I think this is an interesting idea and definitely worth pursuing, definitely happy to contribute whatever mathematics I know.
I think that the current ANSI standards are still quite good, though it offloads the responsibility of interpretation to the user (which may or may not be a bad thing). Ironically, part of why it’s good is that it does not take into account human perception, which (as we’ve seen plenty of times in discussions about the lumen as a unit) tends to introduce ambiguities and make things mathematically ill-defined. One may also argue that factors such as the human eye’s ability to see in the distance varies from to person to person, and is not intrinsic to the light itself.
Designing a new metric is often a difficult pursuit: among other things, one needs to ensure that it satisfies various desirable properties, and just to list a few:
Is it well-defined? Given the same light and different ways to measure the proposed quantity, would they always return the same result? Under identical conditions, does every measurement yield the same result?
Does it actually measure what we want it to, and is it self-consistent?
Is it the “only” sensible definition (up to scaling) that captures the notion of interest, or is it arbitrary? That is, do there exist necessary/natural assumptions that force this definition to be the only candidate?
Satisfying all of them is no easy feat, especially the last one.
But it already does for colours (as it should). All the luminous units use ‘luminosity function’ - a purely empirical perception-based construct.
The only reason mathematicians don’t riot against it is that it’s been packaged as an infallible law of nature expressed as an elegant curve, conveniently unchanged since 1924.
Good point, I should have specified that it only takes into account human perception via the use of lux as a unit, but does not assume anything about, for example, how perceived brightness scales with true brightness given the same fixed spectrum.
I’m really truly not a huge fan of the luminosity function. The very simple fact that the human eye can white-balance to different CCTs implies that the luminosity function is ill-defined, in particular violating
since exposure to prior stimulus changes the relative weighing of wavelengths for the same source.
Can you please explain what you mean by a “5 number summary”. Is it simply 5 flashlight characteristics you would like manufacturers to include with their product descriptions?
Then, please describe each of the 5 items in layman’s terms. Thank you.
A small number of values that best describes the light
min lm is the lowest level. You may not be happy with the light if it’s too high.
max lm is what everybody wants to know (kind of like megapixels in cameras). 30 s is a reasonable cutoff point for unsteady turbos
ANSI distance is up to the point where the beam illumination is 0.25 lux. Halving it implies 1 lux illumination at the target and is what people are doing anyway to make it more realistic. Calculations also simplify as it is just a square root of max candelas.
√(cd/lm) is a nice and intuitive index for beam spotiness, independent of lumens.
lm×hr is a comparative measure of how much light total one gets from a light+battery. If you know that, you can estimate runtimes if you can estimate the output
A small number of values that best describes the light Are you talking about package & website labeling?
min lm is the lowest level. You may not be happy with the light if it’s too high. Which is usually included with every flashlight.
max lm is what everybody wants to know (kind of like megapixels in cameras). 30 s is a reasonable cutoff point for unsteady turbos Also usually included.
ANSI distance is up to the point where the beam illumination is 0.25 lux. Halving it implies 1 lux illumination at the target and is what people are doing anyway to make it more realistic. Calculations also simplify as it is just a square root of max candelas.
**Including the manufacturer’s calculation for throw based on the FL1 standard is useless for anything other than bragging rights, which is why I hope we as a community can come up with a more valuable standard based on real world usage by those that would like to know how the flashlight they’re considering will meet their real world needs. Such would be particularly useful for consumers if included within professional reviews! **
√(cd/lm) is a nice and intuitive index for beam spotiness, independent of lumens. Layman’s terms please, lol.
lm×hr is a comparative measure of how much light total one gets from a light+battery. If you know that, you can estimate runtimes if you can estimate the output So you want runtime included with the manufacturer’s description?
With all due respect and unless I’m misunderstanding your posts, I think you’re getting a bit too far off the topic of quantifying beam shape, and the added topic of defining meaningful beam throw.
@QReciprocity42 questions for you (and of course anyone else who likes to chime in please):
My Nightwatch LW55 tests 8200lm on Turbo. To find out what the LED itself generates (such as the numbers published by Koef), I added 10% for optical loss:
1. Is that 10% reasonable for modern flashlights with AR lens? Without AR I would tack on 3-5% (my findings with Wurkkos TS30S and TS30S Pro), is that about right? 2. Circuit loss is about another 10% correct? But I should NOT add it here? 3. I’ve read Orange Peel causes 2-4% lumen loss vs Smooth Reflector, about right? 4. This LW55’s 5 cm reflector should have less optical loss than eg my Convoy T8’s ~2.5 cm? Just wondering how significant this difference is? 0-10%? 10-20%? TIA
Going from bare LED to fully assembled reflector-based light with AR lens, this test suggests that a loss of 15% is a good estimate. But this will vary due to reflector quality, AR quality, etc.
As far as I know there are no reliable estimates on circuit loss–do you mean loss by resistive elements (wires, contacts, etc.) or do you also account for driver efficiency?
I have not seen any evidence that OP reflectors are more lossy than smooth.
For a reflector, loss is independent of reflector size. A reflector only incurs absorption losses (no scattering losses like a TIR), which is not dependent on size/area, just on the albedo of the surface, which can be assumed to not depend on size.
Thanks very helpful. For the above question I was asking more about “how to do it” rather than the science behind the loss. Namely, how to estimate raw LED output from flashlight output. I use the formula: Raw LED output = Flashlight output + optical loss, no circuit loss added.
Assume optical loss is 15%, and circuit loss is 10%, and my flashlight’s output is 8000 lm at max. To estimate the bare LED output (as in Koef’s test), my calculation is 8000 + 1200 (15%)=9200 lm.
If you want to estimate electrical losses, you need to measure V and A at the emitter, V and A at the battery, and compute the ratio of power out to power in.
Unclear where your 1200 came from, and the percentage calculations are not quite correct. In particular, the inverse of -15% is not +15% but +17.65 (do you see why?). A better estimate is
That could be the source of a potential misunderstanding, thanks for pointing it out. But even lux reduction is heavily dependent on the extent of OP texture and the LES of the emitter, both of which varying wildly across lights. So even a ballpark estimate that works universally cannot exist.
As OP texture increases (or as LES decreases), the lux loss can go from 0 (smooth reflector) to arbitrarily close to 100%.
Remember back when we could just generalize LOP, MOP, and HOP? I guess the scattering effect would be similar to a high transmittance diffuser film, except with much less transmittance losses in terms of lumens. In terms of beam shaping, would beam smoothing techniques be applicable to this thread?
Sorry, I couldn’t immediately recognize the acronyms LOP, MOP, HOP, could you remind me?
I think the effect of going from smooth to OP reflector is comparable to applying diffusion film, even though the principles are somewhat different. I bet many ideas about beam smoothing can be modified/generalized to this context, if not applied directly.
start that other thread already!
