Segmentation tool, performance drop

[PablerasBCN]

Hi,

I think I did read something similar but I could not find, yet the original was that performance degraded as you did segmentation but I noticed that performance drops drastically when entering setmentation mdoe.

It may be more perfectible now becaseu I've a much higher visible points in realtime than I had before because I upgraded the graphics card.

Prior to segmentation, it moves the cloud nicely, as I launch the tool its all hitter when moving the view 2.13 alph, windows. 3090ti.

[daniel]

That is surprising, because on the contrary we have changed the strategy when starting the tool (we don't activate the 'per-point visibility filter' right away). And once you start segmenting, of course this filter is activated as before, this it will slow down the display.

On my side I don't observe any performance drop when starting this tool anymore. Are you sure you are using the latest 2.13.alpha version by the way?

[PablerasBCN]

I will redownload to make sure I'm using latest, wil re post once I do so.

[PablerasBCN]

I downloaded the latest alpha, nice new import window for the scalar fields, by the way.

Starting the tool, does not affect performnce, yet after the first cut, performance drops massively. I don´t recall so agressive performance loss form previous stable version. I'll try with the non alpha and my current max render number of points, since the current issue may be related to the number of points I'm rendering. Since After installing my new card I raised a lot because I don´t like to wait to the octree optimization thing.

[daniel]

I don't remember any particular change on this side. The way points are displayed in this 'partial visibility' mode has always been sub-optimal and ver inefficient.

[PablerasBCN]

I've done further testing,

I've found when this occurs.

as I was supecting, with a high point numbers rendering without thinning something like 170M, while in viewer works nice, after first cut with the segmentation tool, performance collapses.

If I lower the numbers to 70M, when segmenting a 120M cloud, all seems to go smoothly.

I can provide a sample cloud to further narrow what is going on.