Sample culling based on linearized 5D edge equations

• US 9,569,878 B2
• Filed: 12/07/2011
• Issued: 02/14/2017
• Est. Priority Date: 12/07/2011
• Status: Active Grant

• Current Assignee: Intel Corporation
• Original Assignee: Intel Corporation
• Inventors: Munkberg, Carl J., Clarberg, Franz P.
• Primary Examiner(s): Chin, Michelle
• Application Number:
  US13/993,729
• Publication Number:
  US 20130265301A1
• Time in Patent Office: 1,896 Days
• Field of Search: None
• US Class Current: 1/1
• CPC Class Codes: G06T 15/00 3D [Three Dimensional] image rendering
• CPC Class Codes: G06T 15/005 General purpose rendering architectures
• CPC Class Codes: G06T 15/40 Hidden part removal

Abstract

Thin invention introduces a five-dimensional rasterization technique that uses a test based on triangle edges in order to obtain high efficiency. A compact formulation of five-dimensional edge equations is used to derive a conservative triangle edge versus tile test in five dimensions, expressed as an affine hyperplane.

Claims

1. A method comprising:

using a processor to generate an affine hyperplane approximation for a tile to cull samples;

using a processor to cull a tile if a hypercube of valid samples is entirely outside the hyperplane; and

using a processor to pick a corner of the hypercube farthest from the negative direction of the hyperplane normal.

2. The method of claim 1 including formulating bounds as a hyperplane per primitive edge.

3. The method of claim 2 including reducing to an affine hyperplane in uvxyt space where u and v are lens coordinates, t is time, x and y are screen space spatial coordinates.

4. The method of claim 3 including using said affine hyperplane bounds to bound the individual ranges of x, y, u, v, t for each primitive edge.

5. The method of claim 2 including splitting the five dimensional space into two half spaces using said hyperplane and cull all samples falling in one of the halfspaces.

6. The method of claim 5 including recomputing the said hyperplane for each screen space tile.

7. The method of claim 6 including culling samples by evaluating a lower bound of the hyperplane within a screen space tile.

8. The method of claim 7 including evaluating the hyperplane for a more coarse screen space tile, and reusing at a finer screen space tile to cull samples.

9. The method of claim 8 including completing a full visibility test only for samples that are not culled.

10. The method of claim 1 including using bounds based on triangle edges as a triangle undergoes both depth of field and motion blur.

11. The method of claim 1 including detecting a bounding function for a screen space tile, testing samples within the tile against the bounding function and culling samples outside the bounding function.

12. A non-transitory computer readable medium storing instructions executed by a graphics processor to:

generate an affine hyperplane approximation for a tile to cull samples;

cull a tile if a hypercube of valid samples is entirely outside the hyperplane; and

pick a corner of the hypercube farthest from the negative direction of the hyperplane normal.

13. The medium of claim 12 further storing instructions to formulate bounds as a hyperplane per primitive edge.

14. The medium of claim 13 further storing instructions to reduce to an affine hyperplane in uvxyt space where u and v are lens coordinates, t is time, x and y are screen space spatial coordinates.

15. The medium of claim 14 further storing instructions to use said affine hyperplane bounds to bound the individual ranges of x, y, u, v, t for each primitive edge.

16. The medium of claim 13 further storing instructions to split the five dimensional space into two half spaces using said hyperplane and cull all samples falling in one of the halfspaces.

17. The medium of claim 16 further storing instructions to recompute said hyperplane for each screen space tile.

18. The medium of claim 17 further storing instructions to cull samples by evaluating a lower bound of the hyperplane within a screen space region.

19. The medium of claim 17 further storing instructions to complete a full visibility test only for samples that are not culled.

20. The medium of claim 12 further storing instructions to use bounds based on triangle edges as a triangle undergoes both depth of field and motion blur.

21. The medium of claim 12 further storing instructions to detect a bounding function for a screen space tile, test samples within the tile against the bounding function and cull samples outside the bounding function.

22. An apparatus comprising:

a graphics processor to generate an affine hyperplane approximation for a tile to cull samples, cull a tile if a hypercube of valid samples is entirely outside the hyperplane, pick a corner of the hypercube farthest from the negative direction of the hyperplane normal; and

a storage coupled to said processor.

23. The apparatus of claim 22, said processor to formulate bounds as a hyperplane per primitive edge.

24. The apparatus of claim 23, said processor to split the five dimensional space into two half spaces using said hyperplane and cull all samples falling in one of the halfspaces.

25. The apparatus of claim 23, said processor to reduce to an affine hyperplane in uvxyt space where u and v are lens coordinates, t is time, x and y are screen space spatial coordinates.

26. The apparatus of claim 25, said processor to recompute said hyperplane for each screen space tile.

27. The apparatus of claim 25, said processor to use said affine hyperplane bounds to bound the individual ranges of x, y, u, v, t for each primitive edge.

28. The apparatus of claim 27, said processor to cull samples by evaluating a lower bound of the hyperplane within a screen space region.

29. The apparatus of claim 22, said processor to use bounds based on triangle edges as a triangle undergoes both depth of field and motion blur.

30. The apparatus of claim 29, said processor to detect a bounding function for a screen space tile, test samples within the tile against the bounding function and cull samples outside the bounding function.

31. The apparatus of claim 30, said processor to complete a full visibility test only for samples that are not culled.