[eric nixon]

Just noticed that if the mxm is setup as a spotlight but also has a bsdf/s then that bsdf shows up in reflections, but remains invisible to camera.

It would be nice to have a discussion / advice about setting up spotlights that can be seen in reflections / in the camera frame. It makes sense to use an extra emitter, but how to judge the intensity so that the reflections match the direct illumination? Use the same hdr for both?

[eric nixon]

I assume its not possible to create a light-house type beam of light with this Spotlight, (or car headlights, or searchlight etc.) I'm assuming the barn-doors option will just clip the edges and we really want a 'concentric lens'/'focussed beam' option. < maybe in the future?

[Mihai]

If you want the light beam to show up, you need to add volumetrics. This hasn't changed...

[eric nixon]

erm thats not what I meant, I mean is it possible to get a collimated beam of light from these spotlights?

[Mihai]

Yes of course, is it not working properly?

[eric nixon]

Ok, so how do we do it?

btw this is a collimated beam;



It starts from the lens as a disc of light and then carries on with a fairly flat beam, how do we do this? all I see is a diverging spread of light starting from a point?

[eric nixon]

Spotlights giving weird lighting, guessing NL have found this out also by now..



The material on the ball has no texture, its just plain bsdf with roughness around 45.



caustics have gone crazy here.

Please can we have a radius option for the next version of spotlights so that its no longer just a point source of light. I have faked a torch beam type look here, by moving the 'spotlight' back into the torch body, but other situations, such as a lighthouse or flashing siren dont allow that kind of positioning.

[Mihai]

But you can use the blur parameter to increase the size of the virtual sphere. That will give you a more evenly sized light beam.

[Mihai]

Can you upload please that scene with the 45 rough sphere?

[eric nixon]

But you can use the blur parameter to increase the size of the virtual sphere. That will give you a more evenly sized light beam.

Hmm, renders slowly, looks weird. Basically same as if I had cloned a 1000 'spotlights' and varied there position, a more logical approach (I'm guessing here that having a point source rather than an area is essential somehow) would be to have one spot but calculate that the emission begins at a certain radius (i.e. virtually offset the lightsource behind itself a certain distance to achieve that radius).

[Mihai]

Of course it will break if you use such a huge sphere near a wall. The sphere intersects the wall. The same as in real life if a huge focusing lens would penetrate the wall if you try to put it this close. In real life the focusing lens couldn't be bigger than the housing of the light. There are some improvements here that can be done for sure, but in many cases increasing that spheres radius (by changing the blur parameter) will work well, it will give you a more parallel beam. IE, if you have a flash light where the front lens is 5cm in diameter, then that's the maximum width that beam of light is going to take. How would you do it in real life if you want to have a 1m wide beam of light emitted from a flash light or focusing lamp, that's placed 10cm from a wall?

[eric nixon]

I wasn't aware that the sphere was intersecting the wall, I wasnt aware of any sphere as I was applying the 'blur' maybe it needs a new name. anyway i'll try again, so thanks for replying with info.

Seems that a sphere is not a good shape, and it needs to be a disc. unless it is a disc and your calling that a sphere, who knows....

[eric nixon]

Working now but still seems slow, second image shows version with 0.01 blur (spotlight moved back) which is rendering a lot better as you can see.



At least now I know what you meant earlier, that 'it just works' yes it does but very slowly and being a sphere shape makes no sense, very difficult to work with.

EDIT: taking a second look, its clear that the comparison isnt so good, because the light beams have different shapes. Maybe the blur-setting isnt the slowdown, maybe its the overall cone angle. (which is a shame, no?)
I do suspect a disc shaped 'blur' would be faster, just a guess... I'll do some more tests tomo.

[Mihai]

I'm not sure, but the sphere shape isn't selected at random, as opposed to a disc shape or other shape, It's to enable the sampling to be done properly in Maxwell. It's the same why a small sphere was chosen for IES, and in fact these spot light emitters work very much in a similar way to IES, except you have interactive control.

[eric nixon]

But if we're representing a lens in this case, which is disc shaped, a sphere is prob slower and difficult to work with.. esp working blind because I cant see the sphere.

Anyway its a fair bit quicker than ies, this is sl 18 32mins;



If its not quicker than ies then whats the point, just for convenience?

I've embedded the spotlight (with 30ish blur) inside a lens roughly the shape of a baked bean can, secondary emitter is also inside the lens to the front.

Unfortunately the reflection shows the hexagon shaped sphere (That reflected hexagon shape thru concentric lens must be the source of the 'rough 45 bug' which actually isnt a bug, mentioned earlier)., also the physical correctness is not perfect.... because the lens is hidden to gi only... but its close to real