Hi Stu,
The square falloff is the one you want for any non parallel light. If the light spreads, it does it in two directions, hence the square. As it becomes less bright with the increasing distance, the term inverse is used.
The word “accurate” leads me to go more into detail:
It is not the full story, as the substance in which the light exists is perhaps filled with moisture and dust, if we talk about air. Those particles will scatters the light and lower its brightness then faster, talking about the light direction axis. Think about dirty water with particles in it, light is not only in air.
To set up a longer falloff, will establish a brighter light source, as the light has to be brighter to reach further (even the value “Intensity” isn’t changing!), compared to a smaller falloff. To set up a light with a certain value and inverse square fall off, might be a perfect simulation of an existing light source, but to pull then the falloff and call it then again a perfect simulation might show the problem clearly.
As long you create no visible light (Volume etc), any set up in the scene is just half of the equation, as only the correct surface for example will make this light visible. If you have not measured the “bounce/reflectivity” of surfaces in your scene, so they have a correct relation to each other and work accordingly to the light source, the set up might not work at all.
The bounce of light is simulated (not very close to the physical correct scattering of light BTW) with the Diffusion setting. This setting is mostly eye-balled in its value. If you like to have somewhat correct values, use a light meter (in the real world ;o). Measure the the light source as well the surface and write down the stops of light you get. This gives you a rule of thumb about the light - surface ratio. It is a fast simulation, to shorten the render time. The light scatters and so the camera position is critical for practical light. This is done as well in C4D with the Diffusion. The Phong/Oren Nayar/Blinn with its values needs to be mentioned here, but for any GI calcualtion, I haven’t found any influence in the scene—in terms of secondary light distribution. The detail slider in the light source might help for the initial bounce, and with that a little bit in the secondary, but it is more useful to simualte the size of the light source (At least that is the result of my observation)
There is another thing, it is called IES light, which works of course only if the scene scale fits, but if all what has been said already is acknowledge, these IES might help to give some “references” in the scene.
Last but not least, rendering for light simulations should be done in floating point and linear. I even don’t start to talk about light temperature. Commonly ignored as “Fix it in Light Room” if we talk about photography, the average spectrum of the light sources might change the brightness representation of the colorful materials, as colors shifts and corrected back, leave always a problem. In the few color science course that I joint, the term white was always set as something that is not really defined. (Which color temperature, which brightness, etc.) The gray card is perhaps and only to a certain degree the only help here, as setting this neutrally might take the scene mood away.
Of course this might be a little bit nit-picking, but light is such a deep discussion and to get it “right”—even as rule of thumb—is tricky. Sadly enough, there is a lot of wrong data published, which makes it not simple to even start explaining light and its nature.
What is mostly missing is, that any light source is visible (e.g., the bulb) and that there is a reflection of this on the surface. There is practically nearly no surface without reflectivity—even charcoal has some…) Besides that, atmospheric (moisture etc) brightens the space and adds to the captured result as well. (I avoid talking about lens flare here ;o)
However, we are used to see the world centered around a middle value, what ever we decide is the middle value, from there the darks and brightness values follow the “gamma” curve. Light works linearly and our perception not. It becomes even more difficult when we think about celluloid/film or its simulation in DSLRs etc. The key of film and its patience with sloppy exposure is certainly based on the main part which was logarithmic and to on the lower end (Toe) and on the hight end (Shoulder) the material was even more forgiving with light (Hence the high light roll off). Why do I mention that, because many people use photos as reference to set up a scene. Not knowing perhaps what camera-responce curve (film curve simulation) was used. Perhaps even worse, render in integer and avoiding 100% + values to not clip, which can be only done if the scene was eye-balled in its values. This was done over decades and as an result we got to a certain degree used to it. CG look is certainly (partly) an result of it.
In short, if you need a rule of thumb, don’t trust your eyes, get a light meter and check practically and use the reference in your scene. Work in “linear-light and floating point precision” and measure the values. E.g., mid gray is then ~0.18 and a stop above is ~0.36. Nothing that can be measured in a gamma space, like sRGB 2.2.
I hope that was not too long, but I like to answer fully, and not just separate the rule of thumb from a needed context while mentioning the word accurate. Light is a complex and wonderful tool in our art. It was certainly my first love, starting doing light-shows on stage for bands when I was just thirteen ;o)
Good luck with your set up.
Sassi
