Granted Patent B1

US 10,467,775 · App. 15/585,610 · Granted Nov 5, 2019

Inventors: Charles Benjamin Franklin Waggoner (Portland, OR); Yongjun Wu (Bellevue, WA)

Assignee: Amazon Technologies, Inc.

G06T9/20G06T3/005G06T3/0056G06T3/0087G06T19/006

Abstract

Claims (60)

1. A computer-implemented method, comprising:

receiving a two-dimensional (2D) rectangular representation of virtual reality (VR) content, the 2D rectangular representation corresponding to a three-dimensional (3D) projection space and having first pixel locations;

identifying a coordinate transformation function between the 2D rectangular representation and the 3D projection space;

generating second pixel locations from the first pixel locations using the identified coordinate transformation function, the first pixel locations including a first subset corresponding to nondisplay pixels that do not contain image content in the 2D rectangular representation, and the second pixel locations including a second subset corresponding to display pixels in the 3D projection space, wherein the first pixel locations represent 2D coordinates, and wherein the second pixel location represents 3D coordinates;

receiving pixel requests from an encoder, the pixel requests indicating at least one pixel location in the first subset for the 2D representation;

identifying pixels according to the second pixel locations; and

providing, in response to the pixel requests, the identified pixels including at least one pixel value for a pixel location in the second subset instead of a pixel value for a pixel location in the first subset to the encoder.

2. The method of claim 1 , wherein the 3D projection space is one of a cylinder, a tetrahedron, a cube, an octahedron, a dodecahedron, or an icosahedron.

3. A computer-implemented method, comprising:

receiving a two-dimensional (2D) representation of virtual reality (VR) content, the 2D representation corresponding to a three-dimensional (3D) projection space and having first pixel locations;

identifying a coordinate transformation function between the 2D representation and the 3D projection space;

receiving, from an encoder, a first request indicating a first one of the first pixel locations in the 2D representation;

generating a second pixel location from the first one of the first pixel locations using the identified coordinate transformation function, the first one of the first pixel locations corresponding to a nondisplay pixel that does not contain image content in the 2D representation, and the second pixel location corresponding to a display pixel in the 3D projection space, wherein the first pixel locations represent 2D coordinates, and wherein the second pixel location represents 3D coordinates; and

providing, in response to the first request, a pixel value for the second pixel location instead of a pixel value for the first one of the first pixel locations to the encoder.

4. The method of claim 3 , wherein the projection space is a three-dimensional (3D) polygonal projection space, and wherein the transformation function is identified based on the 3D polygonal projection.

5. The method of claim 3 , wherein generating the second pixel location includes:

identifying a vertical position of the first one of the first pixel locations;

determining a curvature of a playback projection at the vertical position of the first pixel location;

determining a horizontal position of the second pixel location according to the curvature; and

identifying the second pixel location using the vertical position and the horizontal position.

6. The method of claim 3 , further comprising:

generating a plurality of second pixel locations from the first pixel locations using the transformation function;

identifying a first set of vertices from the plurality of second pixel locations;

determining that the first set of vertices are associated with a first face of the projection space;

identifying a second set of vertices from the plurality of second pixel locations;

determining that the second set of vertices are associated with a second face of the projection space;

determining that the first set of vertices and the second set of vertices include a shared vertex; and

determining that the first face and the second face are adjacent according to the shared vertex.

7. The method of claim 6 , further comprising:

receiving a request from an encoder for a block of pixels;

identifying pixels according to pixel locations corresponding to the first face; and

providing the identified pixels to the encoder.

8. The method of claim 3 , wherein the projection space is one of an equirectangular projection, a tetrahedron projection, a cubic projection, an octahedron projection, a dodecahedron projection, or an icosahedron projection.

9. The method of claim 3 , wherein a pixel at the second location includes an average pixel value based on two or more pixels corresponding to the first pixel locations.

10. A system, comprising one or more processors and memory configured to:

receive a two-dimensional (2D) representation of virtual reality (VR) content, the 2D representation corresponding to a three-dimensional (3D) projection space and having first pixel locations;

identify a coordinate transformation function between the 2D representation and the 3D projection space;

receive, from an encoder, a first request indicating a first one of the first pixel locations in the 2D representation;

generate a second pixel location from the first one of the first pixel locations using the identified coordinate transformation function, the first one of the first pixel locations corresponding to a nondisplay pixel that does not contain image content in the 2D representation, and the second pixel location corresponding to a display pixel in the 3D projection space, wherein the first pixel locations represent 2D coordinates, and wherein the second pixel location represents 3D coordinates; and

provide, in response to the first request, a pixel value for the second pixel location instead of a pixel value for the first one of the first pixel locations to the encoder.

11. The system of claim 10 , wherein the projection space is a three-dimensional (3D) polygonal projection space, and wherein the transformation function is identified based on the 3D polygonal projection.

12. The system of claim 10 , wherein the one or more processors and memory are configured to generate the second pixel location by:

identifying a vertical position of the first one of the first pixel locations;

determining a curvature of a playback projection at the vertical position of the first pixel location;

determining a horizontal position of the second pixel location according to the curvature; and

identifying the second pixel location using the vertical position and the horizontal position.

13. The system of claim 10 , wherein the one or more processors and memory are configured to:

generate a plurality of second pixel locations from the first pixel locations using the transformation function;

identify a first set of vertices from the plurality of second pixel locations;

determine that the first set of vertices are associated with a first face of the projection space;

identify a second set of vertices from the plurality of second pixel locations;

determine that the second set of vertices are associated with a second face of the projection space;

determine that the first set of vertices and the second set of vertices include a shared vertex; and

determine that the first face and the second face are adjacent according to the shared vertex.

14. The system of claim 13 , wherein the one or more processors and memory are further configured to:

receive a request from an encoder for a block of pixels;

identify pixels according to pixel locations corresponding to the first face; and

provide the identified pixels to the encoder.

15. The system of claim 10 , wherein the projection space is one of an equirectangular projection, a tetrahedron projection, a cubic projection, an octahedron projection, a dodecahedron projection, or an icosahedron projection.

16. The system of claim 10 , wherein a pixel at the second location includes an average pixel value based on two or more pixels corresponding to the first pixel locations.

Assignments (1)

Cited By (14)

US 12,192,617 US 12,314,553 US 12,394,077 US 12,401,889 US 12,495,204 US 12,498,224 US 12,506,953 US 12,530,116 US 12,602,154 US 12,647,673 US 12,659,569 US 12,666,132 US 12,671,891 US 12,701,334