Jump to content


Photo

Particles Move Position When Motion Blur Tweaked


  • Please log in to reply
9 replies to this topic

#1 John Paul Docherty

John Paul Docherty

    Power Pig

  • Adv Members
  • PipPipPipPip
  • 43 posts

Posted 21 March 2014 - 02:40 AM

I'm working on a v quick turnaround shot that involves replicate3D and particles and am a bit confused about how motion blur works with them - I was hoping one of the particle wizards on the forum might be able to give me a few clues.

I've got some geometry going into a replicate 3D node (some squashed bullet slugs) and a particle set up that shoots them out from a central mesh.

Basically I have a cylinder that emits particles with no velocity for a single frame then applies a one frame point force from the centre to blow them out radially from the surface of the cylinder (roughly at right angles to the polygons making up the cylinder surface). I've done it this because I couldn't find a way to make the particles radiate out in straight lines from the surface using the emitter alone - the angle setting pushes them in one one direction and if I use angle variance at 360 degrees I get particles shooting out from the back (inward facing) polygons as well, so when they eventually pass out through the other side of the cylinder it looks like they were emitted later, which sort of ruins the effect. I'm sure there's a better way to do this but I've been scratching my head and my method seems to work, although it makes the initial velocity of the slugs a bit uniform. Not my problem but if anyone has a better idea . . .

Anyway, the real problem. I've set a seed number (3 in this case) to make sure the particles repeat from render to render. My problem is that when I apply motion blur (with identical settings in the pRender and render nodes) I get some blur, although it looks a bit odd, seems to be a lot less than I would expect even though the shutter is set to 280. However, when I change the shutter angle on both nodes the particles all now start from completely different positions. Preroll is set to automatic, so it's not that. I find that if I even change the sampling from say, 8 to 6, again the particles again start from a different position on the surface of the cylinder. This is a pain and as the shot is 3k resolution very tedious and time consuming to test and deal with.

I wondered if anyone could explain how this works and why even a sampling change can alter the positions - I know Gregory is the particle whiz and sure he and others must have come up against this - and if there is way to get around it. Many thanks in advance.

PS While I'm here, just a small addition to the Fu7 debate and Tilt etcs various excellent points about what is needed in Fu7 as opposed to what is already in Nuke - the fusion tracker itself is ancient, very basic and has quite a few problems. I'm not even talking about planar or 3D tracks, just simple 2D tracking. This is a core comp package capability used on most shots and the current tools failings has caused me problems again and again, some of which has been discussed in the past on this forum. Even blender has a better tracking system, and it's free. I'm just mentioning this in case the eyeon people ever look at the thread . . .

#2 SecondMan

SecondMan

    Flying Pig

  • Adv Members
  • PipPipPipPipPip
  • 1,797 posts

Posted 21 March 2014 - 11:48 AM

Hey Paul,


Been a while :) Hope you're well.

The particle thing is a known issue. And really not a lot you can do about it, apart from setting up particles with motion blur from the start. Extremely annoying especially if you have a carefully tweaked setup.

There was a bit more on that here: http://www.pigsfly.c...?showtopic=7138

You could get in touch with Chad to check if he ever pursued that further.

Agreed on the tracker.

#3 ChadCapeland

ChadCapeland

    Flying Pig

  • Adv Members
  • PipPipPipPipPip
  • 1,975 posts

Posted 21 March 2014 - 12:01 PM

It's not just motion blur that causes this. pMerge does too. Eyeon has this logged.

#4 John Paul Docherty

John Paul Docherty

    Power Pig

  • Adv Members
  • PipPipPipPip
  • 43 posts

Posted 21 March 2014 - 01:47 PM

Feck. Many thanks Chad and Pieter - at least I won't waste any more time trying to get around it, time being what I don't have at the moment (it's a trailer shot and the deadline is Monday).

Bit of a nasty one that, hard to explain to the director that all the positions he carefully signed off on are now different. Or that strobing is a good thing.

As a matter of interest, any thoughts on the radial emission thing? Although I've finagled my way around it I have a sneaking feeling that I'm missing something painfully obvious, as radiating particles from the surface of an object (or at least at right angles to the surface of the poly doing the emanating) would seem to be a pretty common thing to do.

Thanks again both of you.


PS Good to hear from you Pieter, we should catch up soon. Hope the family is well!

#5 Tilt

Tilt

    Flying Pig

  • Adv Members
  • PipPipPipPipPip
  • 2,137 posts

Posted 21 March 2014 - 02:09 PM

motion vectors from Renderer3D. Soon in a Fusion version near you. Maybe you can render passes of non-overlapping bullets seperately and use dimension to vectorblur them.

#6 SecondMan

SecondMan

    Flying Pig

  • Adv Members
  • PipPipPipPipPip
  • 1,797 posts

Posted 21 March 2014 - 02:41 PM

I would have to test this so I might be wrong, but perhaps setting an angle variance of no more than 180 degrees would do what it is you're looking for?

360 would mean a full sphere around the angle of origin, so those would indeed shoot out the back...

#7 ChadCapeland

ChadCapeland

    Flying Pig

  • Adv Members
  • PipPipPipPipPip
  • 1,975 posts

Posted 21 March 2014 - 04:10 PM

Is it ok if you cache the particles out first?

What about using a Time3D?



#8 doctoritcata

doctoritcata

    Advanced Pig

  • Adv Members
  • PipPipPip
  • 30 posts

Posted 22 March 2014 - 03:24 PM

Will Reel Smart Motion Blur do the job maybe?

#9 vladi013

vladi013

    Advanced Pig

  • Adv Members
  • PipPipPip
  • 35 posts

Posted 24 March 2014 - 07:16 AM

I was gonna say too just go with the RSMB.

#10 John Paul Docherty

John Paul Docherty

    Power Pig

  • Adv Members
  • PipPipPipPip
  • 43 posts

Posted 27 March 2014 - 11:15 AM

Thanks for the thoughts - kludged my around it and trailer delivered.
Pieter - tried that, still get particles flying from one side through the other. Chad - haven't tried caching them, will do when I get the chance.
thanks again




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users