?Bug? Discontinuity in Pflow Particle Rendering

[im.thatoneguy]

I'm getting gaps between GeoSamples in Pflow. It's like it's miscalculating the start position of every geo sample by precisely one width and one height. If I had to guess you've got a counting from 0 vs counting from 1 distance calculation. So instead of calculating the end point as Position2-Position1-Width you're subtracting (2*Width).

[Morbid Angel]

Hi Gavin,

I had a look at your scene. I think the issue is that the particles are traveling very fast, and duration and geo samples do not accomodate for each other. If I set the geo samples to something like 40-50 it looks good.

[vlado]

Dmitry is right; in general the velocity of the particles is only approximate - it doesn't quite exactly tell where the particle will end up in the next time moment. So increasing the geometric samples will help.

This can probably be improved, but the code for calculating velocity in the renderer becomes more complicated (we need to track which particle from one time moment is which particle from the other time moment).

Best regards,
Vlado

[joconnell]

Vlado, just as a general knowledge thing, when vray calculates motion blur is it tracking the vertices of a mesh from where it starts and where it ends and thus the amount of geometry held in memory will have to increase by the amount of geometry steps, almost as if it's duplicating the geometry for where it ends up at the next frame so it can trace vectors for each vertex?

[vlado]

Vlado, just as a general knowledge thing, when vray calculates motion blur is it tracking the vertices of a mesh from where it starts and where it ends and thus the amount of geometry held in memory will have to increase by the amount of geometry steps, almost as if it's duplicating the geometry for where it ends up at the next frame so it can trace vectors for each vertex?

Yes. Which is why it is better to increase the moblur samples only for objects that really need it, rather than globally for the entire scene.

Best regards,
Vlado

[joconnell]

Perfect, was always curious about that. I'd say pflow was an utter pain in the ass to deal with too! Autodesk have messed around with Oleg quite a bit on it!

[im.thatoneguy]

Ok, well the problem is, the more geo-samples I increase the more spotty the sampling gets on really fine particles. So if I leave it at 3 samples I get big gaps but if I crank it to say 32 or 48 I get just general patchy noise that never resolves.

[Morbid Angel]

can you show us that result?