Pre-Integrated Volume Rendering: Past, Present, Future

1. Pre-Integrated Volume Rendering:Past, Present, Future Martin Kraus December 16, 2004

2. Contents • past: • publications summarized in the chapter “Pre-Integrated Volume Rendering” of “The Visualization Handbook.” • present: • new publications in 2004 • future: • open problems/work in progress Pre-Integrate Volume Rendering: Past, Present, Future

3. Pre-Integration: The Past • reference: Kraus and Ertl, “Pre-Integrated Volume Rendering” in Johnson and Hansen (eds.), “The Visualization Handbook”, 2004 • written: Spring 2003 • goals: • structured summary of pre-integrated techniques • consistent notation • show the big picture Pre-Integrate Volume Rendering: Past, Present, Future

4. Pre-Integration: The Past • contents of chapter 10 in the Vis Handbook: • introduction to pre-int. volume rendering • pre-integrated volume rendering algorithms • accelerated pre-integration • pre-integrated rendering techniques • open problems Pre-Integrate Volume Rendering: Past, Present, Future

5. Pre-Integration: The Past • introduction to pre-int. volume rendering • volume rendering integral • pre- and post- classification • numerical integration • pre-integrated classification viewing ray pixel sampling points eye point viewing ray back front sampling sampling point eye point point Pre-Integrate Volume Rendering: Past, Present, Future

6. Pre-Integration: The Past • pre-integrated volume rendering algorithms • pre-integrated cell projection • Max et al. 1990 (VolVis) • Röttger et al. 2000 (Vis), 2002 (VolVis) • Guthe et al. 2002 (HWWS) • Weiler et al. 2002 (VolVis), 2003 (TVCG) • pre-int. texture-based volume rendering • Engel et al. 2001 (HWWS) • ... Pre-Integrate Volume Rendering: Past, Present, Future

7. Pre-Integration: The Past • pre-integrated volume rendering algorithms • ... • pre-integrated ray casting • Knittel 2002 (EG) • Meißner et al. 2002 (HWWS) • Roettger et al. 2003 (VisSym) • Weiler et al. 2003 (Vis) • pre-integrated shear-warp algorithm • Schulze et al. 2003 (VG) Pre-Integrate Volume Rendering: Past, Present, Future

8. Pre-Integration: The Past • accelerated pre-integration • local updates of lookup tables • particular optical models • pre-int. for a single sampling distance • pre-int. without self-attenuation of segments • hardware-accelerated pre-integration • Roettger et al. 2002 (VolVis) • incremental pre-integration • Weiler et al. 2003 (Vis) Pre-Integrate Volume Rendering: Past, Present, Future

9. Pre-Integration: The Past • pre-integrated rendering techniques • isosurface rendering • Röttger et al. 2000 (Vis) • Engel et al. 2001 (HWWS) • volume shading • Meißner et al. 2002 (GI) • volume clipping • Roettger et al. 2003 (VisSym) Pre-Integrate Volume Rendering: Past, Present, Future

10. Pre-Integration: The Past • open problems • multidimensional transfer functions • see also Kniss et al. 2003 (Vis) • other volume rendering algorithms • additional acceleration techniques • further illumination models • higher-order interpolation Pre-Integrate Volume Rendering: Past, Present, Future

11. Pre-Integration: The Present • important techniques published in 2004: • basics • perspective interpolation • partial pre-integration • accelerated pre-integration • incremental pre-integration of 2D tables • logarithmic scale of third dimension • pre-integrated volume lighting Pre-Integrate Volume Rendering: Past, Present, Future

12. Pre-Integration: The Present • perspective interpolation • published by Kraus et al. 2004 (Vis) • implements perspective correct interpolation of sampled data for projected tetrahedra Pre-Integrate Volume Rendering: Past, Present, Future

13. Pre-Integration: The Present • partial pre-integration • published by Moreland, Angel 2004 (VolVis) • pre-integrates linear transfer functions • advantage: • tables are independent of transfer functions • disadvantage: • requires decomposition into linear segments • implemented for projection of tetrahedra Pre-Integrate Volume Rendering: Past, Present, Future

14. Pre-Integration: The Present • incremental pre-integration of 2D tables • published by Lum et al. 2004 (VisSym) • computes entries of 2D lookup tables incrementally along diagonals • can be combined with incremental integration in third dimension (sampling distance) back scalar front scalar Pre-Integrate Volume Rendering: Past, Present, Future

15. Pre-Integration: The Present • logarithmic scale of third dimension • published by Kraus et al. 2004 (Vis) • pre-integrates 3D lookup table with entries for sampling distances 0, d, 2d, 4d, 8d, 16d, ... • uses a fragment program to access table Pre-Integrate Volume Rendering: Past, Present, Future

16. Pre-Integration: The Present • pre-integrated volume lighting • published by Lum et al. 2004 (VisSym) • pre-computes front-weighted and back-weighted transfer functions to interpolate lighting at front and back sampling point back-weighted front-weighted front scalar back scalar Pre-Integrate Volume Rendering: Past, Present, Future

17. Pre-Integration: The Present • pre-integrated volume lighting (continued) Pre-Integrate Volume Rendering: Past, Present, Future

18. Pre-Integration: The Present • current state of the art: • solutions for rendering artifacts available • efficient pre-integration available • combination with volume lighting available • conclusions: • pre-integrated volume rendering with 1D transfer functions is well understood • there are no excuses for bad rendering quality Pre-Integrate Volume Rendering: Past, Present, Future

19. Pre-Integration: The Future • open problems / work in progress • multidimensional transfer functions • pre-integration in image plane • pre-integrated texture sampling Pre-Integrate Volume Rendering: Past, Present, Future

20. Pre-Integration: The Future • multidimensional transfer functions • is a straight line the correct path of integration between (s1f,s2f) and (s1b,s2b)? • yes, for piecewise- linear data, e.g., in tetrahedral meshes s2 (s1b,s2b) ? (s1f,s2f) s1 Pre-Integrate Volume Rendering: Past, Present, Future

21. Pre-Integration: The Future • multidimensional transfer functions (cont.) • but straight lines can cause problems: “fine” sampling “coarse” sampling s2 s2 s2iso s2iso point missed s1 s1 s1iso s1iso Pre-Integrate Volume Rendering: Past, Present, Future

22. Pre-Integration: The Future • multidimensional transfer functions (cont.) • alternative approach: integrate over bounding box of each pair of sampling points • implementable with summed area tables (Green, GDC 2003) s2 s2iso point covered s1 s1iso Pre-Integrate Volume Rendering: Past, Present, Future

23. Pre-Integration: The Future • pre-integration in image plane • problem: viewing rays might miss thin lines • pre-integration in the image plane might be able to catch them s2 s2iso s1 s1iso Pre-Integrate Volume Rendering: Past, Present, Future

24. Pre-Integration: The Future • pre-integrated texture sampling • problem: • mip maps are good for pre-classification • no obvious solution for post-classification • approach: • apply pre-integration to the filtering Pre-Integrate Volume Rendering: Past, Present, Future

25. Pre-Integration: The Future • conclusions: • next step: pre-integration for multidimensional transfer functions • pre-integration might be applicable to many more (filtered) sampling processes in computer graphics Pre-Integrate Volume Rendering: Past, Present, Future