There are some bottlenecks which have a drastic impact on the performance:
-Exporting frames causes a cpu stall followed by a gpu stall for the next frame. This problem can be solved by a more lazy approach of using fences and the rendering thread frequently checking if new copied textures are availiable.
-All objects are rendered on all cubemap faces, making the vertex rate the only bottleneck for large meshes which can be improved by sorting all faces into a Kd tree and rendering only visible cubes.
There are some bottlenecks which have a drastic impact on the performance:
-Exporting frames causes a cpu stall followed by a gpu stall for the next frame. This problem can be solved by a more lazy approach of using fences and the rendering thread frequently checking if new copied textures are availiable.
-All objects are rendered on all cubemap faces, making the vertex rate the only bottleneck for large meshes which can be improved by sorting all faces into a Kd tree and rendering only visible cubes.