Mesh update at runtime

[davide445]

In an upcoming project we are discussing the possibility to integrate a SLAM  sensor outcome (both 3d data and location/direction) into a 3d visualization with data augmenting real visuals.

Just as example looking at this video you see a mesh is created during person walk in the location.

Since SLAM sensor will export 3d data in form of point cloud or mesh, was wondering if will be possible to update at runtime a visualized mesh, or we will need to reload every few frames the whole new created mesh.

 

Edited April 8, 2020 by davide445

[morbid]

Hello Davide,

It depends on the design. Unigine Engine has dynamic meshes that can be modified in runtime. You can either use SLAM data to build such mesh via the engine API in real-time or convert and import already completed mesh into the engine.

[davide445]

@morbidsince SLAM algo will result in a point cloud, will be possible to implement a dynamically loaded point cloud, or will be better to first transform all in mesh (but we will need to manage the related meshing problems)?

Was reading about a couple of forumeers working on point cloud visualization, but not sure if their solution might be useful for a dynamic loading.

 

 

[silent]

Hi Davide,

It's not possible to say which approach will fit your needs, the information provided is not enough. You need test different theories and see how it goes in practice. Maybe some hybrid rendering will be better (like meshes for small and non-detailed objects and gpu-particles for other).