Think of the room you’re doing your ceiling bounce readings in as a big lightbox because in effect it is, and I’ll call it that to keep things simple. A lux is one lumen per square meter but what we really want to know is, how do lux as measured in your particular lightbox translate into lumens? That’s what the regression formula is for. We need two columns of data to come up with it: 1) lux as measured in the lightbox (your ceiling bounce readings), and 2) actual lumens. Ideally, we’d get the latter from calibrated integrating sphere readings of each light but not everyone has a calibrated integrating sphere handy and if we did we wouldn’t need to do ceiling bounce readings, so we’ll use manufacturer reported lumens that we believe are credible as a proxy. What the regression formula does is quantify the relationship between lux, as measured by your readings, and lumens which can then be applied to other lights. In other words, you develop the formula with maybe ten or so lights (using your lux readings versus manufacturer reported lumens) and then use it to estimate the lumens of other, different lights by plugging the ceiling bounce readings of those lights into the regression formula.
Ya see that is the problem. (Finding the second column of actual lumens)
We have to remove all ****fire type of lights because they all have extremely exaggerated lumens. That leaves us with Thrunite basically. Who knows maybe they are exaggerated as well.
I'm not sure where you can find accurate lumen numbers for these lights.
Appreciate your detailed answers to my questions. Looking to put a good size flood on my shelf soon and your information & comments were helpful.
The problem is that I only have a few lights that I feel confident in the manufacturer lumens. (from ThruNite so six flashlights)
I also think your data would be much closer to 1 if you used an exponential regression line. A polynomial one would get you ever closer to 1 but it would be unrealistic.
I've thought about doing what you are doing before but I never had lights from a manufacturer that didn't exaggerate the lumens.
The reason I mentioned exponential is because your numbers don't look linear. They suggest an upward curve. I don't see why you say it is terrible because I bet it gives you a better R^2 value. I agree a linear line is ideal because that "should be" the relationship between lux and lumens.
Ya I know polynomial isn't realistic that's why I said it would be "unrealistic". You can make almost any set of data R^2=1 with a polynomial regression line which isn't a good thing when there is a limited amount of data points.
I don't claim to know everything about this. I am just questioning why you trust manufacturer lumens. I know we have to start somewhere but why bash me for giving approximate lumens?
I know you are trying to help and your light box and lux reading knowledge greatly surpasses mine. But I'm not exactly "just starting out"... not sure how you gathered that? I live and breathe flashlights and run my own flashlight business. (currently expanding online) I am currently seeking my BA in Marketing because I would like to become Product Manger/Sales Manager/etc. for a flashlight company one day.
The reason I posted about reading my lux meter is because the numbers were so off that I assumed I must have been reading it wrong. I was reading it right. My numbers were off because it isn't exactly spelled out that everyone did there lux readings at more than 1m and then converted back. I think anyone doing lux readings the first time would assume the same thing.
I just love flashlights and am proud to have BLF to talk about these amazing gems.
This week I will try to get a regression line figured out that I feel confident and will update my estimated lumens readings. I think they are close but I would like to get them closer. Thanks luvlites for sparking my interest in getting my numbers closer perfect.