Spread should be in degree (see documentation, and flag_disable_z case)

[matrem]

I think particle spread should be taken as degree.
From these points it is :

• the documentation : http://docs.godotengine.org/en/latest/classes/class_particlesmaterial.html?highlight=particles
• the FLAG_DISABLE_Z case :

  godot/scene/3d/particles.cpp

  Line 643 in 8b41ebf

  code += " float angle1 = (rand_from_seed(alt_seed)*2.0-1.0)*spread/180.0*3.1416;\n";

• the default value (45)

So I think this line should be fixed :

godot/scene/3d/particles.cpp

Line 649 in 8b41ebf

code += " float angle1 = rand_from_seed(alt_seed)*spread*3.1416;\n";

[reduz]

can you check this again? I did a few changes to particles recently. Also i may be converting many values to radians wrong from what i see in the code, so checking all usages of 3.14 may be good

[matrem]

Ok I can look into this, perhaps I'll ask you some things about the code to understand it :p
Should I fix this also :
#13488
usage of mix in particles.cpp seems a little buggy (random from 1 is a little strange)?

[matrem]

• factorized a bit pi usage, and random function between -1,1.
• fixed spread
• fixed flatness parameter, it was not implemented (I hope I understand it correctly)
• it seems pi usage is ok, but as it's mentioned in 3D particles cannot rotate #10182 base angle / angular velocity and other things are not well implemented in 3d particles.

Also I think these :

• Particle emission improvement proposal #14697
• More control on Particles randomness #13488
  could be interesting to have a smoother experience of particle material

[reduz]

Seems good

Regarding the suggestions, I think randomness works more or less OK I think for most cases, any alternative I can think of makes it more complicated for the common use cases. Lifetime cant be randomized due to how the shader works, or more likely I suppose It can (can be made shorter I guess) but it wont affect the amount of particles computed simultaneously.

[matrem]

Ok personally I can't understand the current randomness behaviour (for example the scale) :

• You set the scale to 5
• Then you set the randomness to full (1)
  The scale will be chosen between 1 and 5. Where does the 1 comes from? I never say that my random value should get in the [1, 5] range.
  -If you set to middle (0.5), the scale will be chosen in [3,5], ok... why 3?

It seems to mee, that it is really much more understandable and practical to define randomness with a mean value and a variation (absolute or relative) around this mean value.
For example 5 and 2 get me the [3, 7] range.

But perhaps there is something I didn't catch.