Smart Hardware-Accelerated Volume Rendering

1
Smart Hardware-Accelerated Volume Rendering

• Stefan Roettger
• Stefan Guthe
• Daniel Weiskopf

2
Overview

• Current State of the Art in Direct Volume
  Rendering
• What can be improved?
• Segmentation
• HW Raycasting

3
Direct Volume Rendering

• 3D Slicing Approach (Akeley)
• Pre-Integration (Roettger 00, Engel 01)
• Material Properties (Meissner 02)
• Hardware-Accelerated Pre-Integration (Roettger
  02, Guthe 02)
• Volume Clipping (Weiskopf 02)

4
What is missing?

• From a medical point of view
• Pre-integration is difficult to apply to
  segmented medical data
• Pre-integration quality is still not good enough
• 8 bit frame buffer produces artifacts

5
Pre-Integration

• Ray integral depends on three variables Sf, Sb,
  and l, where l is assumed to be constant
• Pre-compute a table for all combinations of Sf
  and Sb and store in 2D dependent texture

6
Volume Clipping

• Use additional 3D clip volume C(x,y,z)
• Iso surface defines clip geometry
• Adjust Sf, Sb, and l according to clip volume
• for the case Cflt0.5ltCb
• w Cb -0.5 /Cb -Cf
• Sf (1-w)SbwSf
• l lw

C0.5
7
Pre-Integrated Segmentation

• Segmentation of two materials
• Define second transfer function
• In the pixel shader
• Make a lookup in TF1
• for the blue area
• Blend with the lookup in TF2
• for the grey area

C0.5
8
Quality Comparison
clipped Bonsai
with naive clipping
with correct adjust- ment
9
Supersampling Quality
10
Supersampling Quality
11
Supersampling Quality
12
Supersampling Quality
13
Supersampling Quality
14
Supersampling Quality
15
Drawback of Pre-Integration

• Linear interpolation assumed in slab
• But in fact the interpolation is trilinear
• Inside the slab one may cross a voxel boundary
• Lighting is also non-linear
• Conclusion For superior quality one needs at
  least 2-times, better 4-times oversampling

16
Ray Casting

• Supersampling is slow, but fortunately we do not
  need to supersample everywhere
• Define importance volume which tells where to
  sample more precisely
• Depends on 2nd derivative of scalar volume and
  1st derivative of TF
• Perform adaptive ray casting on the graphics
  hardware

17
Hardware-Accelerated Ray Casting

• Can be done on the ATI Radeon 9700 with multiple
  floating point render targets
• Need to process all pixels at once
• Cannot exploit ray coherence
• Early ray termination by hierarchical Z-test
• Adaptive sampling includes space leaping
• Stop if all pixels are terminated (asynchronous
  occlusion query)

18
Hardware-Accelerated Ray Casting

• Store ray parameter to determine actual position
• Complete PS 2.0 code given in the paper

19
Quality Comparison
4-times oversampling 8 bit frame buffer
HW ray casting full floating point RT
20
Performance

• Same performance as 4 times over-sampling
• But much better quality (about 2s/frame)

21
ANTSCD Bonsai Examples
Note Raw data of all three Bonsai is available
on my homepage
22
Conclusions

• With respect to quality HW Raycasting is superior
  to the traditional slicing approach
• By reducing the number of adaptive samples
  interactive frame rates can be achieved easily
  without compromising quality
• Now switching to Stefan Guthe who has preparerd
  THE LIVE DEMO

23
Fin

• Thanks for your attention!
• For further reading see my home page
• GoogleBeHappy Stefan Roettger