Object collision for particle nodes #77266

New Issue

Sebastian Parborg · 2020-06-02T13:53:18+02:00

Sebastian Parborg commented

2020-06-02 13:53:18 +02:00

Here is the current status of the node particle collisions:

It currently handles:

Static objects	Moving objects	Objects with mesh deformations
recording_static.mp4	recording_moving_plane.mp4	recording_moving_wave.mp4

What is left to do is:

- Add a recovery method for when particles are pushed or somehow gets inside colliders (when particles are less than their radius away from a collider). This will also allow particles to have a dynamic radius. (Haven't tried with dynamic radius yet, but it should work in theory)
- Change how face normals are calculated. Now it simply uses the face normal. But this is an issue when the particles collides on the edges or verts of the triangles. This leads discontinuities in the collison mesh as the transition area between the faces is not smooth.
- Make colliders be able to push particles along even if the dampening or friction is so high that the particle will not have any velocity.
- We will probably have to track acceleration in addition to velocity to get high quality simulation (like in the old particle system). Now I think some weirdness can be attributed to us lacking acceleration tracking.
- Look over and port older properties from the old particle system (friction, stickyness, and so on). Flocking (Boids) and other behaviors will be handled in other nodes (not inside this collision node).
- Hook up variables to the particle nodes themselves. Currently most things are hard coded.

Here is the current status of the node particle collisions: It currently handles: | Static objects| Moving objects | Objects with mesh deformations | | -- | -- | -- | | [recording_static.mp4](https://archive.blender.org/developer/F8571960/recording_static.mp4) |[recording_moving_plane.mp4](https://archive.blender.org/developer/F8571961/recording_moving_plane.mp4)|[recording_moving_wave.mp4](https://archive.blender.org/developer/F8571963/recording_moving_wave.mp4)| What is left to do is: - - [x] Add a recovery method for when particles are pushed or somehow gets inside colliders (when particles are less than their radius away from a collider). This will also allow particles to have a dynamic radius. (Haven't tried with dynamic radius yet, but it should work in theory) - - [x] Change how face normals are calculated. Now it simply uses the face normal. But this is an issue when the particles collides on the edges or verts of the triangles. This leads discontinuities in the collison mesh as the transition area between the faces is not smooth. - - [x] Make colliders be able to push particles along even if the dampening or friction is so high that the particle will not have any velocity. - - [ ] We will probably have to track acceleration in addition to velocity to get high quality simulation (like in the old particle system). Now I think some weirdness can be attributed to us lacking acceleration tracking. - - [ ] Look over and port older properties from the old particle system (friction, stickyness, and so on). Flocking (Boids) and other behaviors will be handled in other nodes (not inside this collision node). - - [ ] Hook up variables to the particle nodes themselves. Currently most things are hard coded.

Sebastian Parborg self-assigned this 2020-06-02 13:53:18 +02:00

Sebastian Parborg commented

2020-06-02 13:53:18 +02:00

Changed status from 'Needs Triage' to: 'Confirmed'

Sebastian Parborg commented

2020-06-02 13:53:18 +02:00

Added subscriber: @ZedDB

Jacques Lucke commented

2020-06-02 14:11:05 +02:00

Added subscriber: @JacquesLucke

Christian Beckers commented

2020-06-03 00:29:08 +02:00

Added subscriber: @BeckersC

Sebastian Parborg commented

2020-06-03 14:37:00 +02:00

I noticed that I had some tunneling issues still. Especially with the moving wave mesh.

I'm currently looking into it. But it is one of those problem where you discover even more issues.
For example the particles seems to "pop out" of the plane sometimes even though the max distance it can forcefully be moved is just the particle radius to the plane.
So this is going to take a while too I guess :/
recording_prog.mp4

I noticed that I had some tunneling issues still. Especially with the moving wave mesh. I'm currently looking into it. But it is one of those problem where you discover even more issues. For example the particles seems to "pop out" of the plane sometimes even though the max distance it can forcefully be moved is just the particle radius to the plane. So this is going to take a while too I guess :/ [recording_prog.mp4](https://archive.blender.org/developer/F8574138/recording_prog.mp4)

Sebastian Parborg commented

2020-06-04 19:13:19 +02:00

I think I managed to solve the tunneling and popping issues.
The popping was because of an older coding mistake in the collision_newton_rhapson function that I copied from the old particle system.

recording_prog.mp4

I think I managed to solve the tunneling and popping issues. The popping was because of an older coding mistake in the `collision_newton_rhapson` function that I copied from the old particle system. [recording_prog.mp4](https://archive.blender.org/developer/F8577148/recording_prog.mp4)

Sebastian Parborg commented

2020-06-10 03:49:23 +02:00

I think I managed to get something that handles forced movement of particles nicely.
In this example the dampening is set to 1, so the particles will have zero velocity as soon as they collide with anything:

recording_full_damp.mp4

I think I managed to get something that handles forced movement of particles nicely. In this example the dampening is set to 1, so the particles will have zero velocity as soon as they collide with anything: [recording_full_damp.mp4](https://archive.blender.org/developer/F8604892/recording_full_damp.mp4)

Jacob Merrill commented

2020-07-29 01:44:43 +02:00

Added subscriber: @JacobMerrill-1

Jacob Merrill commented

2020-07-29 01:44:44 +02:00

is there a way to loop the thing to run continuous?

like take the end of the sim and use it as the start frame next cycle?

we would like to use this in upbge but are struggling to understand how we can use it in realtime.

is there a way to loop the thing to run continuous? like take the end of the sim and use it as the start frame next cycle? we would like to use this in upbge but are struggling to understand how we can use it in realtime.

Alain Caviggia commented

2020-10-01 16:18:13 +02:00

Added subscriber: @Impact

Alain Caviggia commented

2020-10-01 16:18:13 +02:00

Hello, someone can explain me why the Particle Nodes doesn't appear in my Blender 2.91 for Mac/Apple please ?
Also, in my Experimental bouton preferences, I don't have the option : New particle System #73324.
Should I will use Window version ?
Thank you a lot :-)
Alain

Hello, someone can explain me why the Particle Nodes doesn't appear in my Blender 2.91 for Mac/Apple please ? Also, in my Experimental bouton preferences, I don't have the option : New particle System #73324. Should I will use Window version ? Thank you a lot :-) Alain

Philipp Oeser removed the

Download

What's New

Blender Studio

Manual

Developers Blog

Documentation

Benchmark

Blender Conference

Development Fund

One-time Donations

Object collision for particle nodes #77266