Niagara particle penetration phenomenon

I am developing a function for a vehicle to spawn particles on the snow, but when the vehicle is moving, the phenomenon of particles passing through the mold appears. Even if I enable the particle collision detection module, it will still be like this. (I tried iteratively to switch the channels and presets for collision detection)

When I was debugging, I added two obvious collision boxes and hung the Niagara particle next to the wheel, adjusted its direction, and relieved gravity. I found that when the vehicle is stationary or moving very slowly, the collision detection of the collision box is effective, but once the vehicle starts to move fast, the particles will pass directly.

I checked a lot of information on the Internet and realized that this is related to the updated frame rate. When the particle speed is too fast, the next frame has passed through the collision box, so there is no way to detect and realize the feedback of the collision. I tried to make the collision box very thin, but it still didn’t work. I would like to ask if there is a solution to this situation? If so, what should I do?

Hi there, I think I am encountering the same problem!

When using CPU mesh particles and collision with “CPU collision type” set to “raytraced” I can’t find any way to prevent particles passing through the colliding object, if the colliding object is moving towards the particles while colliding.

I think it is because the raytraced collisions are calculated with one frame delay, so it detects the particle only once it’s already past the colliding surface. I would think that there should be a way to easily solve this with some kind of bias or offset, but two days of research and experimenting did not get me anywhere.

Imagine you drop particles on an elevator floor while the elevator is going upwards: those CPU particles will bounce off and stay on the floor when it’s not moving, but as soon as the floor is moving upwards, they will pass through.

Have you found a solution for this?