Discard the cell-portal and start loose octal and hierarchical encirclement.

If you want to ask what is the most efficient organization and Rendering Method for relatively large-scale and complex indoor scenarios, cell-portal and derived PVS will certainly be the best choice, this is also the reason why a series of cell-portals such as BSP and BW will be viewed at the beginning. The eight forks and hierarchies have not been paid enough attention.

Last weekend, I looked at the scenarios to be separated and found that there were basically a lot of unclosed spaces in the open space, and some scenarios did not even properly arrange the portal. For non-closed spaces, the advantages of cell-portal itself are completely lost. Even if BSP can be used for such a space, it is difficult to easily split the space into what we hope. Because cameras are basically isolated from each other, these segmentation methods are meaningless.

The boss said that in this case, it can only be surrounded by eight forks and layers.

It is true that there are many, and many deficiencies in the eight forks and hierarchies. Few may consider splitting them as space: Compared with BSP, the details they may produce may be even more terrible: A large number of small spaces, each containing a small number of triangles, and these triangles may belong to multiple materials and textures. Compared with these disadvantages, the advantages of allowing a large space change are like chicken ribs, which makes no sense at all.

However, this is, after all, a "abnormal" World, and in a "abnormal" world, any "correct" approach is wrong. Therefore, we may also be forced to use it like this.Algorithm-- You know it is inefficient, but you cannot find a more efficient method. Therefore, it is the best. I feel so evil ......

Back to the last century, back to the last era, and re-pick up the eight forks and hierarchical Encirclement-these two algorithms that have been widely used in Object Management so far, but have already been out of the game in Scenario Management, it is used to meet the scenario segmentation requirements. Pai_^

If there are more abnormalities, it will not be abnormal. But is this really a good scenario ?...... After all, that's not the question I should ask.

Perhaps such doubts and struggles are unnecessary, or the sentence:

"In scenario segmentation, there is no optimal solution, but only the optimal solution ."

Hope we can end it as soon as possible ......