Improve point render performance with spatial index

Hi @swharden & Team,

perhaps I had mentioned it before. The point drawing performance becomes very slow with large dataset’s Here is an example with 2.5 million points. In this view i zoomed in so that about 1000 points are visible:

As you can see, a lot of render time is wasted. Probably all points are rendered? The AddScatterPoints function was used. Is it perhaps useful to use a spatial index like the R-tree? You can think about it 😃

Version: ScottPlot 4.1.9-beta Thank You!

[AlgoExecutor]

Hi @StendProg ,

that’s totally ok. The conversation with you was really nice and I could implement the small quickfix. And as you say, even this can be improved with integer rounding. Would be nice if you could find point cloud solution from which other users of ScottPlot can benefit without writing their own renderer 😉

If I think of anything else on the subject I’ll write it here…

[StendProg]

Hi @AlgoExecutor,

I played a little bit 😃 Just filtering out the points that lie on top of each other brings considerable progress in the initial view. So it is possible to reduce the points from 2.557.503 to 871.267. 66% reduction. Not bad far a start. Code I used:

The solution is very interesting for its simplicity, I think you can get a greater effect if the coordinates are previously rounded to integer values.

It was very pleasant to participate in the discussion and many ideas were expressed. But I guess I’ll take a break until I get to coding and try to dig deeper on my own, otherwise we’ll just talk about what.