Please view the video.

Is Radiosity the only way to get the desired effect I am looking for? Is there a way to make a group a light deflector or blocker? As you would a deflector with cloth or hair pr dynamics?

Click to view attachment

Do you have cast shadows turned ON in the light properties? The shadow darkness will need to be set at 100% as well. If the darkness is set to 80%, then a percentage of light will pass through to lighten the shadow by 20%.

Mark's reply is most likely, but objects also have a "cast Shadows" prop which should be on. Default is on, but it's always possible to switch it by accident.

Hmm, I did some testing and it appears that the light still penatrates through double thick walls. I would setup a test project and send it to A:M reports.

Your suggestion although interesting does not work.

I think I have found a possible solution "Light Lists"

Yup light lists seem to do the trick. I wish there was a real world light option so light rays couldn't just pass through solid impermeable objects but Light lists at least let me force and object to not receive light from the light you desire.

Ok, did more testing. If you set the "distribute in passes" in the cast shadow properties to ON or set the number of rays to more than 1, the light does not penatrate. I would still send this to A:M reports, because a 100% shadow should not allow any light to pass through an object.

Light widths greater than 0 simulate occlusion falloff of shadows. Set the light to 0 width for stark black shadows.

Click to view attachment

Slight correction...Light with only one ray with widths greater than 0 simulate occlusion falloff of shadows. Set the light to 0 width for stark black shadows or set the light to have more than one ray.

It's not a bug, it's a feature!

That's interesting. Never would have thought of that. I did play with the width, but never set it to 0.

Did you view the video Robert? I dont feel we are talking about the same thing. The property you are speaking of controls whether the light has fall off or not but thats not practical In my context of my problem.

Thanks that was very helpful and It also forced me to learn about light lists.

Interesting data

Click to view attachment

Setup, notice no lights are in this completely sealed room.

Click to view attachment

I did watch the video. The property is not about light fall off. It's about shadow being made less than completely black.

Can you post your test case? In all of your shots above you have light width not set to zero.

Click to view attachment

hear you go.

Does anyone know what is happening when you turn on distribute in passes in the cast shadows, light properties? I can not find any info on it anywhere.

I took a look at your project - not sure this is what you are trying to do

FIRST - you probably need to set the options for the ground plane (in the chor) to CAST SHADOWS = ON - currently it's off

then what I did for testing:

1) turned off all lights except light 4 (left the white light)

2) made the width of white light = 0, made darkness of shadows = 100%

3) moved the light outside the room

4) did on screen render

5) changed width to 10 - did on screen render

You know that you have Darkness set to only 80% on all those lights, right? That guarantees light will come through the walls. Is that still what we are trying to solve?

Hey Guys

I appreciate the the help but the problem was solved a while back by mtpeak2 . Change darkness to 100% and turn distribute in passes to on OR change the rays cast to 2.

The problem with Robcats solution is that changing the light light width to 0 gives extremely undesirable results. I cant think of a situation where I would want a razor sharp hard shadow.

What I would like to know is where I can find information about distribute in passes

Distribute in passes is like setting the rays to more than 1. It will multiple that 1 ray by as many passes set.

Oh, so if I render with 1 multipass and 1 ray and set distribute in passes to on, it is actually off because multipass is equal to 1?

First, Z-buffered Klieg lights are great for soft shadows and are way faster than those bulb lights.

But you want soft bulb lights?

Recipe:

Darkness 100%
rays >2 (much much greater to avoid grainy shadows)
width >0

Show me case where that doesn't work.

Remember the original premise of your question was how to make lights behave like real lights. The answer is to not use light behaviours only exist in CG. 80% dark shadows don't exist in real life. In real life if a shadow is less than 100% black it is because light from some other direction is filling it in, not because the original light's shadow is only 80% strong.

I remember Yves describing it differently. If you had 10 rays and five passes it would do 2 rays per pass. This was so high pass renders would not be doing time consuming over kill on multi rayed lights.

Or is it the other way around... it works that way if you have distribute off?

Not quite. It's equal to 1 pass of a multipass render. Lights jitter in multipass, so it distributes the ray once, so the more passes the more it distributes the ray in a different location, creating a softer look. It's kind of like an AO render, the more passes the softer the shadowed area looks. Also, the larger the width of the light, the wider spread the rays will be distributed. The width of the light is related to how much the light jitters between passes.

I'm not an expert with lights, so take it for what it's worth. Hopefully someone will correct me if I'm wrong.

Robert, to my understanding, 10 rays in 5 passes and distribute in passes OFF=2 rays in 5 passes and distribute is passes ON. Distribute in passes allows you to set the number of rays to less than what you want. I'll have to see if I can find Ives explanation.

My Eyes just crossed. Sounds like we need a a concrete video tutorial or feature focus explaining this and showing the major differences and why you would choose cirtain options over the other.

The "1 ray" term is iffy to me anyway. If you really had just one ray... there'd only be one lit pixel on your render.

That is correct.

Ok, I did some test, Robert is correct.

10 rays distribute Off=50 rays with it On with 5 passes. Render times are the same. I guess it will help more with multi lights.

With multilights, normally, you should balance the total number of samples among the lights according to their importance in the scene. In other words, if your scene have 4 lights, it is not very efficient to set them all to say 50 samples. 50 samples total among all lights should be good enough. Brighter and larger lamps would require more samples than dimer and smaller lamps. It can be tricky to figure the right balance but in the end, it is worth it because of the reduced render time.

are "rays" = "samples" ?


I was experimenting with a light today and found that even 100 rays still makes a noticeably grainy shadow, but 100 rays is the upper limit allowed.

Thanks Yves for the explanation.

I'll try to explain.

Real lights have a width or some size, area or volume. Because of this, the shadow they cast is soft. The shadow softness depends on the light width and the distance from the shadow casting objects. In computer graphics, the most used trick to generate soft shadows is to sample the surface of the light many times and count the proportion of shadow rays that are occluded by any object to determine the strength of the shadow at some point in the scene. A technique very similar to ambience occlusion.

If you cast only one shadow ray, then this ray is either occluded or not and thus the shadow is either on or off producing sharp shadows. There are no ways, with one single shadow rays to generate soft shadows (note that there are techniques to generate soft shadows without doing multi-sampling but they don't use shadow rays). Because of that, traditional CG renderers use the shadow falloff trick. This trick just uses the distance fron the light to the shadow casting object, the distance from the shadow casting object to the shadow receiving object, and the width of the light to compute a shadow "strength". This is why the light seems to pass through walls.

In A:M, if you use only one ray on a light and the light have some width, then the shadow falloff trick kicks in. It kicks in even if light width is 0 but the math makes the shadow falloff only at infinity so in practice, there are no visible falloff. The minute you use more than one ray on a light, then the lighting algo tries to use that to compute true soft shadows.

The shadow darkness property is used in all cases: single ray or multiple rays. It gives more control to the artist. In principle, it should be set to 100%.

One last note. If you are looking for as real light behavior as possible, then you should gamma correct your final render. Of course, this also means that you should correct your material amd map colors too to compensate. This is called "Linear Workflow". If you don't do that, then your light falloff will be too short on distance and too quick on object edges, even your solf shadows won't be so soft and in multiple light scenes, the light additions will look too strong.

Yes. Sorry, I'm just too much into unbiased renderers these days.


It depends on many factors. If you have a very large light, or the light is very close to the shadow casting object, then indeed, 100 rays may not be enough. So the total number of rays is very scene dependent. For normal size lights, though, 50 rays is usually already quite large.

Hi Y man

Always good to here your thoughts on a subject. Is there any way I can get you to update the AM tech ref with this info? The shadow options do not seem to be in there, like "distribute in passes, Darkness, rays and so on. I would gladly purchase some screen capture software and usb headphones and mic for you if you were interested in doing a feature focus video tutorial on this subject.