Iray, Lightbulbs, Lumens and Physics
Neil Clennan
Posts: 45
I've started running some experiments to better understand Iray, in particular, the Emission settings in the Surfaces tab. I am puzzled because Luminance setting does not seem to be working the way it should in a physics based system when I switch between lumens and cd/cm^2. I'm hoping someone can tell me what I'm doing wrong, whether I'm making some incorrect assumptions, ot my math is wrong, or something else. Technical nuts and bolts seemed like the right forum.
To start, I set up a scene with a sphere (8 cm diameter) as my light source. I set the Environment Mode to Scene Only, so the sphere will be the only source of light. I also add a box around the to simulate having walls on all sides to reflect light back (the set did not come with 4 walls).
I want the sphere to simulate a 100 watt light bulb, so on it's surface I set the Luminence Units to lm (lumens) and the Luminance value to 1600. That matches the value indicated on the packaging on a light bulb I'm looking at now, but I could get the same result by setting the Luminence to 100 W (watts) with a Luminous Efficacy of 16 lm/W.
To use reasonable camera settings for poor indoor lighting (a single 100 W lightbulb), I go to the Tone Mapping tab of the Render Settings and set the Exposure Value to 5.70. I acheive this by setting F/Stop to 1.8 and a Shutter Speed (1/x) of 16, but it turns out it doesn't matter which combination I use as long as I arrive at the same EV.
When I render, I get this:
Now I want to understand how the cd/cm^2 units work. the cd stands for cadlepower, so to convert lumens to candlepower I divide 1600 by 12.57 to get 127.287. The surface area of my sphere is 4π 4^2 or 201.06 cm^2. So to get my value Luminence it's 127.287/201.06 to get 0.633 cd/cm^2. I put that value in and set my units to cd/cm^2 and I should get the same amount of light, but instead I get this:
Does anyone know what I did wrong?
(Sorry for the huge gaps; I have not figured out this editor and how to embed images the way others have.)
Daz 3D is part of 
Connect
DAZ Productions, Inc.
224 S 200 W, Suite #250
Salt Lake City, UT 84101
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
I don't want to be obtuse (although I've been told I have a gift for it), but if setting your light using watts works, then why use something that isn't working?
use this https://www.rapidtables.com/calc/light/watt-to-lumen-calculator.html
Rather than 0.633 try 63.00.
https://books.google.co.uk/books?id=ETQvBQAAQBAJ&pg=PA109&lpg=PA109&dq=cd/cm2+of+incandescent+light+bulb&source=bl&ots=3jDSW_Sh7j&sig=ACfU3U0phuBkNH82xBUypwT_fJZGDocEpg&hl=en&sa=X&ved=2ahUKEwjhrfOa8srpAhUHRBUIHVy1DaUQ6AEwC3oECB0QAQ#v=onepage&q=cd/cm2 of incandescent light bulb&f=false
It's a fair question. Mostly it's because I want a deaper understanding of it, and have aparently failed. There may be edge cases where it is useful, and luminence over area is the default used by DAZ Studio.
That calculater is a simple multiplier to give you total lumens rather than lumens over area. The calculator does confirm that I'm doing the watts to lumens part of it correctly.
I'm not sure where you get tthe 63.0 value from... well, clearly from the book you linked to, but there are a lot of pages there and I'm not sure what part tells me what I want to know. I'll spend some time with it though. Thanks.
The math for the lights is correct.
But, there's part of the equation that isn't being considered, Render settings.
The default setting are closer to an exterior photographic setup where one is dealing with bright light.
When you move to an interior, many of those settings need to be adjusted to lower light conditions.
F/stop, Shutter speed and Film ISO, should be your first go to's.
F/stop's and Shutter speed need to be adjusted to smaller numbers, film ISO to a higher number to increase the "Brightness" of lights.
I don't recommend adjusting Exposure value directly, as it simply changes the shutter speed without adjusting the other settings.
Well, you're very gracious to say that mine was a fair question, Neil, but right after I posted it on impulse, I had to go out for a long while, and the entire time out I was cringing for having posted it. There's an old wisdom: "What you resist, you become." And I realized that in a distracted and abstracted moment, I had done just that; you see, I simply cannot abide people who leap into a question thread with "Why would you want to ..."
I detest it. And I did it.
I will not resort to Medieval methods of self-inflicted pain to atone, but I will certainly beg your forgiveness, and I definitely repent.
I admire your quest for greater understanding of this question—which is one I should have, too. So I will follow it with interest, and know I will learn something.
You are right that I didn't try changing the the film speed (I did vary the F/stop and shutter speed), and when I did that, even though I arrived at the same EV, the result was very different. That was very good for me to learn, so thank you for that.
However the point of my question wasn't how to get good indoor lighting and camera settings. The question is, given the same camera settings (tone mapping) did not change at all from one render to the next, and given that the math is telling me that the light source output should not have changed, why was there such a huge difference between the two pictures? Why did the light output, in fact, change?
Don't kick yourself too much mate.
Asking a question like you did, "why do x if y is working?" is a legit question.
It not only informs you of what the thought process of the person you're asking, but also gives the next person that reads the thread something to consider.
There's a lot of things that people do without ever simply asking the question you did.
I think your problem is that you are using the bulb size as reference, whereas it actually is the wire that radiates.
I'm not sure if i will explain this right but here goes :)
Tone Mapping at 30/4.00/200
The left hand image is the 8 cm sphere at 100 watts.
The middle image is cd/cm² at Luminance 8.12 This was the 650 lumen output of the filament in the book by 80, for no other reason that that was the 0.1x8x10 which was the sphere lowered to the size of the filament (although one is cm² and the other is cm) :)
The right hand image is a 1 cm sphere at cd/cm² at Luminance 650 which is the 65 cd from the book for the 0.1 cm² filament multiplied by 10, 0.1x10=1 which makes the software think it is a 1 cm² filament :)
Nothing scientific just playing about with different scenarios :)
Click on image for full size.
The real issue is that Iray simulates light behavior / physics. That doesn't extend to the actual illumination provided by a given lumen or cd/cm2 setting. Further the ISO, shutter speed, and F stop do not simulate an actual camera.
You will need to make your klighting setting based on results. Do note that the Iray spotlights do have shape and geometry settings and will accurately simulate the effect of photography softboxes of similar size and shape.
With Iray, the old practice of using emissive lights for realistic llighting is no longer necessary and will take much longer to render than a Iray spotlight for any geometry more complex than a plane with a very few divisions.
Just to play devil's advocate, what lead to the assumption that the lumens/watt was for an incandescent light bulb? I've not come across claim this in any documentation. And are we talking about a light bulb that's frosted or one that isn't? A 100 W LED would produce much more light, while a toaster operating at 100 W would produce much less.
Personally, I don't care what the numbers or units are. I'm not trying to automate my lighting, so there's no need to. If there's enough light of the type I want coming from the direction I want, I'm happy. Besides, you can render your light sources to canvasses and do all that light level and colour adjusting in post.
That is just the way some people like to do it. I can count on one hand the number of renders I have tried Photoshop on and had to do very little to them. I come from a Photographic background having been enveloped in it for sixty plus year so lighting and camera settings is how I do it, and I have put more photographs through Photoshop than I care to remember. These questions are asked in here because this is the Technical Help section where they might be answered and not in the Commons where the answer is usually "don't do it that way when it is easier this way" which isn't what the poster wants to know.
I don't think so. I'm calling the sphere a lightbulb, but the sphere is actually a thoretical/mathematical representation, so I could just as easily call it the filament. The math problem remains.
I'll probably try messing with those settings in my scene, though I don't think it will solve my math problem. Based on what you're showing me here, I think I can see the part of the book you were referring to. That example in the book seems to match the math I was doing, seamingly proving that I made no mistakes, but I'll give it a closer look.
Well I don't think you can believe advertising on LED boxes in the stores. When I replaced all the 60 W incandescent bulbs in my house with LED lights the claim was by the manufacturers that their 800 Lumens LED lightbulbs were the equivalent of 60 Watt incandescent bulbs. Having bought them for every room in the house I don't believe it though. At the source the LED lamps seem even brighter than the incandescent at the source but the falloff from the LED light seems to happen much faster than the fall off for the incandescent. I'd guess twice as fast as more. Since light doesn't actually change behaviour based on it's source then I think those 800 lumens bulbs are only about the equivalent of 30 watts or less, not 60 watts.
By the way, when I 1st started trying to use iRay & iRay lighting in DAZ Studio I also thought that I could do that by using real life lumen, watt, and other physics measurements but the truth was iRay is only a simulation of a very limited amount of light that would be present in the real world so you will find that your scenes are underlit if you try to use real world measurements to add light to your scenes.
Finally what I've had to resort to is just caring whether of not the lighting looks good in the render. I don't even look at the strength of it, I just crank it way up by a factor of 1000 or more, eg change candula/meter squared to kilocandula/meter squared and so on and then start lowering the intensity, sometimes, by lowering the number (which is 1500 by default in the emissive settings).
I don't think I see your point. If Iray is not accurately simulating actual illumination provided by a given lumen, then I don't think we can say that it is an accurate simulation of light behavior or physics. Unless you are pointing out that no simulation can anticipate what the results will look like on a given monitor or when printed by a particular process. I would certainly agree with that, but my purpose in posing the question is not to learn how to set up a realistic render (though that is certainly a worthwhile endeavor), but to understand a particular set of settings. Your other comments about alternate ways to light are also useful, but not in the context of what I am trying to understand.
What I have is an age old process of establishing a frame of reference (the first picture and the settings that produced it), changing a single variable, and observing the result (the second picture). In this case, I expected no change in the result, but instead got a significant change, and I'm trying to understand why. I see 2 possible outcomes.
1) I've misunderstood the math or physics of the situation, and those 2 values are not equivalent after all.
2) Iray (or DAZ Studio's implementation of it) is broken as a physics based simulation.
I've used 3Delight for many years, and a results based workflow is very familiar to me, I had heard that Iray was useful to architects who need a good representation of reality.
iRay's handling of illimination is improving over time but it is not yet accurate to what a specific real world light source produces.
That doesn't mean it doesn't accurately model the physics of light bouncing etc. It just fails at turning that into how much illumination hits the camera lens.
You can easily verify that iRay is modeling light bouncing by positioning two mirrors for the infinity effect, putting enough light to see the reflections and an object and camera between the mirrors and do the render.
Yeah, the number of rays of light really bouncing around are orders of magnitude higher in the real world
Candela/Candlepower and Lumen are two different things. Candela is the measurement of the light produced by the light source, Lumen is the measurement of the area illuminated by the light source.
Something else that occurred to me is that I read somewhere that the values in the Luminance setting is based on Kcd/m² so when using cd/cm² you have to shift the decimal point e.g. 127.00 becomes 12.70.
The point is that it is the filament within the lightbulb that produces the light, and the filament has a much smaller surface area than the glass bulb that surrounds it. By treating the surface of the bulb as the emissive surface, you're dividing by too big a number and ending up with too dark an image.
Personally, I just put the scene on Iray preview and increase the lumens on my lights (usually resized cube primitives with emissive surfaces) till it looks how I want it. Sure, my iray people must buy their lightbulbs from the places that supply lighthouses, but if it looks right, who cares?
@Sevrin "Just to play devil's advocate, what lead to the assumption that the lumens/watt was for an incandescent light bulb"
The OP also used a specific Luminous Efficacy value. While I didn't check the value, that is what would make it an incadescent bulb.
But the luminosity was set for the sphere as a whole, not for a filament, so if there is a problem with the calculation I don't think this is it.
@Neil Clennan "I don't think I see your point."
See @kenshaw011267 response. I consider behavior and physics to cover bouncing, reflection, refraction, absorbence and translucency, not actual illumination values.
Perhaps more salient, if one is adament about using acual lighting parameters and actual camera settings, one will be disappointed and frustrated. Of course you can use a results based workflow, Iray is predictable once you understand the results you get for particular settings. No different as 3DL, except for Iray's basis in reality (objects cast shadows and reflect light among others).
No PBR is perfect at modeling light bouncing, it is impossible to do since there are effectively infinite rays coming from any light source. So they all have to fudge at determining how much light is actually hitting the lens. I have noticed a definite improvement over the last several versions of iRay in this.