AWS Thinkbox Discussion Forums

Motion blur with attach to node

Ok, I’m loading particles from disk, and attaching them to a node. In this case a point. The point is keyframed to move around the scene.



The motion blur passes on don’t show this, though, essentially representing the particle system as static.

Tried something different…



I am rendering scene particles with pflow.



Position Icon operator with Lock on Emitter, the PF Source is path constrained to a star shaped path, and travels along that path once every frame.



Integration step 1 tick (!), motion blur segments set to 25, shutter 360



Basically, I should see a star shaped smear.



But instead I see a linear smear. Odd?


Ok, tried it with 3 gravity forces, each set to a waveform controller. Each gravity is pointing in a different direction, so I end up with a pleasant periodic motion. In the viewport I can see the particles moving in little circles.



Render is completely linear.



So I’m thinking that the segment number isn’t actually updating the particles.

Render is completely linear. <<



I’m looking at the same thing right now. I have a couple of cases where

this is very clear, but I need to talk to the engineer to see if this is a

bug, or by design.

Parenting the camera and target to a dummy that is on a path constraint on a spiral that runs through the spiral once per frame DOES make spiral particles.



A very hackish way of making very large particles.



I have a scene with 10 million particles, but with them so fuzzy with the spiral motion blur, you’d think there was a lot more. Of course, nothing it sharp or distinct, like you would get with more particles.

Render is completely linear. <<



I was able to get some clarification on this. Motion blur is 3d, but

linear. Particle Flow gives us only position and velocity per frame, so

there is no practical way to extract the sub-frame data needed for curved

blur. In our experience, linear blur has given us sufficient results, but

if you have a real-world case that fails we would be happy to consider

alternatives.

That’s odd. Even though Pflow can cache 2400 ticks per second?



Hmmm… If I could change the frame rate, to say 180fps or 360fps, this would be less of an issue. We’ve done that many time before, but with particles, it’s a bit trickier.



Can you do frame offset rendering? Like Brazil 2 has? Where you can render not 1, 2, 3… but 1.2, 2.2, 3.3, etc? Then we could set a smaller shutter angle and just keep rendering offset time values (via deadline) until the motion was smooth enough?



Hmmm… The application I was using this for, the “real world” example, was having low density (spacially) particles move in vary fast ocilations, while the high density stayed put. That way the low density particles would come both larger and less dense at the same time. Space filling and such, and less reliance on density mapping at rendertime.

Oh, what about the initial problem? The particles don’t have motion blur when attached to a moving scene node?

>>Oh, what about the initial problem? The particles don't have motion blur when attached to a moving scene node? <<

I'm definitely not seeing this in my tests.  My motion blur is definitely following the motion of the node.  Do you have a .max file we can use to reproduce this?

Privacy | Site terms | Cookie preferences