Question about https://developer.blender.org/rB058514aa0ab

Other Development Topics User Feedback
#1

Hey @Hypersomniac have a pair of questions about the point cloud rendering technique decided :slight_smile:

1.- How does this affect to scalability?

I mean, what happens if we have 100 or 200 million points in the point cloud, something quite normal for a granular/sand simulation, could using geometry instead of an analytic shape present some memory trouble?

I mean at render time, at viewport time I imagine it could be optimized by showing less points.

2.- how does it behave and interact with lighting when a huge amount of geometry is being used?

3.- how does it work with vertex color? I mean, a point cloud from photogrammetry could have vertex color implemented, should it bake the color of one vertex to one sphere, or will it do an interpolation between points?

4.- will we have a screen space fixed size option?

Thanks!

#2

1 - Scalability was my first concern. We can change the amount of geometry for all points or even use point rendering if wanted. The geometry is instanciated so it does not present any memory issue.

Cullin points however would be difficult but could be considered as an optimization just like for particle system. The issue might be that the point order might not be that much random (but this is easily fixable).

2 - Just like regular geometry.

3 - There will be one vertex color per sphere.

4 - Yes that’s possible.

1 Like
#3

Awesome, it feels like you are designing a very robust system :slight_smile:

Thanks!