Renderman Compliant?

[HyFrmn]

I've been working on getting Houdini to render using Render Drive from ART-VPS (www.artvps.com) I have had a little bit of success so far, but light intensity across the renders do not seem to match. I'm using a very basic lighting setup with three spotlights. In order to get some lighting to appear from my RenderDrive renders I had to crank the intensity up on my spotlights to 3. I'm using the spotlight shader for the renderman shader on each spotlight, and checked use expressions, so when I output the RIB file for rendering my lightshader has a color value of [3 3 3] (I double checked the RIB to be sure this what was being output). The settings work fine and look similar for both RenderDrive and Aqsis, but are obviously too hot for Mantra. Can any one tell me why or think of a reason I would to be blowing out my lights like this for a Renderman Complaint Renderer?

Aqsis render:

Render Drive render:

Mantra render:

~MP

[old school]

Just a guess here but try switching the lights for the Mantra render to physically correct (inverse square falloff).

[Jordan]

Just a guess here but try switching the lights for the Mantra render to physically correct (inverse square falloff).

Now i know physically correct = inverse square falloff in other packages

Is that half density distance = quadratic falloff in other packages?

What about cubic falloff ?

It will be great, if u can explain that. thanks in advance.

[lukasdesign]

Now i know physically correct = inverse square falloff in other packages

Is that half density distance = quadratic falloff in other packages?

What about cubic falloff ?

It will be great, if u can explain that. thanks in advance.

CUBIC = Quadratic, so it should be the same I guess

[Mario Marengo]

CUBIC = Quadratic, so it should be the same I guess

As far as I know, "cubic" != "quadratic", at least in the typical use of those terms. They usually describe the degree of an expression or polynomial (the highest exponent found in the expression). In that context, a "quadratic" expression (polynomial) is one whose highest exponent is 2, a "cubic" would have a term with an exponent of 3, "quartic" = 4, "quintic" = 5, etc.

In the case of light "falloff", physicists tell us that the rate at which light intensity drops is inversely proportional to the square of the distance traveled (from the light source) -- so, this would be the numerical inverse of a quadratic (degree 2) expression, like 1 / d^2.

The problem with the literal use of 1 / d^2 as an intensity multiplier is that it approaches infinity at the light source (d=0), so instead, you'll more likely see 1 / ( d^2 + 1 ) used in an actual shader (this results in an intensity value of 1 at the source, while still retaining the "physically correct" falloff rate).

So, in a light shader, "quadratic falloff" can be usually taken to mean something like 1 / ( d^2 + 1 ), "cubic" would be 1 / ( d^3 + 1 ), "quartic" = 1 / ( d^4 + 1 ), and so on.

Ultimately though, you're better off looking inside the shader and seeing what they used

HTH.

[Jason]

In the case of light "falloff", physicists tell us that the rate at which light intensity drops is inversely proportional to the square of the distance traveled (from the light source)

The most intuitive explanation of this phenomenon is this: if you imagine yourself looking up at large radiant surface- the sun, for instance, you'll be seeing it taking up a portion of the sky. You are exposed to light emitted from all over the surface of the sun. Now as you move the sun away from you, the area of the sky the sun covers shrinks in both it's horizontal and vertical dimensions, and so shrinks at an inverse square rate. Thus you are exposed to a smaller area of the sun, and the sun is emitting the same amount of energy. So note that the light source is not become dimmer or the light energy is not suffering attrition by the atmosphere - you are merely being exposed to less of it. So this formula that Mario posted is needed because CG lights are infinitely small (they have no virtual size at all) and so you need to emulate the diminishing exposure somehow.

Moreoever, there could be a small amount of scatter by the atmosphere but that depends on the how clear the air is; and this might be what cubic is attempting to model. I read somewhere that a laser beam gets cut down by 20% over a kilometer on a clear day and by 80% on a hazy day; so it's quite variable.

Hope this "illuminates" the issue a little more:)

Jason.

PS. The above is proved in GI simulations when you make a surface emissive and merely animate the emissive surface away from target. There is no hardcorded attentuation of indirect light in GI simulations - and it just looks right, because it is.