Lots and lots and lots of lights (how to render)

[Atom]

I played round with Matt's VDB cone example file. I was able to get the cones to align to the hemisphere but I am not sure how to actually make them emit light?

ap_me_volume_cones.hipnc

[Atom]

I went ahead and removed the GeoLight and installed a VolumeLight instead. I set the Geometry field to the VDB object and assigned a BillowySmoke to the Material field. Inside the BillowySmoke I mapped Density to the emit feature so I could use the color ramp.

While this does improve the quality as far as noise is concerned, I am not sure if the spotlights are casting shadows anymore?

 

Enabling Self Shadowing does help.

 

ap_me_volume_cones.hipnc

Edited May 13, 2016 by Atom

[MENOZ]

Hello!

I have a question about the hip from Oskar. I don't understand how the light shader is computed in order to get the dot product. It really put in question many aspects of what I tought I had clear before. is this a special case?

The light shader is called by the surface shader of the currently shaded sample isn't it?

What I don't get is that the dot product seems to be between I and N. N of what? emitting surface? I is the vector from camera to surface. Is I changed to something else in this case? like L, vector from surface to light? this would make more sense to me.

thank you!

 

[anim]

10 hours ago, MENOZ said:

... I is the vector from camera to surface. Is I changed to something else in this case? like L, vector from surface to light? ...

AFAIK yes, N of the geometry the shader is attached to (emitting surface)

I is not always ray from the camera to surface, tooltip says from Eye to Surface, but you can think of it as incoming ray direction, so in this case the Eye is the surface that is sampling the light, so exactly like L in proper light shader

[mestela]

On 12 May 2016 at 3:54 AM, bandini said:

Yes, VDB volumes instanced to points is where I ended up. The issue with the setup is accounting for occlusions and shadow casting. I think some kind of raytracing would have to be implemented in the volume lookups.

Yep, and such a setup is probably possible (I haven't given it much thought yet), I was more thinking this would be a way to setup the most pared down volume possible, that you'd then hit with regular lights. 

[MENOZ]

thanks a lot Anim.

is there a way for me to know how houdini interprets "eye" on a given situation? how would I know that I is the surface that is sampling the light in this case? How could I guess it? How should I think? Is it something I can deduce somehow?

[anim]

as I said, think of it as Incoming ray

so when you see surface A directly the Eye is the camera

when the surface B sends a ray to surface A the Eye is the location where that ray was sent from

- so in this case as each sample in the volume sends the ray towards geometry light to evaluate it's color and figure out occlusion, the ray is not sent from Camera, but from the volume sample, therefore I is direction from there

but if you turn on Render Light Geometry, then the actual rendered geometry surface will use Camera as Eye, since  that's where the ray is coming from

- likewise, when reflective surface sends the reflect ray, for the hit surface the Eye will be location where the reflection ray was sent from, not Camera, the same from refractions, etc...

[MENOZ]

thank you, now that you make me think of it makes totally sense