Adaptive depth of field sampling

• US 8,493,383 B1
• Filed: 12/10/2009
• Issued: 07/23/2013
• Est. Priority Date: 12/10/2009
• Status: Active Grant

• Current Assignee: Pixar (The Walt Disney Company)
• Original Assignee: Pixar (The Walt Disney Company)
• Inventors: Cook, Robert L., Carpenter, Loren
• Primary Examiner(s): Nguyen, Kimbinh T
• Application Number:
  US12/634,960
• Time in Patent Office: 1,321 Days
• Field of Search: 345/419, 345/421, 345/422, 345/423, 345/426, 345/427, 345/428, 345/581, 345/582, 345/589
• US Class Current: 345/419
• CPC Class Codes: G06T 15/06 Ray-tracing
• CPC Class Codes: G06T 15/503 Blending, e.g. for anti-aliasing
• CPC Class Codes: G06T 2200/12 involving antialiasing

Abstract

Adaptive sampling alleviates aliasing by partitioning the field of view of an image sample point into depth regions. Portions of the scene are sampled within a depth region using sample rays. If a sample ray is not completely occluded in the depth region, corresponding sample rays are evaluated in adjacent depth regions. Sample rays can be recursively evaluated in further depth regions until all the subsamples intersect opaque objects or a depth limit or transparency threshold is reached. The value of an image sample point is the weighted combination of sample rays. The number of sample rays in each depth region may increase monotonically with distance along a line of sight from an image sample point for effects such as reflection, refraction, and illumination. The number of sample rays in each depth region may increase monotonically with distance from a focal plane for effects such as depth of field.

Claims

1. A method of generating an image from three-dimensional scene data, the method comprising:

defining a first sampling frustum;

associating a first depth region with the first sampling frustum;

defining a first set of samples within the first depth region adapted to sample a scene to determine at least a first portion of an attribute value of a first image sample;

associating a second depth region with the first sampling frustum;

defining a second set of samples within the second depth region adapted to sample the scene to determine at least a second portion of the attribute value of the first image sample; and

sampling the scene with at least a portion of the first set of samples and a portion of the second set of samples using a processor;

wherein the second set of samples is larger than the first set of samples.

2. The method of claim 1, wherein the second depth region is further from the first image sample than the first depth region.

3. The method of claim 2, wherein distances of the first depth region and the second depth region are based on distances from a plane.

4. The method of claim 2, wherein distances of the first depth region and the second depth region are based on distances along at least one ray associated with the first sampling frustum.

5. The method of claim 1, wherein the first image sample includes a color value in response to at least the first and second sets of samples.

6. The method of claim 1, wherein the first image sample includes a value in response to at least the first and second sets of samples, wherein the value is selected from a group consisting of:

an illumination value;

a reflection value;

a refraction value; and

a scattering value.

7. The method of claim 1, wherein the size of the first and second sets of samples is based on a function of distance from an entity, wherein the entity is selected from a group consisting of:

a focal plane;

a camera viewpoint;

an image plane, a surface point of an object;

a particle;

an interior point of an object;

a volume;

an implicit surface; and

a scene element.

8. The method of claim 7, wherein the first image sample includes a color value in response to a depth of field effect from at least some of the first and second sets of samples.

9. The method of claim 1, wherein the first and second depth regions are located at regular intervals of a function of depth.

10. The method of claim 1, wherein the size of the first and second sets of samples is proportional to the volumes of their associated depth regions.

11. The method of claim 1, wherein the first set of samples are distributed within the first depth region according to a Halton sequence.

12. The method of claim 1, wherein the first set of samples are randomly or pseudo-randomly distributed within the first depth region.

13. The method of claim 1, wherein the first set of samples are distributed deterministically within the first depth region.

14. The method of claim 1, wherein the second set of samples within the second depth region is defined in response to a determination that the first set of samples is not occluded by the scene in the first depth region.

15. The method of claim 1, wherein the first set of samples is associated with at least one sampling ray adapted to intersect at least a portion of the scene to determine the first portion of the attribute value of the first image sample.

16. The method of claim 1, wherein the first set of samples is adapted to sample a rasterized portion of the scene to determine the first portion of the attribute value of the first image sample.

17. The method of claim 1, wherein the second depth region is further from a sample origin than the first depth region, and wherein sampling the scene with at least a portion of the first set of samples and a portion of the second set of samples using a processor comprises:

determining if at least a first sample in the portion of the second set of samples intersects a first scene element in the second depth region;

in response to the determination that the first sample intersects the first scene element in the second depth regions, back-tracing the first sample towards the sample origin at least through the first depth region;

determining if the back-tracing sample intersects a closer scene element in the first depth region; and

in response to the determination that the back-tracing sample intersects the closer scene element in the first depth region, excluding an attribute value of the first scene element from the attribute value of the first image sample.

18. The method of claim 17, wherein the first sample in the portion of the second set of samples does not correspond with any of the first set of samples.

19. A non-transitory computer-readable medium including instructions adapted to direct a computer to perform an operation, the operation comprising:

defining an image plane;

defining a focal plane;

defining a first sampling frustum on a first side of the focal plane between the focal plane and the image plane;

defining a second sampling frustum on a second side of the focal plane, wherein the second side is opposite to the first side

associating at least a first depth region and a second depth region with the first sampling frustum;

associating at least a third depth region and a fourth depth region with the second sampling frustum;

defining a first set of samples associated the first depth region, a second set of samples associated the second depth region, a third set of samples associated the third depth region, and a fourth set of samples associated the fourth depth region, wherein the first, second, third, and fourth sets of samples are adapted to sample a scene to determine at least a first portion of an attribute value of a first image sample; and

wherein sizes of the first, second, third, and fourth sets of samples increases in proportion to distances of their respective associated depth regions from the focal plane.

20. The non-transitory computer-readable medium of claim 19, wherein the distances of the depth regions to the focal plane are based on distances from the focal plane.

21. The non-transitory computer-readable medium of claim 19, wherein distances of the depth regions to the focal plane are based on distances along at least one ray associated with the first and second sampling frustums.

22. The non-transitory computer-readable medium of claim 19, wherein sizes of the first, second, third, and fourth sets of samples increases monotonically as a function of a distance from the focal plane.

23. The non-transitory computer-readable medium of claim 19, wherein:

second depth region and the third depth region are adjacent to the focal plane; and

the size of the second set of samples is different than the size of the third set of samples.

24. The non-transitory computer-readable medium of claim 19, wherein the first image sample includes a color value in response to at least one of the first, second, third, and fourth sets of samples.

25. The non-transitory computer-readable medium of claim 19, wherein the first, second, third, and fourth depth regions are located at regular intervals of a function of depth.

26. The non-transitory computer-readable medium of claim 19, wherein the sizes of the first, second, third, and fourth sets of samples are proportional to the volumes of their associated depth regions.

27. The non-transitory computer-readable medium of claim 19, wherein the first set of samples is distributed within the first depth region according to a Halton sequence.

28. The non-transitory computer-readable medium of claim 19, wherein the first set of samples is randomly or pseudo-randomly distributed within the first depth region.

29. The non-transitory computer-readable medium of claim 19, wherein the first set of samples is distributed deterministically within the first depth region.

30. The non-transitory computer-readable medium of claim 19, wherein the second, third, and fourth sets of samples within their respective associated depth regions are defined in response to a determination that the first set of samples is not occluded by the scene in the first depth region.

31. The non-transitory computer-readable medium of claim 19, wherein the first set of samples is associated with at least one sampling ray adapted to intersect at least a portion of the scene to determine the first portion of the attribute value of the first image sample.

32. The non-transitory computer-readable medium of claim 19, wherein the first set of samples is adapted to sample a rasterized portion of the scene to determine the first portion of the attribute value of the first image sample.