Hi, I am trying figure out, best approach, for 3D flood algorithm. Not asking about algorithm of such, but rather suggestions and / or directions. Just for starter, I have 2D representation (rather 3D) of what I want to flood search. Vid as part of the project, to get better idea, what I am dealing with (ECS). Spoiler So as current, 1. I got neighbouring working. Meaning, each side of 3D box, stores relevant information, regarding neighbour 3D boxes. Hence, I can traverse to every near boxes directly, knowing its location. 2. I have octree, which I can ray cast if needed, or search groups of near 3D boxes. 3. I don't use K-D tree, since I don't use points, but rather variety size 3D cubes. Octree in such case are more suitable, as I believe. Also, my structure can change many times over and often, and can be quite of the size. Hence as far I am concerned, K-D will not be suitable for that reason, as I would need regenerate it every time change occurs. On side note, an interesting article regarding K-D tree. https://blog.mapbox.com/a-dive-into-spatial-search-algorithms-ebd0c5e39d2a So for now, I am thinking of utilizing searching by traverse though neighbours, to build something like https://en.wikipedia.org/wiki/A*_search_algorithm but for 3D. Other than I don't need path finding part at this point. However, while this is rather straight forward for equally spaced grid, my tinkering is around "irregular grid" (see first picture). Alternatively, I am thinking about division of 3D boxes, into smaller equal sized cubes, to put them into the regular 3D grid. That seams resealable, but if I could avoid such, it would be nice. Or, are there better methods? I hope my problem is clear enough, but if not, give me a shout. Any thoughts, suggestions, comments?