Best Practices & Tips
mika edited this page Jul 12, 2022
·
4 revisions
--WIP--
- If you cannot convert large LAZ/LAS files, split it to multiple las files, then use batch convert option (to convert multiple smaller las files into single V3 pcroot cloud)
- Use smaller grid size, if single tile contains too many points (you can view point count per tile, from the .pcroot header file), each row shows filename and then point count, ideally one tile shouldnt contain several millions of points. see format: https://github.com/unitycoder/UnityPointCloudViewer/wiki/Binary-File-Format-Structure#custom-v3-tiles-pcroot-and-pct-rgb
- Adjust viewer start distance to 0, and end distance to some suitable value, try 500m for example
- Use strong falloff should be enabled usually (to filter points more aggressively)
- GPU upload steps, 0 = upload all data in 1 frame (slower performance), 4 = spread upload in 4 frames (better performance, but tiles update slower)
- Tile resolution can be lowered to test how it looks 1 = full resolution (but rather filter the point cloud first, before bringing into unity)
- Using Unity 2019.1 or later, should enable useNativeArrays and releaseTileMemory (to cleanup memory if tile becomes invisible)
- Use color packing, in the external converter its -pack=true https://github.com/unitycoder/PointCloudConverter/wiki/Commandline-Arguments and then use V3-packed material in the viewer
- You can use Unity Occlusion Culling to hide tiles behind objects https://docs.unity3d.com/Manual/OcclusionCulling.html