Motion Blur Dilemma

Krakatoa for 3ds Max
1

Hi, I have an issue here. At first I thought It was my particula particle flow setup, and then I realized, it affects everything.

The motion blur in Krakatoa does not take the trajectory into consideration, but only the velocity, which is only a vector value.

This means, only straight lines for particles? I am surprised I only realized this myself now, maybe because I was using so many particles before that I didnt notice. It becomes especially apparent in many examples though.

Ky_Trails is one plugin which needs velocity to work, but it actually traces the path of each object or particle, WITHOUT having to simulate the last 50 frames with pflow.

Can we expect a motion blur in Krakato v2.0 which follows the trajectory and not just a pure vector value?

2

You can render non-straight motion blur by using multi-pass motion blur, which Krakatoa is fine with.

If you wanted to store the velocity as something other than a vector, what would you store it as?

  • Chad
3

yes velocity of course is only a straight vector…

can you please tell me how i can render multipass motion blur with krakatoa? can u show an example?
of particles with smooth “real trailed” motion blur???

:open_mouth:

4

i tried adding motion blur effectbut had no effect

5

ok, but u meant of course the in-camera motion blur effect!!

wow chad, thanks alot :slight_smile:

Advantages
The 3ds Max Multi-Pass Motion Blur does sub-frame sampling of the scene, so it takes into account the actual trajectory of particles between frames.

HUGE ADVANTAGE!!!

would be cool if krakatoa 2 will have its own style of drawing motion blur with trajectory, maybe so that if u render 1000frame animation, it does not have to render every frame 50x but take the information rendered for each already

6

What do you mean with the last sentence? What could you cache between frames?

7

well if the camera isnt moving for example, the frames it has to redraw alreay have been rendered, so lets say a 10frame duration motion blur on frame 20, then it already has “calculated” the last 10frames, instead of doing recalculation just take the information that has been done…

make sense?

8

But they’ll be out of depth order… would only work if the particles were always moving toward the camera.

9

oh ok,
what about a faded opacity?

10

Still wouldn’t help. The idea of rendering without depth sorting would only work for additive or other associative modes.

Besides, it’s effectively a solved problem, you can just do the accumulation of frames in a composite.

  • Chad
11

too bad the pixol technology from zbrush isnt used, just an extra channel with depth information!

12

But you’d need to store that for every point, not every pixel. The rendered files would become enormous. It would be the equivalent of RPF’s, but with massive amounts of layers.

  • Chad
13

but the points already have position information inthem,

but jea,

i would just like for kraka to have its own trajectory motion blur, would be better i think :slight_smile:

14

But the position data is stored in the PRT. Each frame that you render could be as large as ALL the PRT used to render it AND the previous frames. :open_mouth:

15

but the points already have position information inthem, = But the position data is stored in the PRT

i just meant, forget about that technique i mentioned, i would just like for kraka to have its own version of the incamera multipass motion blur, so it would natively support trajectory motion blur

16

But it already supports multipass motion blur. It’s actually a more accurate way of doing motion blur, as it accounts for more things. The position extrapolation technique is there mainly because it’s very fast and light on I/O.

  • Chad
17

yea your right :slight_smile:

im glad it does! next time ill chek the documents before starting a thread :slight_smile:

18

You can also combine Max’s multi-pass motion blur with Krakatoa’s motion blur to get better motion sampling. For example, you might try 4 multi-pass passes, and reduce the Krakatoa motion blur interval by 1/4. The motion will be piece-wise linear with 4 sub-frame samples.