Store with IM for non-rectangualr lights?

[Micha_cg]

Hi,

I ask me, is the rectangular light option "store with IM" possible for distant, spot and point lights too? At 3dsmax?

This option help to much speed up the calculation and I miss it for the distant light very much. Last I replaced a distant light (sun) with a rect light in IM mode and the render time was down at ~50%.

-Micha

[ALTO]

Store with IM for non-rectangualr lights?

Link

[Micha_cg]

Store with IM for non-rectangualr lights?

If I understand right, the "store with IM" should work for any light or? What is a Vray light? All lights?

[dalomar_cg]

Store with IM for non-rectangualr lights?

I don't think the store with IR map option is available for anything other that Rec lights in Rhino and vray lights in max. I'm not exactle sure why, but I think it has to do with how the actual illumination would be calculated. Here's my best guess...

Basically when the IR samples are being calculated there will most likely be several hemispheric subdivisions which would capture the illumination from a vray light (an area/rec light basically). With point lights however (or linear lights to a certain degree) there is only one vector which connects a given point of geometry to a light. Therefore IR's way of sampling would most likely not actually sample the exact point that would relate to the point light. This would not significantly contribute to the irradiance calculate at a given point and would not produce the desired effect. I'm not sure if this is the exact explanation (maybe vlado, wouter, or gijs could confirm/deny) but I think it would have something to do with it.

The thing about the store with IR option is that its trying to get around tracing rays from each light. With rec lights this is a much more teadious task, but is also benifited by using the ir option. each plane has in theory an infinite amount of points (and hench the need for subdivs) and, therefore a significant number of rays to trace. Pointlights are only one point, so comparitively rec lights are significantly more work as far as raytracing goes (but they look so beautiful )

[Micha_cg]

Store with IM for non-rectangualr lights?

Dalomar, so, an area light in portal mode would act like an emitter material. But:
- lights can get a radius, so point light could be not a point only and could simulate with an emitter sphere for the IM rays
- rect light can be very small and the options "store ..." is possible
so, I suppose so, that the IM mode could implemented for standard lights too, but bring good results with not to small radius only (see test).

small rect light box (radius 0.05 units) and the shadow of a box (size 3 units)


My goal is, to get a distant light with IM option to simulate a sun with corona like a light hazed days. So the light source could get a quite big radius and the "store ..." would speed up the calcualtion.

An other question or wish I have: could it be possible to set a light in emitter mode? So, the user could easy jump from raytraced light mode to plain emitter mode.

[ALTO]

Store with IM for non-rectangualr lights?

In VfMax there are 3 types of Vray Lights:
Plane, Sphere, Dome
(and other nice parameters!) >> LINK

...something to implement in VfR!!!

[Micha_cg]

Store with IM for non-rectangualr lights?

Thats true, the light parameters are quite incomplete now and I very miss some of the parameters like radiant power or affect specular for hidden lights (I need the speculars of hidden lights).

[dalomar_cg]

Store with IM for non-rectangualr lights?

I was just stipulating with my previous explanation, so whether a point/directional light would have to be "tricked" to be put into an "emitter" mode I wouldn't really know. To claify however, emitters are sampled much differently that standard lights are, and for the most part its not a very good way to light a scene (we found that out in Sketch Up).

What may help with what your looking for is either a "parrallel" light or an option on a rec light that would only trace rays normal to the surface, so that it could be used for what your looking for.

Or here's another idea, maybe if you had a simple "direct light pass" that could precalculate those rays (from any light) it might be useful. If it was done before IM or LC passes then maybe those prepasses could use that information in their light calculations, or if it was done after the IM pass then the rays could only be traced from the IM samples themselves and the illumination values simply added to the irradiance info. Basically the benefit could be two fold. First if your only tracing rays from a set number of points (each of the IR samples), then that operation becomes much effiecient than tracing for the whole scene. Secondly, since the ray tracing wouldn't need to load any geometry (again all it needs is points), and since it would be its own operation (its not calculating other things at the same time) I think it might be possible to have this process significantly speed up the rendertime associated with directional or first bounce light. This is also be able to be for any light not just vray light, and it wouldn't need to be an option on a per light basis but would be a global thing.

I don't really know where this idea came from, but here it is. If you have any questions or thoughts about it I'd like to hear them as I pretty much just came up with this idea. As for many of the light parameters in VfMax I'd really like to get some more of those in vfr, especially the ability to load a maps for the light.

[joe_b]

Store with IM for non-rectangualr lights?

I believe it needs to be an area light to be stored with irradiance map. Currently the only area lights we have is the Rect light ( well the "linear light" is actually a rect light too in our VfR implementation ). We are missing parameters to handle units of our area lights. Its something we need to add...