Experiments

For a couple of weeks, I have been experimenting with raycasting algorithms to render my compressed voxel format. Initial tests were unpromising but after trying various approaches, I have now found an algorithm which allows a few hundred raycasts per second. The speed of the algorithm increases as the frequency of the data decreases. Assuming I can achieve even half that on the GPU per Pixel (by performing the raycasts in parallel), I should easily achieve interactive frame rates.

I discovered the algorithm by drawing and analysing an image of a Segment which covers many layers of Z:



This demonstrates all the unique properties of my compression. These are:

• Any Segment is known to be inside the bounds of (0, 0, SegmentStartZ) and(MaxX, MaxY, SegmentEndZ)

• If SegmentStartZ is Equal to SegmentEndZ, the Segment is known to be Inside the Bounds of (0, SegmentStartY, SegmentStartZ) and(MaxX, SegmentEndY, SegmentEndZ)

• If SegmentStartY is Equal to SegmentEndY, the Segment is known to be Inside the Bounds of (SegmentStartX, SegmentStartY, SegmentStartZ) and(SegmentStartX, SegmentEndY, SegmentEndZ)

• When SegmentStartZ is Not Equal to SegmentEndZ, If the RayHitZ is Not Equal to SegmentStartZ or SegmentEndZ an intersection exists.

• When SegmentStartY is Not Equal to SegmentEndY, If the RayHitY is Not Equal to SegmentStartY or SegmentEndY an intersection exists.

• When SegmentStartY is Equal to SegmentEndY, an intersection exists.