Converting Particle Coordinates to Matrix?

Hi All,

I am trying to convert the LOC/ROT/SCALE information contained in the .co property of each particle into a matrix_world compatible matrix. The result of such conversion would result in an object being placed at the same position and orientation as the original particle.

I have some code to fetch the .co properties. But I am stuck on how to convert to a matrix?

import bpy

def valid_particle(pa, cfra):
    return not (pa.birth_time > cfra or (pa.birth_time + pa.die_time) < cfra)

def returnSampledParticles(scene, psys):
    loc = []
    rot = []
    size = []
    birth = []
    death = []
    for p in psys.particles:
        if valid_particle(p, scene.frame_current):
            loc.extend( p.location )
            rot.extend( p.rotation )
    return (loc, rot, size, birth, death)

ob_emitter ="emitter")
scene =[0]
psys = ob_emitter.particle_systems[0]
samples = [returnSampledParticles(scene, psys)]        # This def takes longer the larger the particle count.
# Loop through the particles.
for loc, rot, size, birth, death in samples:
    # Here is where I am stuck.
    # Are particle rotations quaternion?
    matrix_particle = ob_emitter.matrix_world * returnMatrix(loc, rot, size)

# What would returnMatrix code look like?

I think this is close…

def returnParticleMatrix(particle):
    locmat = Matrix.Translation(particle.location)
    #particle.rotation    #Quaternion.
    rotmat = particle.rotation.to_matrix().to_4x4()

    # Particles are uniformly scaled.
    scalemat = Matrix().Scale(particle.size, 4, Vector((1,0,0)))
    scalemat *= Matrix().Scale(particle.size, 4, Vector((0,1,0)))
    scalemat *= Matrix().Scale(particle.size, 4, Vector((0,0,1)))
    mat = locmat * rotmat * scalemat

    return mat

But the end result seems to be an order of operation failure. Like the particles are rotated then translated instead of translation then rotation…?

The viewport before render:

The result of the render:

If you want them to translate then rotate then you just use:

mat = rotmat * locmat * scalemat

When people build these matrices they usually think of them like:

  • I want to translate the matrix:
    mat = locmat

  • Now I want to rotate it:
    mat = locmat * rotmat

  • Then I want to scale it:
    mat = locmat * rotmat * scalemat

But the the operation seems ‘reversed’ because the final result is evaluated like this:
finalResult = (locmat * (rotmat * (scalemat * point)))

First the point is scaled, then rotated, then translated.


I am still having trouble with this. If I use your suggested code I get no particles rendered within my camera view. They do appear in my export file but the numbers must place them behind the camera or out of view.

I basically have location and scale working using.

mat = particle_point_mat * group_object.matrix_world * scalemat

This gets me my group participants in the correct location of a particle. This image shows two particles with a whole group deployment.

When I add in the rotation matrix all particle disappear. I am calculating the rotation matrix from the particle system quaternions like so…

rotmat = particle.rotation.to_matrix().to_4x4()

Is that correct?
Or do I need some more math because the rotation is a quaternion value?
What about degrees or radians…does that matter at the matrix level?

Basically I just need what I have rotated at the particle_point_mat level.
And as I type the above it finally hits me.

mat = (<i>particle_point_mat</i>*rotmat* scalemat) * group_object.matrix_world 


Removed the comment, I should have noticed the ‘solved’ tag!