Richard Korzekwa

Researcher at AI Impacts.


Why indoor lighting is hard to get right and how to fix it

Another thing that might be worth considering is that if you want to estimate your body's response to light, a camera-based illuminance meter with a very wide field of view and some amount of clipping is probably better than a standard lux meter with a diffuser, since it will account for whether the light is spread over your full field of view or not.

Why indoor lighting is hard to get right and how to fix it

This is a good question. I have not looked, and the closest thing I have done is to measure the color temperature and intensity according to my camera for direct sunlight just before sunset (seems to be about 2700K, 1900lx).

I would think that you could get a decent estimate using SMARTS if you adapted the input file to a ray going through a lot more atmosphere, but I'm not sure. I haven't looked at the code and it might do something like make approximations that only work for short path lengths or low optical densities or something.

Why indoor lighting is hard to get right and how to fix it

I just compared my phone (Pixel 3a) to my Uceri meter, using the app that you link to here.

It seems okayish. The main issue with it is that the camera has a limited field of view, so it does not capture light coming from every direction. With the lux meter and the phone both pointed directly at the lamp near my desk, the phone reads about 35% higher than the lux meter, and when they're both pointed at an angle that barely excludes the lamp from the field of view, the phone reads 45% lower than the lux meter. The reading on the lux meter only changed by 4% between these two angles.

I'd say if you're okay with getting within a factor of two, the phone app is fine. The phone and the lux meter seem to agree the best when the lamp is just barely inside the field of view of the camera.

Automated intelligence is not AI

It seems to me that there is an important distinction here between "the thing that replaces human cognitive labor" and "the thing that automates human cognitive labor". For example, a toaster might eliminate the need for human cognitive labor or replace it with a non-cognitive task that accomplishes the same goal, but it does not automate that labor. A machine that automates the cognitive labor normally involved in toasting bread over a fire would need to make decisions about where to hold the bread, when to turn it, and when it is finished toasting, by having access to information about how the bread is doing, how hot the fire is, etc. Or maybe people are using these phrases differently than I am expecting?

Why indoor lighting is hard to get right and how to fix it

Visors I've seen do not illuminate all that much of your field of view. Plus I'd prefer not to have to wear a thing all the time. But maybe there are better products like this now? Or were you thinking of building something?

Why indoor lighting is hard to get right and how to fix it

This is really cool!

I thought about how to do this a while back, but I couldn't think of a good way to do it without it being huge, expensive, or a fire hazard. At the time, I do not think there were bright enough LEDs to really make this work, and I simply hadn't thought of a good way to solve the Rayleigh scattering problem.

I also hadn't thought of the trick with a bunch of boxes with separate LEDs. It seems like you could make a bunch of smaller versions of his larger one (that is, using smaller LEDs with reflective collimators and a layer of colloid for scattering), and get a similar effect without needing such a large volume. I may try this out.

Why indoor lighting is hard to get right and how to fix it

Thanks! The last time I was shopping for LED strips, they just weren't bright enough to really light up a room, but it looks like that has changed. I'll add something about this in the post.

BTW, what I was using LED strips for was "UFO lighting" around the base of my bed. The relatively dim, low CCT strips were attached around the bottom edge of my bed frame, with the LEDs pointing toward the floor. The scattered light was just enough to read by, and having the light mainly on the floor was good for not tripping over things if I had to get up in the middle of the night.

Why indoor lighting is hard to get right and how to fix it

The biggest reasons why I do not try to do this myself are:

  1. I haven't done enough research to know what a good dose is in terms of wavelength, intensity, and time
  2. It seems hard to build a source that gets the dose right, without basically building a tanning bed
  3. I think it's hard to know what dose you're actually getting

It's also possible that I am unreasonably worried about shining artificial UV light on myself.

Assuming you have an answer for 1 (and that paper looks promising for having a good answer to this), you need to build a source that illuminates your skin in some reasonably even way. This can be hard, since most light sources radiate over a wide angle, so that you get 1/r^2 drop off in intensity. For example, just now I used my lux meter to measure the intensity of the light near my desk on my forehead and my stomach, and they varied by almost a factor of two. One potentially pretty neat way to fix this would be to use some kind of large collimator like the one in that video that Robert Miles mentioned. The intensity will drop off much more slowly with distance from the source, which should make it easier to get a predictable dose, plus our intuitions about dose from the sun will will still sort of work (for example "The part of my skin that is farther from the sun but is normal to the sun's rays is getting more"). If I did build a big sunlight simulator like that, I would be tempted to add in a little UV.

I recently learned that a common way to measure UV dose is to paint something with a white pigment that absorbs UV and reflects visible/IR light, and see how much it heats up when illuminated. I'm not sure how well this would work when the dose rate is less than 10 mJ/cm^2/minute as in that paper. You probably would just want to use a UV index meter, like they did in the Nature paper you linked.

What would be good is to have some kind of inexpensive cumulative dosimeters that you could place on yourself and around in the area where the UV is so that you can check that you're not inadvertently getting way more or way less than you want. A quick Google search for "UVB dosimeter" looks like there are options, but I have not looked at any of them enough to know if they're any good.

What risks concern you which don't seem to have been seriously considered by the community?

I do not mean technological development productivity, I mean economic productivity (how much stuff we're making, how many services we're providing).

What risks concern you which don't seem to have been seriously considered by the community?

It is plausible to me that this would be fatal to our civilization, in the long run. Eventually we need to stop being biological humans living on the surface of Earth. It is not clear to me that we can move past that without much higher productivity than present day US.

Load More