1. All the videos are in QuickTime .mov format. The player can be downloaded from: http://www.apple.com/quicktime/download/ 2. All the videos were captured in real time from our system, however, the real time rendering has been slowed down by the video capturing software. The actual rendering speed (without video capturing) is faster than the displayed fps at the top-left corner. More accurate results on the speed performance are provided in the paper. 3. Due to the limitation of the video capturing software, which captured only 8 frames per second, as a result, there is some leaping in the video, which actually does not exist in our system. 4. Please note the change of the T_min and the number of active blocks (displayed at the top of the video) when the transfer function is frequently modified by the user. 5. The purpose of these videos is to demonstrate: 1) speedup_factor: this video switches between our algorithm and the benchmark algorithm several times at runtime, in order to let users to feel the different speeds of them. The displayed parameter, ProxyCube, shows which algorithm is used currently at each frame: ˇ°ProxyCube=0ˇ± means our algorithm is used, while ˇ°ProxyCube=1ˇ± means the benchmark is used. 2) all the other videos show the rendering quality and speed when transfer function and view parameters are changed dynamically by user interaction.