Geometry conversion for 360-degree video coding
Processing a 360-degree video content for video coding may include receiving the video content in a first geometry. The video content may include unaligned chroma and luma components associated with a first chroma sampling scheme. The unaligned chroma and luma components may be aligned to a sampling grid associated with a second chroma sampling scheme that has aligned chroma and luma components. A geometric conversion to the video content may be performed. The video content, that may comprise the aligned chroma and luma components, in the first geometry may be converted to a second geometry. The first geometry may be a stitched geometry, and the second geometry may be a coding geometry. The converted video content in the second geometry may include the chroma and luma components aligned to the sampling grid associated with the second chroma sampling scheme.
1 . A method for video decoding comprising:
obtaining a video content, wherein the video content comprises a 360-degree video content, wherein the 360-degree video content is associated with a first projection format;
based on the obtained video content, identifying a current sample location, wherein the current sample location is associated with a current sample of the video content;
determining whether the current sample location is located at a content boundary;
based on a determination that the current sample location is located at the content boundary, obtaining a spatially neighboring sample location, wherein the spatially neighboring sample location is located outside of the content boundary and is spatially continuous to the current sample location; and
converting the first projection format to a second projection format based on the obtained spatially neighboring sample location.
2 . The method of claim 1 , wherein the spatially neighboring sample location is obtained based on a circular characteristic associated with the 360-degree video content.
3 . The method of claim 1 , wherein the first projection format comprises an equirectangular projection format and the second projection format comprises a cubemap projection format.
4 . The method of claim 1 , wherein the content boundary comprises at least one of a frame packed picture boundary or a face boundary.
5 . The method of claim 1 , wherein the video content is associated with a frame-packed picture comprises a plurality of faces that is associated with the first projection format, and wherein the method further comprises:
obtaining a first face that includes the current sample location;
obtaining a 3D position of the spatially neighboring sample location, wherein the spatially neighboring sample location is associated with the first face;
based on the 3D position of the spatially neighboring sample location, obtaining a second face that includes a padding sample location; and
applying a geometry projection with the 3D position of the spatially neighboring sample location to derive a 2D planar position of the spatially neighboring sample location in the second face as the padding sample location.
6 . The method of claim 1 , wherein the spatially neighboring sample location is associated with a spatially neighboring sample, and wherein the method comprises:
based on the obtained spatially neighboring sample location, determining that the spatially neighboring sample is misaligned with the current sample; and
based on the determination that the spatially neighboring sample is misaligned with the current sample, applying an interpolation filter to obtain the spatially neighboring sample.
7 . The method of claim 1 , wherein a spatially neighboring sample is associated with the spatially neighboring sample location and shares the same content boundary as the current sample.
8 . An apparatus for video decoding comprising:
a processor configured to:
obtain a video content, wherein the video content comprises a 360-degree video content, wherein the 360-degree video content is associated with a first projection format;
based on the obtained video content, identify a current sample location, wherein the current sample location is associated with a current sample of the video content;
determine whether the current sample location is located at a content boundary;
based on a determination that the current sample location is located at the content boundary, obtain a spatially neighboring sample location, wherein the spatially neighboring sample location is located outside of the content boundary and is spatially continuous to the current sample location; and
based on the obtained spatially neighboring sample location, convert the first projection format to a second projection format.
9 . The apparatus of claim 8 , wherein the spatially neighboring sample location is obtained based on a circular characteristic associated with the 360-degree video content.
10 . The apparatus of claim 8 , wherein the first projection format comprises an equirectangular projection format and the second projection format comprises a cubemap projection format.
11 . The apparatus of claim 8 , wherein the content boundary comprises at least one of a frame packed picture boundary or a face boundary.
12 . The apparatus of claim 8 , wherein the video content is associated with a frame-packed picture comprises a plurality of faces associated with the first projection format, and wherein the processor is further configured to:
obtain a first face that includes the current sample location;
obtain a 3D position of the spatially neighboring sample location, wherein the spatially neighboring sample location is associated with the first face;
based on the 3D position of the spatially neighboring sample location, obtain a second face that includes a padding sample location; and
apply a geometry projection with the 3D position of the spatially neighboring sample location to derive a 2D planar position of the spatially neighboring sample location in the second face as the padding sample location.
13 . The apparatus of claim 8 , wherein the spatially neighboring sample location is associated with a spatially neighboring sample, and wherein the processor is configured to:
based on the obtained spatially neighboring sample location, determine that the spatially neighboring sample is misaligned with the current sample; and
based on the determination that the spatially neighboring sample is misaligned with the current sample, apply an interpolation filter to obtain the spatially neighboring sample.
14 . A method for video encoding comprising:
obtaining a video content, wherein the video content comprises a 360-degree video content, wherein the 360-degree video content is associated with a first projection format;
based on the obtained video content, identifying a current sample location, wherein the current sample location is associated with a current sample of the video content;
determining whether the current sample location is located at a content boundary;
based on a determination that the current sample location is located at the content boundary, obtaining a spatially neighboring sample location, wherein the spatially neighboring sample location is located outside of the content boundary and is spatially continuous to the current sample location; and
including the spatially neighboring sample location in video data.
15 . The method of claim 14 , wherein the spatially neighboring sample is configured to convert the first projection format to a second projection format.
16 . The method of claim 15 , and wherein the first projection format comprises an equirectangular projection format and the second projection format comprises a cubemap projection format.
17 . The method of claim 14 , wherein the spatially neighboring sample location is obtained based on a circular characteristic associated with the 360-degree video content.
18 . The method of claim 14 , wherein the content boundary comprises at least one of a frame packed picture boundary or a face boundary.
19 . The method of claim 14 , wherein the video content is associated with a frame-packed picture comprises a plurality of faces that is associated with the first projection format, and wherein the method further comprises:
obtaining a first face that includes the current sample location;
obtaining a 3D position of the spatially neighboring sample location, wherein the spatially neighboring sample location is associated with the first face;
based on the 3D position of the spatially neighboring sample location, obtaining a second face that includes a padding sample location; and
deriving a 2D planar position of the spatially neighboring sample location in the second face as the padding sample location based on applying a geometry projection with the 3D position of the spatially neighboring sample location.
20 . The method of claim 14 , wherein a spatially neighboring sample is associated with the spatially neighboring sample location and shares the same content boundary as the current sample.