VRay Rectangular Lights- A Few Queries

[Jackson_cg]

I could probably work some of these out with more testing, but I've just got too much on at the moment to do lots of test renders, so I thought I'd finally gather my thoughts and ask about some of the VRay Rectangular Light settings and see if we can get something together for the Tips, Tricks and Tutorials forum.

1) "Double-Sided" is self-explanatory, but what about "Ignore Light Normals"? Leaving this unchecked should logically mean light normals are taken into account, so presumably light only emits from the "front" normal, but if you check it does it not mean light will be emitted from both sides- i.e. the same thing as "Double-Sided"?

2) "Invisible"- self explanatory, though a hover tip to remind people that it will still show up in reflections unless they untick "Affect Reflections" would be nice.

3) "No Decay"- why is this a toggle parameter, rather than Linear, Inverse, Inverse Square like Omni Lights? Hasn't been a issue in any of my renders yet, but I'm just curious.

4) "Skylight Portal"- does this completely override emitting colour, multipliers, etc.... and can rain not get in this way?

5) Store with Irradiance Map - This is the one that really stumps me: why would you not want a rectangular light to be stored with IMs? Shouldn't this at least be ticked by default? If this is a useful function, why don't Omni Lights have it?

That'll do just now, maybe someone who has the VRay Bible can illuminate me (sorry, couldn't resist :P).

Jackson

[Micha_cg]

Re: VRay Rectangular Lights- A Few Queries

So far I understand it ...

1) ... it influence the light distributation. If you turn "ignore.." on, than more light will be send to the sides. Physical correct is to turn it off, but for example if you place a rect light as a window light, than "ignore.." is needed.

3) I missed "inverse" some times, for example rendering a small airplane lavatory with a rect light at the ceiling. Inverse square (decay on) was to much decay and linear/"no decay" was not enough decay for looking good.

4) It helps to sample bright areas without "real" lights, for example it can be placed in front off emitters. This area will be "raytraced" like a raytraced light - less errors, but slower.

5) "store..." switch the light from a standard raytracing calculation to a random GI sampling. This cause less accurate shadows/lighting, less noise and more speed.

6) Omni lights don't have a physical size or a quite small size that is not easy to sample by a random GI sampling of the IM calculation sampling. A omni light without radius is infinte small and can't be random sampled

random GI vs. raytracing - an example: you can place hundred of emitters at a scene without a big speed lost, but never place hundred of raytraced lights. Emitters are samples by a random GI calculation, for example 50 IM subdivs means 50x50=2500 sample rays at each shading point. If you place 10 lights with shadow subdiv 20, than 20x20x10=4000 sample rays are needed for each shading point. More lights, more samples must be calculated. The GI sampling stay the same, independent from the emitter count.
I'm not sure I explain it correct, but it helps me to work with Vray.

[Jackson_cg]

Re: VRay Rectangular Lights- A Few Queries

Thanks Micha!

I'll have to bookmark this explanation and re-read a few times to get my head around it, but the "Ignore Light Normals" was particularly helpful- I hadn't realised it had any effect on the light emitted from edges/sides or that it should be checked when used as window lights.

Regards,
Jackson

[dalomar_cg]

Re: VRay Rectangular Lights- A Few Queries

Micha took care of most of it, but here are a few additional points...

3) I miss inverse too....I'll see what options there are for this.

5) with Store with Irradiance Map the lights are taken out of the per-pixel calculations and switched over to a per GI sample point calculation...they are NOT randomly sampled, just sampled via raytracing at a different level. The accuracy of the result is really based on your IR solution. So if you are doing a low quality IR calc, your lights will be lower quality. If you're doing a high quality IR calc, then the lights will be of higher quality as well. This will almost always fall short of a normal calculation, but it does get very close and speed things up significantly.

[Micha_cg]

Re: VRay Rectangular Lights- A Few Queries

So far I understand the vray engine, than if lights are used, than Vray know where are the lights and how much samples must be taken per light - thats the reason why many lights slow down the calculation.
In GI mode random samples are shot in the scene, the engine dosn't know where are "lights" to sample. And at this point a light portal can be used to focus a determinated sampling to an area. I thought, the GI sampling is random controlled and that cause, that small, bright light source are difficult to sample. That's the reason to be carefull with HDRI with bright suns or little bright emitters. Both are difficult to catch by the samples and cause a noise. For HDRI it helps to blur the HDRI so that the sun intensity is washed out over a bigger area.

[dalomar_cg]

Re: VRay Rectangular Lights- A Few Queries

Your understanding is correct. However, this does NOT apply to using Store With Irradiance Map. Without Store enabled, for each pixel sample V-Ray goes through and samples all the lights. With Store enabled, that process happens per IR sample point, and no longer happens on a per pixel basis.

As far as I know, Light Portal mode is for allow an environment (the color and intensity that is behind the light) to be calculated in a similar manner to direct light. Under normal conditions that light would have to be sampled through "random" GI processes, but with a light in portal mode, V-Ray still has to calculate the direct light, but it does so with the "properties" of the skylight behind it.

[Micha_cg]

Re: VRay Rectangular Lights- A Few Queries

Damien, do you remember on a post by Vlado, something like a cornell box test, but with an emitter light source like a light bulb in the middle? The surprise for me was, that Vlado placed light portals around the bulb.

... I found it:

http://www.chaosgroup.com/forums/vbu...ell+box&page=5

(*) I used light portals around the bulb to tell v-ray where to look for the light source. This matches more exactly what maxwell does - it knows that the light is an emitter and samples it specifically, whereas by default V-Ray doesn't do anything special about a VRayLightMtl. This optimization reduces the noise a lot. The illumination is still coming from the VRayLightMtl on the light bulb, but V-Ray knows where to look for it.

[thomthom]

Re: VRay Rectangular Lights- A Few Queries

I don't see anything related to Cornell box test on that link...

[Micha_cg]

Re: VRay Rectangular Lights- A Few Queries

At page 4 the same image like I posted should be visible - or can't you see the thread, is it for registered users only?

[thomthom]

Re: VRay Rectangular Lights- A Few Queries

Gotcha. Didn't realise it was on another page.

[thomthom]

Re: VRay Rectangular Lights- A Few Queries

I need to read up on Sky Portals. I've never used them before, and after reading this they seem to be different from what I thought they'd be.

[thomthom]

Re: VRay Rectangular Lights- A Few Queries

What I don't understand is the point where he says that it's only the material that's the light source; the light portal only forces VRay to look for light at that point. How can it be that the light portal doesn't contribute to the light?

[Jackson_cg]

Re: VRay Rectangular Lights- A Few Queries

What I REALLY don't get is the general concept of skylight portals forcing VRay to look for light sources- I mean if you had a rectangular emitter plane and the placed a skylight portal directly coplanar with it I could understand VRay being made to look for the light source, but just placing skylight portals in the vicinity of an emitter source seems bizarre. It's seem like VRay is having to make some kind of volumetric/spatial calculation between portal and emitter just to work out if the portal should have any effect on the result. I mean how close should the portal be to the emitter, how many should there be, etc, etc. How does one calculate how many portals, what size they should be and how close they should be to surround a cylindrical or organic formed emitter source?

[Micha_cg]

Re: VRay Rectangular Lights- A Few Queries

What I don't understand is the point where he says that it's only the material that's the light source; the light portal only forces VRay to look for light at that point. How can it be that the light portal doesn't contribute to the light?

I imagine it so, that sample rays goes from a sampled surface in direction to the light portal and than straight through until a surface is matched. The intensity of this surface point is used as light color and intensity.

[dalomar_cg]

Re: VRay Rectangular Lights- A Few Queries

Let me explain what's happening here and why Vlado is using Light portals in this instance. To do this I actually have to explain material emissiveness more than anything else. As you've heard me say, emissive materials are NOT lights and are NOT sampled like lights. In fact they are sampled like materials which means that if they "get seen" by the sampling for a given point then okay, but there's nothing that requires that they be seen. This is not really a very effective way to get accurate samples of the illumination of the scene.

So, by placing rec lights as skylight portals in front of a light emitting material, this is a backwards way to get emissive materials to be sampled as accurately as normal lights. Since they are so close to the light itself, the process that determines what the illumination and color of the light is, simply can't miss the emissive material that is behind it.

As of VfMax 1.5 SP3 this is no longer needed as Vlado added a means to calculate emissive materials as direct lights. All this does is effectively "add" the object to the list of lights to be sampled directly from any given point which forces every point to see that light as opposed to leaving it up to chance. This is one of the biggest things that I would like to get into the new build of V-Ray because it really changes the playing field as far as emissive materials go.