Cineversity

Hello,

I’m working on an atmosphere plate, creating dust particles that will eventually be incorporated into live action shots, and am currently figuring out the best way to generate the dust particles.

For simplicity’s sake, I’m using a MoGraph Cloner, with two separate Random Effectors (one for position and one for rotation), on a collection of five or six unique splines that are cloned, depending on the needs of the scene, sometimes going as high as a 10 x 10 x 10 grid of objects.

Now, so far, I’ve tried skinning this cat two different ways:

1. using a hair texture, applied directly to each spline (haven’t tried simply applying it to the cloner, will have to test that).

advantages: fast viewport.
disadvantages: slow start up on renders (calculating hair).

2. using a sweep object to “dress” a square along the spline while trying to minimize the number of polygons generated.

advantages: starts to render right away.
disadvantages: my viewport becomes too slow (I’m running on a machine dating back from 2009 with only 512MBs of VRAM, so…).

I created an xPresso to switch the sweep objects on and off.  This helps, but after working on the scene for a few minutes, the system starts to slow down too much and I see the spinning wheel again.

My questions are:

A. Would applying the hair texture to the cloner instead of each spline help reduce the “calculating hair” time?
B. Most importantly, is there a better way to do this?

To help illustrate my poor explanation(s), I’ve included the scene file.

Thank you to anyone who can help / offer suggestions.

All the best,

Jerome

PS
I own and have used with great results Trapcode Particular, but am looking at different ways to get the job done.

Hi Jerome,

A. You can test that certainly as fast as I can do it. I think to use “Render Instances” would be more of interest ;o) I have a 6K/4K de-noise rendering in the background and my results are not stable, but similar for all four cases.

B. Render Instances is certainly better way here. But why do you need a volume, you could have them all use the camera as target (Target Effector), and then use only a small polygon object, perhaps three polygons.

If there is nothing else in the scene, e.g., interacting and moving the air, then the set up should work. But I have more the impression that dust follows very easily any little change of air, sometimes even hotter objects (black iron) might change their motion.

As an alternative, you could pre-render some layers and use them as alpha information on planes perpendicular to the camera. Which means the camera move should be evaluated before.

To get a scene specific “dust light” I think everyone likes to adapt to the scene and its requirements. Which means testing…

I would be perhaps have an XPresso that switches the amount of particles in the air for the editing time down, and with a checkbox then higher.

This is similar to the way of Editor Menu>Options>Level of Detail, which normally works nicely with generated objects, e.g., Sweep.

All the best

Sassi

Sorry of the late response.  Was busy making my MoGraph playlist.  ;^)

Dr. Sassi - 03 September 2014 01:31 AM

Hi Jerome,

A. You can test that certainly as fast as I can do it. I think to use “Render Instances” would be more of interest ;o) I have a 6K/4K de-noise rendering in the background and my results are not stable, but similar for all four cases.

Render instances doesn’t seem to work with me.  I don’t think render instances supports splines with hair, though I could be completely wrong on that.  Replacing the splines with a small cube reveals that there is no problem on my system with render instances, so it’s definitely a spline / hair issue.

Also, applying the hair material to the cloner doesn’t work.  It’s not like applying a color to cloned polygons.  C4D seems to behave differently with the hair material.

B. Render Instances is certainly better way here. But why do you need a volume, you could have them all use the camera as target (Target Effector), and then use only a small polygon object, perhaps three polygons.

If there is nothing else in the scene, e.g., interacting and moving the air, then the set up should work. But I have more the impression that dust follows very easily any little change of air, sometimes even hotter objects (black iron) might change their motion.

I’ll definitely have to look into that, though the reason for using a volume is to accentuate perspective and parallax.  Since most of the shots are handheld, I was thinking of trying my hand at motion tracking.

As an alternative, you could pre-render some layers and use them as alpha information on planes perpendicular to the camera. Which means the camera move should be evaluated before.

You mean create plates in C4D, then composite them in AE?  Not sure I follow.

Jerome

Hi Jerome,

Instances was meant for the Sweep solution. I did not had Hair at all in mind, sorry for the confusion, I see you asked for the Hair.

Hand held shots are normally done with random motion blur, no constant motion is expected. Which means that the dust would be affected as well. I’m not certain how much parallax you will get from a single dust particle, especially given your set up with a shallow depth of field, and your idea of a subjective camera. The dust, at least how I understand it, marks the empty space and allows light to be seen where you have normally none. It is a little bit less of an effect to show time, as it can indicated slow motion or time-lapse, besides showing that a space is in silent (no wind, etc). As an cinematographer I would have my attention only shortly on it, just to establish a mood, but longer that a second a main theme must have a very specific reason, if one likes to tell a story. If you motion track, use a camera that has no rolling shutter, DSLRs are mostly equipped with that problem, a reason why handheld-dslr-motion tracking would ring all alarm bells for me. If you like to get only the dust effect in it, that might work, as anything that is not connected to a practical surface is much simpler to integrate of course.

I don’t think about Ae here, it is an technique that is used in digital matte-painting, not for dust, but for haze and fog. I have attached a sketch, with way too much movement, to make the point, as well with way too much “dots” on the texture. The key, however, is here that the dust interacts with the light in the scene. If that is not a given, you might take the reality out of your shot (perhaps a wanted effect).

All the best

Sassi

Thank you for the insight.  I’ll play around with it and get back to you if I have any questions.

As for the camera, no worries.  It was shot on Alexa.  ;^)

You’re welcome, Jerome.

Of course less dots and more planes with different textures.

Not to say anything bad about Alexa, or any other CMOS camera (I have several…), but for the sake of Camera Tracking:
There is rolling shutter in the Alexa
https://www.youtube.com/watch?v=hnqANUgzzGw
Stop the clip when it “pans” fast. Check the vertical lines. Without the Motion Mount, the RED shows that as well!

All the best

Sassi