Accelerating Volume rendering by Space-Leaping []

Similar approach that in []. There is a great deal of coherence between any pair of consecutive images ( few rendering parameters change).
They save the coordinates of the first non-transparent voxel encountered by the ray emitted at each pixel. The save it in the ``C-Buffer''( See figure ).
The change in position, color, or intensity of a light source is accelerated directly for the data set. Rotation of the world space implies the C-Buffer[i,j] coordinates are transform to C'-buffer[i',j'].

  
Figure: 2D illustration of the passage of rays through the grid when employing space-leaping

Then we have two cases:

1. Empty spaces just trace a ray again.
2. More than one sample per pixel.
   • Z-buffer doesn't work. This approach fails due to discretization artifacts. Some different objects can be mapped to the same coordinates.
   • Solution to the previous they use the approach of [].
   • Solution to the previous splatting approach.
   • The rays that hit no object have background and no value in the C-Buffer. Bounding Boxes or their solution the back-faces are objects (They must be removed when they become a front face).

Difference between [] and []:

• [] have image degradation can only support first opaque ray traversal.
• [] no image degradation and independent of the image composition.
• [] just rotation.
• [] enables rotation and change in the shading parameters.