Processing system for streaming volumetric video to a client device
A network processing system obtains a viewport of a client device for volumetric video and a two-dimensional (2D) subframe of a frame of volumetric video is obtained associated with the viewport. Viewports may be obtained from the client device or be predicted. 2D subframes and reduced resolution versions of frames can be transmitted to the client device. A client device may request volumetric video from the network processing system and provides a viewport to the network processing system. The client device may obtain from the network processing system reduced resolution versions of volumetric video frames and 2D subframes in accordance with the viewport. The client device may determine whether a current viewport matches the viewport associated with the obtained 2D subframe and provides a display based on either that subframe (upon a match) or a 2D perspective of the reduced resolution frame associated with the current viewport (if no match).
1. A method comprising:
predicting, by a server including at least one processor, a viewport needed by a client device for a volumetric video, the predicted viewport comprising a viewing perspective for use in rendering two-dimensional content based on three-dimensional content;
generating, by the server, a two-dimensional subframe from a frame of the volumetric video, wherein the two-dimensional subframe is generated based upon the predicted viewport of the client device and comprising a two-dimensional view of the frame of the volumetric video as such frame appears from the viewing perspective of the predicted viewport;
generating, by the server, a three-dimensional reduced resolution version of the frame of the volumetric video, wherein the three-dimensional reduced resolution version of the frame of the volumetric video is of a reduced video quality compared with the frame of the volumetric video, and wherein the three-dimensional reduced resolution version of the frame of the volumetric video is generated without use of the predicted viewport; and
transmitting, by the server to the client device, the two-dimensional subframe and the three-dimensional reduced resolution version of the frame of the volumetric video.
2. The method of claim 1 , wherein the predicted viewport is predicted in accordance with a machine learning model.
3. The method of claim 2 , wherein the predicted viewport is predicted in accordance with the machine learning model based upon at least one viewport received from the client device.
4. The method of claim 1 , further comprising:
obtaining the frame of the volumetric video.
5. The method of claim 4 , wherein the frame of the volumetric video is obtained as part of at least a portion of the volumetric video comprising a plurality of encoded frames, wherein the method further comprises:
decoding the frame of the volumetric video from the plurality of encoded frames.
6. The method of claim 4 , wherein the generating the three-dimensional reduced resolution version of the frame comprises:
performing an interframe encoding between the three-dimensional reduced resolution version of the frame and at least one additional three-dimensional reduced resolution version of at least one additional frame of the volumetric video.
7. The method of claim 4 , wherein the generating the three-dimensional reduced resolution version of the frame comprises applying an octree-based compression.
8. The method of claim 1 , wherein the three-dimensional reduced resolution version of the frame is generated based upon a throughput between the client device and the server.
9. The method of claim 1 , wherein the frame of the volumetric video comprises:
a point cloud; or
a three-dimensional mesh.
10. The method of claim 1 , where the generating the two-dimensional subframe comprises generating a plurality of two-dimensional subframes, wherein each of the plurality of two-dimensional subframes is generated from a respective time sequential frame of the volumetric video, the method further comprising:
applying an interframe encoding to the plurality of two-dimensional subframes.
11. The method of claim 1 , further comprising:
caching, by the server, at least one of:
the two-dimensional subframe; or
the three-dimensional reduced resolution version of the frame.
12. The method of claim 1 , wherein the predicted viewport comprises a position and an orientation.
13. The method of claim 12 , wherein the orientation comprises a yaw, a pitch, and a roll.
14. The method of claim 4 , wherein the generating the three-dimensional reduced resolution version of the frame comprises applying compression.
15. A server comprising:
a processor; and
a computer-readable medium storing instructions which, when executed by the processor, cause the processor to perform operations, the operations comprising:
predicting a viewport needed by a client device for a volumetric video, the predicted viewport comprising a viewing perspective for use in rendering two-dimensional content based on three-dimensional content;
generating a two-dimensional subframe from a frame of the volumetric video, wherein the two-dimensional subframe is generated based upon the predicted viewport of the client device and comprising a two-dimensional view of the frame of the volumetric video as such frame appears from the viewing perspective of the predicted viewport;
generating a three-dimensional reduced resolution version of the frame of the volumetric video, wherein the three-dimensional reduced resolution version of the frame of the volumetric video is of a reduced video quality compared with the frame of the volumetric video, and wherein the three-dimensional reduced resolution version of the frame of the volumetric video is generated without use of the predicted viewport; and
transmitting, to the client device, the two-dimensional subframe and the three-dimensional reduced resolution version of the frame of the volumetric video.
16. The server of claim 15 , wherein the predicted viewport is predicted in accordance with a machine learning model.
17. The server of claim 15 , wherein the three-dimensional reduced resolution version of the frame is generated based upon a throughput between the client device and the server.
18. A non-transitory computer-readable medium storing instructions which, when executed by a server including at least one processor, cause the server to perform operations, the operations comprising:
predicting a viewport needed by a client device for a volumetric video, the predicted viewport comprising a viewing perspective for use in rendering two-dimensional content based on three-dimensional content;
generating a two-dimensional subframe from a frame of the volumetric video, wherein the two-dimensional subframe is generated based upon the predicted viewport of the client device and comprising a two-dimensional view of the frame of the volumetric video as such frame appears from the viewing perspective of the predicted viewport;
generating a three-dimensional reduced resolution version of the frame of the volumetric video, wherein the three-dimensional reduced resolution version of the frame of the volumetric video is of a reduced video quality compared with the frame of the volumetric video, and wherein the three-dimensional reduced resolution version of the frame of the volumetric video is generated without use of the predicted viewport; and
transmitting, to the client device, the two-dimensional subframe and the three-dimensional reduced resolution version of the frame of the volumetric video.
19. The non-transitory computer-readable medium of claim 18 , wherein the predicted viewport is predicted in accordance with a machine learning model.
20. The non-transitory computer-readable medium of claim 18 , wherein the three-dimensional reduced resolution version of the frame is generated based upon a throughput between the client device and the server.