Vray Directional lights

[rajdarge]

they are good.
1. They are slow: rendering times are increased significantly in proportional the the directional value.
2. They are noisy: again directly proportional to the amount of directionality.
3. They affect light intensity: also in direct proportion to amount of directionality. So much so that a small dim highly directional light from a far distance can overpower a bright large light closer to the subject.
4. They are restricted to a 1.0 for directionality: you cannot simulate convergent beams of light like you get with a spotlight with a lens focusing adjustment. You can simulate this with a modelled lens and caustic effects but at a big hit to quality and render time.
5. They look fantastic when used.

I assume you modelled the directional factor based up a 100% efficient Parabolic reflector behind the light?
would it be possible to have a check box to make the light intensity at a point in front of the light the same as if it was not directional. something like: "IsoLuminescent" ?

Yes I do use Parabolic reflectors to achieve direct (or specular) lights - but I might also use a gobo or snoot or a grid to achieve the same thing. Most of these just absorb the light outside of the target beam, so do not increase but decrease the intensity. I tend not to use the parabolic reflectors as they produce and efficient but harsh light.
Raj

[vlado]

1. They are slow: rendering times are increased significantly in proportional the the directional value.

That might be because of #2

2. They are noisy: again directly proportional to the amount of directionality.

There might be some increase in the noise, but not that much, and it is certainly not proportional to the directionality. A very directional light is less noisy compared to a light with medium directionality. Still, if you can get me some samples to study, we may be able to optimize this.

3. They affect light intensity...

The total emitted light power stays the same. Obviously as the light gets more directional, areas right in front of the light become brigher to compensate for other areas becoming darker.

4. They are restricted to a 1.0 for directionality: you cannot simulate convergent beams of light like you get with a spotlight with a lens focusing adjustment.

It is true, but this is a different kind of directionality that would be a lot more complex to calculate in the same way.

I assume you modelled the directional factor based up a 100% efficient Parabolic reflector behind the light?

The directionality parameter controls the way each point on the source surface emits light relative to the surface normal - something like a diffuser in front of a perfectly directional light.

would it be possible to have a check box to make the light intensity at a point in front of the light the same as if it was not directional. something like: "IsoLuminescent" ?

Yep, will mark it as a feature request.

Thanks for the feedback!

Best regards,
Vlado

[rajdarge]

I have uploaded some images and the maxfile here:
http://www.comcen.com.au/~raj/NoiseAndLightTesting.rar

I did this precise test on version 2.001 and the results were different.
the noise may be made more apparent as the directionality Increases as the intensity increases.
Some of what i called noise from light was actually GI noise.
you can see however that the render times are increased - though not as significantly as I imagined.

[nils_p]

I'd like to take this opportunity to note that when you increase the directional amount, the light fades. At some point around 0.6 the light isn't visible anymore.

is this normal behaviour?

[vlado]

I'd like to take this opportunity to note that when you increase the directional amount, the light fades. At some point around 0.6 the light isn't visible anymore.

is this normal behaviour?

Yes; as the directionality increases, the surface emits light forward, and not to the sides, so it may happen that no light reaches the camera, depending on the angle that you look at the surface.

Best regards,
Vlado

[nils_p]

thx for the explanation

[nils_p]

oh by the way. The online help isn't updated yet with this feature. At least, I didn't see it.

[Jiri.Matys]

+1 for that "Isoluminescent"

[rajdarge]

in the meantime is there a formula I can use to correct for illuminenscence at a point rather than having the reverese engineer one?

Raj

[vlado]

There probably is; not sure what it is yet I have to do some maths...

Best regards,
Vlado

[rajdarge]

I was doing some armchair logic trying to work out volumes of sectors and assuming the flux of light in the sector was the same yada yada yada... after about an hour of dickering
I thought - just divide the light intensity by (1 - Directionality) as this value is largely linear now.
Is this right or did I leave something out?
Raj