can be rendered at 60 fps
► Disadvantages:
► Consumes more memory, bandwidth and shader
instructions than traditional rendering
35
Particle system with
glowing particles.
Source: Humus 3D
Reference
► Deferred Shading Tutorial:
► http://bat710.univ-
lyon1.fr/~jciehl/Public/educ/GAMA/2007/Deferred_Shading_Tu
torial_SBGAMES2005.pdf
36
Lecture Overview
► Deferred Rendering Techniques
► Deferred Shading
► Screen Space Ambient Occlusion
► Bloom
► Glow
► The Future of Computer Graphics
37
Screen Space Ambient Occlusion
► Screen Space Ambient Occlusion is abbreviated as SSAO
► “Screen Space” refers to this being a deferred rendering approach
► Rendering technique for approximating ambient occlusion in real time
► Developed by Vladimir Kajalin while working at Crytek
► First use in 2007 PC game Crysis
38
SSAO component
Ambient Occlusion
► Attempts to approximate global illumination
► Very crude approximation
► Unlike local methods like Phong shading, ambient
occlusion is a global method
► Illumination at each point is a function of other geometry in
the scene
► Appearance achieved by ambient occlusion is similar to
the way an object appears on an overcast day
► Example: arm pit is hit by a lot less light than top of head
► In the industry, ambient occlusion is often referred to as
"sky light"
39
SSAO Demo
► Screen Space Ambient Occlusion (SSAO) in Crysis
► http://www.youtube.com/watch?v=ifdAILHTcZk
40
Basic SSAO Algorithm
► First pass:
► Render scene normally and write z values to g-buffer’s alpha channel
► Second pass:
► Pixel shader samples depth values around the processed fragment and
computes amount of occlusion, stores result in red channel
► Occlusion depends on depth difference between sampled fragment
and currently processed fragment
41
Ambient occlusion values in red color channel
Source: www.gamerendering.com
SSAO With Normals
► First pass:
► Render scene normally and copy z values to g-buffer’s alpha
channel and scene normals to g-buffer’s RGB channels
► Second pass:
► Use normals and z-values to compute occlusion between
current pixel and several samples around that pixel
42
With SSAO
No SSAO
SSAO Discussion
► Advantages:
► Deferred rendering algorithm: independent of scene complexity
► No pre-processing, no memory allocation in RAM
► Works with dynamic scenes
► Works in the same way for every pixel
► No CPU usage: executed completely on GPU
► Disadvantages:
► Local and view-dependent (dependent on adjacent texel depths)
► Hard to correctly smooth/blur out noise without interfering with depth
discontinuities, such as object edges, which should not be smoothed out
43
References
► Nvidia’s documentation:
► http://developer.download.nvidia.com/SDK/10.5/direct3d/Sourc
e/ScreenSpaceAO/doc/ScreenSpaceAO.pdf
► SSAO shader code from Crysis:
► http://69.163.227.177/forum.php?mod=viewthread&tid=772
► Another implementation:
► http://www.gamerendering.com/2009/01/14/ssao/
44
Lecture Overview
► Deferred Rendering Techniques
► Deferred Shading
► Screen Space Ambient Occlusion
► Bloom
► Glow
► The Future of Computer Graphics
45
Bloom Effect
► Bloom gives a scene a look of bright lighting and
overexposure
46
Left: no bloom, right: bloom.
Source: http://jmonkeyengine.org
Bloom Shader
► Post-processing filter: applied after scene is
rendered normally
► Step 1: Extract all highlights of the rendered
scene, superimpose them and make them
more intense
► Operates on back buffer
► Often done with off-screen buffer smaller
than frame buffer
► Highlights found by thresholding luminance
► Step 2: Blur off-screen buffer, e.g., with
Gaussian blurring
► Step 3: Composite off-screen buffer with
back buffer
47
Bloom shader render steps.
Source: http://www.klopfenstein.net
References
► Bloom Shader
► http://www.klopfenstein.net/lorenz.aspx/gamecomponents
-the-bloom-post-processing-filter
► GLSL Shader for Gaussian Blur
► http://www.ozone3d.net/tutorials/image_filtering_p2.php
48
Lecture Overview
► Deferred Rendering Techniques
► Deferred Shading
► Screen Space Ambient Occlusion
► Bloom
► Glow
► The Future of Computer Graphics
49
Glow Effects
► Glows and halos of light appear
everywhere in the world
► They provide powerful visual cues
about brightness and atmosphere
► In computer graphics, the intensity
of light reaching the eye is limited, so
the only way to distinguish intense
sources of light is by their
surrounding glow and halos
► In everyday life, glows and halos are
caused by light scattering in the
atmosphere or within our eyes
50
A cityscape with and without glow.
Source: GPU Gems
Glow vs. Bloom
► Bloom filter looks for highlights automatically, based on a
threshold value
► If you want to have more control over what glows and
does not glow, a glow filter is needed
► Glow filter adds an additional step to Bloom filter: instead
of thresholding, only the glowing objects are rendered
► Render passes:
► Render entire scene back buffer
► Render only glowing objects to a smaller off-screen glow buffer
► Apply a bloom pixel shader to glow buffer
► Compose back buffer and glow buffer together
51
References
► GPU Gems Chapter on Glow
► http://http.developer.nvidia.com/GPUGems/gpugems_ch21
.html
► Bloom and Glow
► http://jmonkeyengine.org/wiki/doku.php/jme3:advanced:bloom_
and_glow
52
The Future of Computer Graphics
► ACM SIGGRAPH Asia, 19.11.-22.11.2014 in Hong Kong
(3:18)
► http://www.youtube.com/watch?v=FUGVF_eMeo4
► ACM SIGGRAPH, August 10-14, 2014, Vancouver
► Student volunteer application deadline: Feb 9, 2014
► Cryengine 4 Trailer
► http://www.youtube.com/watch?v=aseq4T81P7g
53
Good luck with your final projects!
54
