How to use the "lights" section?

  • We've been discussing practical lighting for night time, and I wanted to get an idea of what effect a typical light bulb in an average room would have. But I had some trouble understanding how to interpret the f.luxometer readings in the "lights" section.

    F.luxometer gives a cutoff point of 10 melanopic lux (m-lux) for "just before bed" - I assume that's about an hour. So is a 25-watt bulb on the other side of the room ok? What about 60 watts?

    The first thing I noticed is that it's rather surprising to read that an iPhone torch is "42% as bright as daylight", and significantly brighter than a halogen lamp! Eventually I realized that the "% as bright as daylight" number isn't really about brightness, but is a measure of the number of minutes of phase shift as a percentage of the maximum 2 hours 39 minutes produced by 100,000 m-lux daylight - but which can also be produced by as little as 5000 m-lux. So maybe a different wording there would be less confusing.

    I also realized that, with the exception of the candle and moonlight, the lights are normalized to 100 standard (photopic) lux. That explains why there's no mention of wattage, and no "distance" sliders in the settings (unlike daylight, with does have a "distance" slider, which is also a little confusing!). So the readings in the "lights" section should be interpreted as, "if you were exposed to 100 lux of this type of light, it would have this much melanopic lux, and produce this much phase shift". I imagine it would take hundreds of iPhone torches to light up a room to that level - though shining it directly in your eyes would probably do it.

    But I still couldn't get any idea of how much light an actual bulb would make in a "real world" situation. Say I have an 800 lumen (60-watt equivalent) Cree 2700K bulb, is there some straightforward way to estimate how many m-lux it would produce in a typical room setting? I tried a couple of online calculators and different formulas, but the results didn't make much sense. I read that an average living room is anywhere from 50-300 lux. Maybe there are too many factors to calculate - light bouncing around the room and so on - and one would really need to use a light meter to measure it?

    Assuming I had a light meter, maybe something like this HDE LX-1010B Digital Luxmeter, could I just multiply its reading by the m-lux/lux ratio listed in f.luxometer, i.e. 0.45 for the Cree 2700, to get the actual m-lux value? It seems like that's a linear relationship.

    But what's the formula for calculating the phase shift? It seems to be non-linear, "using a best fit from several published studies..." I suppose we could just say, "keep it under 10 melanopic lux just before bed". So for example with typical "warm white" LED or incandescent lighting, it should be kept under a 22 lux (22 x ~0.46 = 10 m-lux) reading on a standard light meter at your seating position, late at night. Maybe 27 lux for a CFL. So that seems to be quite dim indeed, something like candlelight, or outdoors with minimal street lighting.

    That doesn't include light from your screen if you're working on the computer though. I assume that f.lux's recommended setting of 1900K takes into account that the screen only takes up a certain portion of your field of view. Maybe ambient environment lighting, even if it's quite dim, might push you over the limit?

    Am I on the right track? I don't have a light meter at the moment, is there anyone who could take a couple of example readings with a 25-watt or 60-watt bulb?

    It would be nice if there was a calculator where you could input the melanopic lux value and get the phase shift for different lengths of exposure, at different times of the day. Could you use a phone's camera to directly measure the melanopic lux, and figure it all out? Now that I think of it, knowing you guys, you're already working on it...

  • f.lux team

    Yes, that's exactly why we put the m-lux/lux number there. It is pretty easy to measure lux, and somewhat hard right now to measure the other stuff. It is very useful to have a cheap illuminance "lux" meter and just multiply for different sources.

    We did measure all those lights with an irradiance diffuser, but it wasn't fair to report the numbers that way, because viewing distance is entirely arbitrary, so that's why we scale to 100 lux. And for screens, we instead measure "radiance" and then scale it in software (we do really well at it), which is why we can have a distance slider.

    It is true that distance and brightness are perhaps the most important factors. As a rule of thumb, light intensity from a point falls off with 1/r^2, and "lux" is in fact "lumens per square meter" -- so a very simple approximation tells you that at one meter (~3 feet), lux=lumens, and at 3 meters, lux=lumens/9.

    Another factor, however, is that the use of illuminance/irradiance to measure what goes into the eyes is somewhat strange -- it has been done for years in the sleep research, but people who do it are holding the meter so it's basically flat on your forehead. The usual way of using a meter (flat on the table) predicts how bright a piece of paper will look when you put it on the table, but it's not an especially useful number for circadian work - vertical to horizontal illuminance can vary by a factor of 10x!

    Anyway, this still means that a "reading light" like you might have on your bedside table can be unexpectedly bright, because you sit right next to it. If you don't look directly at it, you're better off.

    "Dim" as in "doesn't significantly suppress melatonin" we think is around <3-5 m-lux. But light at these levels can still keep you awake, of course. Dim is not dark.

    "Saturating" might be 500 m-lux or more, and light therapy treatment uses brighter levels (10x brighter) than that. Then, because the response curve is logarithmic, the surprising thing is that the midpoint ("half maximal") is somewhere around 50-100 m-lux.

  • @herf said:

    "Saturating" might be 500 m-lux or more, and light therapy treatment uses brighter levels (10x brighter) than that. Then, because the response curve is logarithmic, the surprising thing is that the midpoint ("half maximal") is somewhere around 50-100 m-lux.

    (Oh cool, I copied it and it quoted-nice) So what is m-lux?

  • f.lux team

    It's the portion of the spectrum the melanopsin cells see, scaled so that melanopic lux = lux when you see illuminant E (about 5455K).

    So typically if you're seeing light >5500K the melanopic lux number will be higher than a lux meter will give, and the reverse below.

  • @herf Cool, there's so many factors for how lights affect sleep, it's almost unreal!

Log in to reply

Looks like your connection to f.lux forum was lost, please wait while we try to reconnect.