Current implementation of SphereViewer inspired by PanoDotNet project maps the image onto a geometric mesh approximation of a sphere:
This approach has quite a few downsides. As described in #4, straight lines in the images can be deformed when the number of mesh fragments is too low. As described in #2, nadir and zenith is deformed when the number of mesh fragments is too low. Obviously these issues could be partially resolved by increasing the number of mesh fragments but that would decrease the performance of 360° projection at the same time 🤷🏻♂️
So I guess it's time to abandon the current implementation (which by the way makes it also hard to support huge panoramas in #1) and look for a better alternative.
I would like to experiment with algorithm used in pannellum which I'm using to display equirectangular images on my website and which does not seem to suffer from the issues described above.
Current implementation of
SphereViewerinspired by PanoDotNet project maps the image onto a geometric mesh approximation of a sphere:This approach has quite a few downsides. As described in #4, straight lines in the images can be deformed when the number of mesh fragments is too low. As described in #2, nadir and zenith is deformed when the number of mesh fragments is too low. Obviously these issues could be partially resolved by increasing the number of mesh fragments but that would decrease the performance of 360° projection at the same time 🤷🏻♂️
So I guess it's time to abandon the current implementation (which by the way makes it also hard to support huge panoramas in #1) and look for a better alternative.
I would like to experiment with algorithm used in pannellum which I'm using to display equirectangular images on my website and which does not seem to suffer from the issues described above.