Lumens from Windows

[3ddesign]

Has anyone established a luminous power value that they use for a Vray plane light at an exterior window? Web articles are saying that unobstructed vertical windows should have between 100 - 200 lumens per square foot for unobstructed daylight and up to 500 for direct sunlight. If that is correct, a 4 foot by 4 foot window would produce around 3200 lumens of light, the same as two 100 watt incandescent light bulbs. Surely this can't be right.

In a test render of a recently completed building, a 4 foot by 4 foot Vray plane light needed a Luminous power value of 145000 to match a photo I took of the room using the same Vray camera settings as the digital camera settings used to photograph the room.

I've read all the posts found by searching for Luminous power and/or lumens and am not sure I totally understand how to begin using real world values for daylighting an interior scene. Any help, pointers, links explaining this would be very appreciated.

Craig

[3ddesign]

I've been reading about daylight lux values and decided to do an experiment. The photo I took of the room was taken on a clear noon day in the US mountain west. I read that the CIE clear horizontal value for that time/location should be 25000 lux but that a vertical value should be .39 of the horizontal. So I removed the window assembly in the 3D model, turned the sky model to CIE Clear and entered a value of 9750 (25000x.39) and the render very closely matched the photograph's light levels. Was this just luck or do these numbers represent reality? - Craig

[3ddesign]

I should also report that I set up the exterior building model to match a photo I took on the same clear day. My research told me that a CIE Clear value of 50000 lux was appropriate for exteriors. I set up the Vray camera with the same settings as my digital camera and changed the Sky Model to CIE Clear w/ a value of 50,000 and the lighting levels were very close. Maybe this is old news for everybody but it was new and exciting to me.

But back to the original question: Has anyone figured out how to get a Vray plane light to accurately represent real world daylighting conditions? Or does everybody just use trial and error to find what 'looks' right?

-Craig

[3ddesign]

OK, I think I'm getting somewhere. Our lighting engineer measured the light at the window at 295 footcandles which equals 3175 lux. When I enter 3175 lm/m?/sr it renders correctly. This value equals 147292 lm which is very close to the value 145000 lm I used previously so I guess I'm on the right track. And since using lm/m?/sr means the actual light power is dependent upon the size of the window I should be able to use it on every plane light I use in a window opening and it should render accurately (at least for bright daylight shots at noon at a given camera exposure). I'll do some experiments and get back. - Craig

[3ddesign]

In my limited testing it looks like 3000 lm/m?/sr is a good starting point for daylight accurate vray plane lights. I went to several past projects and plugged the value in for plane lights in windows ranging from 2 feet square to 10 x 30 windows and in all cases the value worked with vray camera settings that represent typical real world camera interior settings (i.e., f16, shutter 1/2, iso 100). The value of 3000 also balances well with IES lights and lights with known lumen values (such as the typical chandelier we put in many of our spaces with 75 40W bulbs for a Lumen value of 32000 and the Kurt Versen downlights we use that have 3200 lumens). Comparing the renders to photos of the real space shows that this value looks pretty good.

I'd like to know if anyone gets a different result.

-Craig

[bardo]

could you show us some images comparison?

I've understood your tests but I'd like to see the result that it seems very interesting.

To do the same analysis I have to get an instrument to measure the real lighting and at the moment I haven't it.

Thanks in advance

[3ddesign]

I knew someone would ask this. Unfortunately all my work is NDA. If I have time I'll try to use a common model like the Flipside room.

[snivlem]

Look forward to it 3d, seems to be an interesting experiment, but not sure what you exactly mean without the pics.

[JeffG2]

ping for reference.