A method is described for
processing an omnidirectional video by a
client device, said omnidirectional video being associated with a 3D coordinate
system for determining a spatial position of video data of said omnidirectional video on a curved surface, preferably a sphere, said method comprising: receiving a
manifest file, the
manifest file comprising a plurality of tile
stream identifiers for identifying a plurality of tile streams, the tile streams comprising video frames having image views, whereby the image views of video frames of different tile streams cover different regions of a 2D projection of the omnidirectional video, a region defining a tile; selecting on the basis of
spatial relation information in the
manifest file and on the basis of a viewpoint of a user of the
client device a first tile streams associated with a first resolution and a first tile position and a second tile
stream associated with a second resolution and a second tile position, the second resolution being lower than the first resolution, wherein the
spatial relation information defines for each tile a tile position and wherein the viewpoint defines a viewing direction of the user, tile position and the viewpoint being defined on the basis of coordinates of the 3D coordinate
system.