Video refinement mechanism

1. An apparatus to facilitate processing video bit stream data, comprising:

one or more processors to decode point cloud data, reconstruct the decoded point cloud data and perform a super-resolution operation to fill one or more holes in reconstructed point cloud frame data using patch metadata and depth data included in the decoded point cloud data, wherein the patch metadata comprises occupancy map data and auxiliary patch information provided as input to upsamplers that perform the super-resolution operation by performing pixel mirroring at patch occupancy edges to provide neighboring pixels input to the upsamplers; and

a memory communicatively coupled to the one or more processors.

2. The apparatus of claim 1 , wherein the super-resolution operation implements a machine-learning algorithm to analyze patches to determine whether a patch has importance to an image to be rendered.

3. The apparatus of claim 1 , wherein the one or more processors further to analyze the patch metadata to determine whether a patch in the decoded cloud data warrants performing the super-resolution process.

4. The apparatus of claim 1 , wherein the one or more processors further to perform preemptive scaling on samples of a captured image.

5. The apparatus of claim 4 , wherein the one or more processors further to perform the preemptive scaling on captured image samples based on a perspective in which the image was captured.

6. The apparatus of claim 4 , wherein performing the preemptive scaling comprises analyzing the samples to determine a relevant distance of objects included in each sample and scaling each of the samples based on a corresponding determined distance.

7. The apparatus of claim 1 , wherein the one or more processors further to apply hysteresis to changing point cloud geometry data to fill the one or more holes due in the point cloud frame data.

8. The apparatus of claim 7 , wherein applying the hysteresis comprises using data from one or more previous point cloud frames to fill the one or more holes due in a current point cloud frame data.

9. The apparatus of claim 8 , wherein applying the hysteresis further comprises storing one or more characteristics corresponding to the data from the one or more previous point cloud frames prior to receiving the current point cloud frame data.

10. The apparatus of claim 1 , wherein the one or more processors further to perform anti-aliasing to fill geometric discontinuities in the point cloud frame data.

11. The apparatus of claim 10 , wherein performing the anti-aliasing comprises performing three-dimensional (3D) anti-aliasing to generate one or more replacement voxels to bridge a geometric discontinuity.

12. The apparatus of claim 1 , wherein the one or more processors further to perform spatio-temporal refining of texture data.

13. The apparatus of claim 12 , wherein performing the spatio-temporal refining comprises generating patch edge texture coordinates.

14. The apparatus of claim 13 , wherein performing the spatio-temporal refining further comprises generating the patch edge by performing a viewpoint transform of an Adaptive Multi-Frequency Shading (AMFS).

15. The apparatus of claim 12 , wherein the one or more processors further to receive the patch edge texture coordinates to generate super resolution textures.

16. A computer generated method to facilitate processing video bit stream data, comprising:

decoding point cloud data;

reconstructing the point cloud data; and

performing a super-resolution operation to fill one or more holes in reconstructed point cloud frame data using patch metadata and depth data included in the decoded point cloud data, wherein the patch metadata comprises occupancy map data and auxiliary patch information provided as input to upsamplers that perform a super-resolution operation by performing pixel minoring at patch occupancy edges based on an importance of an area indicated by the depth data.

17. The method of claim 16 , wherein the super-resolution process implements a machine-learning algorithm to analyze patches to determine whether a patch has importance to an image to be rendered.

18. The method of claim 16 , further comprising performing preemptive scaling on samples of a captured image.

19. The method of claim 18 , wherein performing the preemptive scaling is performed on captured image samples based on a perspective in which the image was captured.

20. The method of claim 19 , wherein performing the preemptive scaling comprises analyzing the samples to determine a relevant distance of objects included in each sample and scaling each of the samples based on a corresponding determined distance.

21. The method of claim 16 , further comprising applying hysteresis to changing point cloud geometry data to fill the one or more holes in the point cloud frame data.

22. The method of claim 21 , wherein applying the hysteresis comprises using data from one or more previous point cloud frames to fill the one or more holes in the current point cloud frame data.

23. The method of claim 22 , wherein applying the hysteresis further comprises storing one or more characteristics corresponding to the data from the one or more previous point cloud frames prior to receiving the current point cloud frame data.

24. The method of claim 16 , further comprising performing three-dimensional (3D) anti-aliasing to generate one or more replacement voxels to bridge a geometric discontinuity in the point cloud frame data.

25. The method of claim 16 , further comprising performing spatio-temporal refining of texture data included in the decoded point cloud data.

26. The method of claim 25 , wherein performing the spatio-temporal refining of the texture data comprises:

generating patch edge texture coordinates; and

generating super resolution textures bases on the patch edge texture coordinates.

27. The method of claim 26 , wherein the patch edge texture coordinates are generated by performing a viewpoint transform of an Adaptive Multi-Frequency Shading (AMFS).