Environment Fog tests

[rfellers]

I think I figured out the animation thing. I set both primary and secondary to light cache, upped the subdivisions to 2000, set the scale to world (I left the sample size at .02 - my gizmo is sphere with an 8 unit radius) and the mode to fly-through w/ auto-save on. Once the light cache was calc'ed, I set the mode to 'from file' and loaded the vrlmap. The light cache took about 15 minutes to calculate. Frames took about 6 minutes. I'm running this on a quad xeon 2.83ghz w/4 GB ram running XP, 64-bit, Max 2008.

http://www.ryanfellers.com/pub/Atmos.mov

HenrikBC - Thanks for the post!

[HenrikBC]

rfellers: Looks stable - cool.
But my problem is with animating moving stuff. I believe irradiance+light cache provides stable solutions as well - for flythough animations when precalculated in that mode.

This one vray_volfog_03b_h264 has the light source moving towards the left (cancelled further frames due to impatience). I ran multiple tests yesterday to no avail. Right now I'm trying brute force for primary and none for secondary. 30 mins per frame in 640x267.... so not really usable, but if it gives me a stable solution then at least I'm that "far".

[rfellers]

I had originally tried animating the phase of the texture, but gave up on that and decided to just get the static scene working and then go from there. I had explored the brute force approach, but when everything is animated (lights, texture, camera) I think the step size (in the vrayfog setup) also has to be extremely small to keep the lighting from jumping around. When I get some free time I will experiment more with this.

[Adam H. Stewart]

This images and animations are so damned cool - I love it when 3d gets away from the shiny and into the murky world of the organic like this - can't wait to do some of me own inner space / blood testing - great !

[stevesideas]

Very nice animation!

[peterguthrie]

here is my inner space movie:



http://vimeo.com/4087738

it was rendered using the universal method with LC on flythrough mode and a noise threshold of 0.03, took about 9 seconds per frame at 720x480px on my macpro. It actually has a moving light source, not that you can tell! didnt turn out as i had hoped really, must try harder..

[RErender]

Cool stuff, I'm not sure though that the flashing in the animation is totally incorrect. When flying through a cloud in an airplane the effect is much like that at the window.

[HenrikBC]

Nice Peter. And stable too - unlike mine. Frustrating. But I think I've figured out why.

My clip with the moving light was an omni inside the volume. The omni's source of light - and therefore the shadow's - is infinitely small, which causes the light throughout the whole scene to change drastically whenever there is even a small change in the volume density around the center of the light source.

So.. to circumvent this, I tried using a vray light shaped as a sphere so the light would be emitted from a bigger area instead of an infinitely small point in space. It seems to work, but the rendertimes are quite stupid - not to say useless - for any production environment (that would want to do this effect like I do). And the result is quite noisy because of lacking vray light subdivs (just looks like artifacts in the quicktime due to compression).

vray_volfog_06_post_h264.mov

Rendertimes were a horrendous ~51 mins per frame - on my dual quad mac pro nehalem 2.26 ghz... Some six days of nonstop rendering while I was in Germany.

My colleague Anders Steensgaard suggested I tried using the omni with an inner attenuation to avoid the zeropoint center problem, but it doesn't work - shadows are still cast from the center of the omni even when an object is placed inside the inner att. area. If someone has more suggestions, I'd love to hear them.

[MasterBercon]

Fade the texture around the light source... You can do this with by multiplying your existing map with Falloff or BerconGradient in the distance to object mode.

[vlado]

You can also use an Omni light with inverse square decay and larger Decay Start value (*not* near/far attenuation) to avoid the intensity going off to infinity near the light position.

Best regards,
Vlado

[HenrikBC]

Bercon: I have previously attempted using falloff to avoid sharp edges on the volume to no avail. But I will try what you suggested later today.

vlado: The previous test I did was with inverse square on the omni. The problem is not the intensity of the light at the center - the problem is the shadows flickering when the point from which they are cast is constantly changing in density (the env fog).

[vlado]

Well, I'm doing a small test now and it seems to render fine... will post it a bit later.

Best regards,
Vlado

[vlado]

vlado: The previous test I did was with inverse square on the omni. The problem is not the intensity of the light at the center - the problem is the shadows flickering when the point from which they are cast is constantly changing in density (the env fog).

That would be somewhat normal, as the light will be blocked to varying degrees depending on which way it goes... in that case MasterBercon's suggestion is probably a better idea.

Best regards,
Vlado

[vlado]

Here is what I got:

http://www.spot3d.com/vray/images/stuff/env_fog.avi

Best regards,
Vlado

[peterguthrie]

another silly test from me: http://vimeo.com/4196011

but with a stationary omni - about 4 mins per frame at 720x480 on my mac pro