Iray practical lighting questions
SnowSultan
Posts: 3,773
There are a couple of things about Iray lighting that I don't quite understand. Maybe I'm just missing something, so please let me know your thoughts.
- Is there absolutely no difference between Point Lights and Linear Point Lights in Iray? Linears don't have any additional options in Iray, and they appear to give exactly the same results when using identical settings.
- If you change the geometry of Point Light (either kind) to anything other than Point, the geometry appears (as expected). However, it still shows up in reflections even if you turn off Render Emitter. Is this a bug?
- I was expecting big differences in lighting when using different light geometry, but aside from the Cylinder option, I don't see what the advantages are over just using a Spotlight and adjusting the Spread Angle and Beam Exponent. You get more options with a Spotlight than with any other light type, so unless you need a clearly defined square spotlight, I don't understand why things like emissive light 'panels' or lights with geometry are any better.
- Can we not use gobos/gels in Iray? Do they always have to be created realistically by light shining through geometry?
- Is there a way to have an object be emissive and either a) not be absolutely blinding white or b) not be visible while still casting the light? Having it be invisible sounds counter-productive, but if you try to place a point light in its place, it will often cast shadows where they shouldn't fall.
Thanks very much for any ideas and suggestions.
Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2025 Daz Productions Inc. All Rights Reserved.
Comments
1. Linear Point lights don't follow inverse square law, as real-world uncollimated lights do
2. Render Emitter controls the direct camera view, but not reflections. You wouldn't want to not see the light in reflections. Reflections are good and aid realism. How else could you see puffy clouds and blue sky in the polished chrome of a 1957 Bel Air?
3. No type of light is any better than the other. Iray purposes to reproduce real lights and scenes, and mimics some of these lights using different kinds of luminaires. Each has its use. For adjusting light geometry: consider that the larger the emitter, the softer the shadows. This, too, follows real world physics.
4. I use gobos all the time. They need point lights, or better yet, a distant light. The sharper the definition of the light (it'll cast a harder shadow), the more defined the gobo edge. And yes, it's *typically* done "realistically" as Iray is a physically-based renderer. When using distant lights, remember that its luminosity is measured as light incident on the scene. Full-noon sun is about 10-12 lumens per square centimeter. The default setting for distant lights is WAY too high.
5. For emissive, choose cd/cm^2 (candles per centimeter square), and dial in a very low number (10-50). This will produce a consistent light output per area of the geometry, regardless of its size. The problem people have is that they don't realize light output in lumens or watts is also based on area. The light appears to diminish when increasing the area of the mesh, but this is only because they've selected the wrong luminance units. To "hide" the mesh, there are a couple of techniques; the easiest is either A) setting Refraction Weight and Refraction Index both to 1.0 (preserves reflected intensity) or B) dialing Cutout Opacity to 0.001 (intensity of the mesh is diminished in reflections).
What do you mean by "cast shadows where they shouldn't fall"? Iray puts shadows where they would physically appear in the real world. Do you mean shadows where you would prefer not to have them? There are some ways to tackle this. Detail your needs a bit for a more concise answer.
"1. Linear Point lights don't follow inverse square law, as real-world uncollimated lights do"
In English please. ;) Do they give different results in any way to normal Point Lights in Iray?
"2. You wouldn't want to not see the light in reflections. "
What I was asking is why the emitting geometry for a Point Light is still visible in reflections when other light types using geometry are not. That's why I asked it it was a bug.
"4. I use gobos all the time."
By shining light through actual geometry, correct? When I said "gobos/gels", I meant a map that can be applied to a light in case you don't have the geometry. For example, using a blurry b/w image to cast a tree's shadow on a scene when the tree isn't visible in your scene.
"5. B) dialing Cutout Opacity to 0.001 (intensity of the mesh is diminished in reflections)."
Thank you for reminding me of that, I remember reading it shortly after Iray's introduction and had forgotten about it. That solves the problem.
"What do you mean by "cast shadows where they shouldn't fall"?
By trying to place a point light where a candle flame would be for example; it may cast shadows in places and with an intensity that an emitter would not.
I tested and linear point lights ts do indeed behave differently than spotlights
I think I know what you're talking about in regards to pointlights casting shadows where they shouldn't. They are rather less invisible than we want aren't they? Basically we can only turn them invisible to the camera (altho set the geometry to point they won't do any of that stuff as they are then infinitely small.) I'm pretty sure that is how they were designed, and personally I like them showing up in reflections, but them casting shadows on occasion can be annoying (I really mis blender's visiblity options here, in Blender you can make any object invisible to a bunch of different things and it is so very handy.) The one thing that is definitely a bug is that If you place a light behind something transparent it turns anti-invisible It keeps on turning my hair white (literally).
Generally I can usually find an alternate location to put my lights. generally I just select my light in the viewport, zoom out, and up the intensity. Since its further away its less likely to be intersecting anything.
For gobos I haven't figured out anything other than doing it physically. Luckily it doesn't seem to incur too much of a time hit. Unluckily, the last time I did it, after a bunch of test renders the light passing throught the gobo decided it was red, and I could not for the life of my figure out why
1. Inverse square law IS English! As I don't know your level of knowledge about this, it's better to start with universal phrases, especially as Google is but a few clicks away. But in any case, for uncollimated light, the light intensity is 1/4 for every doubling in distance. The infographic on this page shows it pretty well:
http://photography.tutsplus.com/articles/rules-for-perfect-lighting-understanding-the-inverse-square-law--photo-3483
As you can see, the light intensity is not linear over distance. Richard H. also explains it here:
http://www.daz3d.com/forums/discussion/comment/36165/#Comment_36165
2. The light from all light sources should be visible in reflections. You are not seeing the emitter, but the light from the emitter. At least, this is how it works for me. If you're seeing something different, please post an example.
4. You're wanting to think of this in unbiased renderer terms. That'll snare you up real fast. A gobo (or scrim or gel) in the real world is placed in front of a light. That's also how it works in Iray. That said, there are workarounds that might work in certain applications, such as using meshes with unusual geometries, using IES lighting profiles, and even creating your own light shaders. There are a couple of products in the Daz store that demonstrate the first; the second is best done by experimenting with freebie profiles from the Web and plugging them in; and the third suffers from woefully inadequate documentation by both nVidia and Daz. What I've discovered along the lines of the latter isn't yet ready for public consumption, but I plan some posts about it soon.
5. A point light is truly omnidirectional, but a candle flame is not. There are a couple of ways to handle this. One good way that preserves realism and takes out the guess work is to find an IES profile that mimics a candle flame. Actually, you could probably use any of a number of profiles that have similar light outputs.Point lights (but not spot lights) accept these profiles, which "shape" the light in various ways. See the sig line of any @mjc1016 post here for a list of IES profile sources.
For the candle problem, I'd think about using a small mesh in the shape of a flame and make it an emitter, if the candle is in the scene. There's always a performance hit when adding gemoetry emitters, but for something like this, it's probably minor, and in-scene lighting it's usually worth it for the added realism it provides.
"I tested and linear point lights ts do indeed behave differently than spotlights"
Thanks, but I do know point lights and spotlights behave differently. ;) My question was if Linear Point Lights and normal Point Lights behave differently.
" I'm pretty sure that is how they were designed, and personally I like them showing up in reflections"
They should unless you turn Render Emitter off, which is why I don't understand why Point Lights do not seem to respond to this. My mentioning shadows being cast where we don't want them is mainly for when you try to use a point light to simulate a visible light source like a bulb; your choices are either to place it in the bulb and have the mesh block the light or place it to the side and get inaccurate shadows.
That problem isn't the light...it's the bulb. To accurately pass the light, the bulb either needs to be modeled correctly...that is have 'thickness' to the glass, otherwise the shader treats the diameter of the bulb as the thickness of the glass OR you need to use a glass shader with 'thin wall' turned ON (and maybe flip the normals of the bulb).
I believe the reflection thing is a bug and as is the fact that a supposedly invisible emitter will be visible when seen through something with transparency.
Try changing the surface you see the reflection on to something like Nickel. When you do this, you will see it is in face the "illumination" of the light's shape you are seeing - not the mesh itself. Depending on the background used, this may "look" like mesh, but you are using a shapped light, so will see the shaped reflection. It can look like mesh depending on the surface, but it isn't. Went round on round on that one here. :)
Another thing we have seen - "shadows". Check the color and make sure you aren't really seeing caustics from bouncing off darker objects. Light bouncing off a purple sphere onto a ceiling may cause it to look like a shadow that when rendered larger can bee seen as purplish in color - thus is really caustics....
We have all gotten so used to faking lighting that our perception of reality has become skewed and this makes Iray look "wrong". LOL
If you want the light to come from the bulb, why not turn the bulb into an emitter? This would be one of the times when a mesh light is better than a built in photometric light. There are sufficient choices in the Daz Uber Iray shaders to get what you want.
Gels can be reasonably modeled by changed the color of the photometric spot. Gobos, as stated by Tobor do need to be phsically between the source and the subject.
I believe Katherine's cooment "We have all gotten so used to faking lighting that our perception of reality has become skewed and this makes Iray look "wrong". LOL" is spot on.
"I believe the reflection thing is a bug and as is the fact that a supposedly invisible emitter will be visible when seen through something with transparency."
I wasn't using transparency in the specific instance I mentioned, but that's good to know as well.
"If you want the light to come from the bulb, why not turn the bulb into an emitter?"
I am, that's why I originally asked about how to make it less blinding or to hide the emitter completely (for candle flames and such).
"Gobos, as stated by Tobor do need to be phsically between the source and the subject."
OK, thanks for confirming.
This also happens if you place a light within a volume to simulate smoke or fog. You're seeing the light from the scattering and absorption. The emitter is still not seen directly, but its effects are. Why the emitter shows with a transparency I don't know, unless it has to do with material reflection and other aspects. In any case, I suspect that if it is a bug, it's not a typical use-case, so probably low down in nVidia's priority list. Most of the time, a light within the scene is intended to be seen..
So the thing is this: we've asked Iray to use physics to render a scene. This greatly adds to realism. But now, we want to take out the odd physics here and there to make the scene easier to model. That just takes us back to biased renderer-land, and ultimately makes achieving realistic renders more difficult. It's much harder to create a fully realistic scene with a biased renderer, which is why RenderMan experts in Hollywood can demand so much money.
Couple that with the limitation that in order to have the speed, everything must fit in the video card's memory. Which means some short cuts NEED to happen...that properly modeled bulb is going to double the poly count for it. which means more memory.
For the last point in the original post, you can have a spotlight or point light not show up or reflect by setting "render emitter" to "Off."
For mesh lights you have to set the cutout opacity to .001 to get a similar effect, but it does work when I've used it. I like to leave them visible and render with them offscreen because of the nice round eye reflections they produce, but that's a personal preference.
"For the last point in the original post, you can have a spotlight or point light not show up or reflect by setting "render emitter" to "Off."
It seems that Point Lights with their geometries set to anything other than "Point" are behaving strangely though; the emitter itself is still visible in reflections. Is this normal?
"Behaving strangely" isn't much to go on. In what way?
Render Emitter only has no effect on a reflected image, for spotlights or point lights. You're not seeing the emitter in this case, but the light from the emitter. There is no doubt this is working as designed, and is normal. Render Emitter is only for the direct view of the light source, with nothing intervening.
There is no "emitter" when either pointlight or spotlight is set to Point, so you won't see it there. The light is infinitely small. But change to an area shape, and you will see the emitter in a reflection. It's normal.
If you really, really don't want to see the effect of reflected light when the emitter has been changed to an area shape, you can try setting the scale of the luminare to something near zero -- 0.001 for example. This will alter the physics of the light emission in some ways (shadows can be affected), but it might work for you.
Finally, if the shadows effects are definitely not what you want for a scene, you can put Iray into a quasi-biased mode by changing tfrom Photoreal to Interactive. (Interactive is a poor choice of words, but just go with it.) In that mode, there are shadow controls for the lights. You can turn shadows on and off for individual lights, and alter the shadow effects in ways similar to 3DL or other biased renderer. You lose some other features when in Interactive mode, but most of the important ones -- like PBR shading -- are retained.
Or, thinking about it another way, perhaps one could say that_all_ the surface effects you see (diffuse, specular, etc) are the results of reflecting the light from the light source?
@SnowSultan "
"If you want the light to come from the bulb, why not turn the bulb into an emitter?"
I am, that's why I originally asked about how to make it less blinding or to hide the emitter completely (for candle flames and such)."
I see. Here's an example. I've attached the test render. Figures skin color is grey. I've also attached the settings. Is this what you wanted?
We should probably keep to the "direct view" definition because it also happens even in clear glass that is made invisible with refraction weight/index set to 1.0. The resulting shown light from the emitter is not from a reflection.
I'm not sure how Iray handles the logic internally, but it may come down to a simple conditional test that looks at ANY geometry between the source and camera, regardless of whether it's a reflecting surface or not.
Elsewhere a couple of weeks ago I posted an explanation of how to hide the geometry of an emissive object, while still allowing for the light. I'll summarize here. These approaches have specific uses, and differ in ways that help you select the one you want for your particular scene. In all cases, apply the default Iray Uber surface first (clears any old values), then apply the emissive shader. Set up the shader to produce the light you want before the next steps.
1. Dial the refraction weight and index to 1.0. This will cause the geometry to disappear, and will stop the geometry from casting shadows caused by other light sources. The light is VISIBLE in reflections at full strength. This setting is intended for planar one-sided meshes, when the unlit side is facing the camera.
2. Dial the Cutout Opacity to 0.0001, the lowest it will go while keeping within the Limits. This has the same effect as #1, but the light is INVISIBLE in reflections.
3. Apply the Advanced Iray Node Properties script to the \geometry, and in the Parameters panel, turn Matte Enable on. This has the same effect as #2, but creates a HOLE where the geometry was (transparency behind) that is intended for post compositing.
(For example, you could use a geometry of a flame as an emissive object. To make it bright enough, you may have to set the intensity so high that the color and details of the flame is burned out. In post, you could put a picture of actual flame there instead. Or if you wan to keep it straight-up 3D, you could re-render the scene with just the candle set at a level to provide proper coloration and details, then composite the two in Photoshop. Such a use is why Iray has the matte object feature.)
Other notes: Use cd/cm^2 for the Luminance Units. The reason is that the centimeter (not foot or meters) is the default scale used in D|S. This helps you to keep your units of measure straight, and avoids confusion. As cd/cm^2 takes into account the area of the emission surface, you do not need to keep adjusting the luminosity if you change the size of the emitter. A value of just 20-25 will produce a strong light. This is 20 to 25 candles per square centimeter emitting from the geometry. There is no absolute direct correlation between candles and lumens, since the two are measured differently, but for basic calculations, you can figure roughly 12 lumens per candle.
Those are very good tips Tobor, thank you. Whatever is happening with the Point Light emitter being visible in reflections, I'll leave that for others to ponder. As for the differences between the Point lights in Iray, I'll assume they're identical.