Linear or logarithmic?

As for multi stage lights (at least have three stages, low - med - high), which do you prefer more, linear or logarithmic? If possible, please also explain the reason.

Thanks.

I voted linear. I don't believe that our eyes just notice brightness changes logarithmatically. I have getting old eyes and I can notice 50% or less changes in brightness, particularly when lux is the key issue. Those fence posts are just brighter when the difference is 25% using a light that is set up for optimal lux. I usually try out my new lights or modded lights before measuring them with my lightmeter for lux and output (bounce) to see if there is a subjective difference compared to my other lights, or what I can recall (subjective) from the light prior it being modded to higher lux/lumen. I know, the old pros will argue this to the end.

Bill

Just noticed voting. I am in the minority.

Bill

I voted logarithmic. I can't argue with the people who voted linear. Sometimes I like my light levels more spread out, and other times I like my light levels closer together. I have an HDS EDC so I can set the levels how I want them.

The idea of logarithmic sensitivity is often misunderstood on CPF.

The human eye is sensitive to small changes in brightness, much smaller than factors of 2, 10, etc. The actual number is irrelevant to whether the appropriate scale is logarithmic or not. The question is how the eye and brain perceive successive changes in brightness.

Suppose that you are designing a light with a maximum output of 100 lumens, and you want to have 6 brightness levels. To simplify, you have two choices regarding how to set up the scaling:

Scale A (linear): 100, 80.5, 61, 41.5, 22, 2.5
Scale B (logarithmic): 100, 50, 25, 12.5, 6.25, 3.125

If you build a multi-level light using scale A, you will find that the differences in perceived brightness at the high end seem rather small. You will notice them, but the degree of dimming won't seem very useful. At the low end, the differences will be very noticeable, and the jumps at the very lowest end may seem too large. (In some situations, you might wish that there was a 5 lumen or 10 lumen level.) But all of the intervals in this scale are 19.5 lumens.

Scale B is something closer to what the Surefire U2 uses. Here, the brightness between successive levels changes by a factor of 2. With this scale, you will find that the steps in brightness between adjacent steps are noticeable and useful, at both the high and low ends of the scale.

I should point out that there is no single "correct" base for a logarithmic scale, since they are all inter-convertible through scaling by a constant factor. If you had more levels and a smaller % change between steps, the scale would still be logarithmic.

Logarithmic steps will appear 'even' to the eye, linear steps will appear 'uneven'. If you use a linear volume control on your stereo, it will not be very nice to adjust, logarithmic adjustments are more natural.

-- Alan B

If you are using the flashlight in the daytime, going from full brightness to total darkness and back again before your eyes can fully adjust, the lowest levels will be useless because your eyes are adapted to bright light. In that case, you would have two useful levels for daytime use with the linear method instead of just one with the logarithmic.

I prefer being able to set my own levels on my HDS and not being restricted to just logarithmic or linear scales. I wish everybody could try out a fully adjustable light and see what it is like to always get exactly as much light as they choose.

afahmic said:

As for multi stage lights (at least have three stages, low - med - high), which do you prefer more, linear or logarithmic? If possible, please also explain the reason.

Thanks.

Click to expand...

I would prefer a logarithmic scaling but with two or three more levels.

Something like this (in %) (numbers are rounded):

3; 6; 13; 25; 50; 100 (factor of 2)

or

1; 3; 6; 16; 40; 100 (factor of 2.5 )

I firmly believe that the brigther the light, the more levels are needed. With single LED flashlights now reaching 200 lumens and up, five or six levels are not exagerated. It all depends on how those levels are implemented or on the U.I. (user interface) if you wish.

As an example, the Fenixes LxD and PxD would be better lights if they had a lower low instead of the poor SOS feature.

Calina said:

As an example, the Fenixes LxD and PxD would be better lights if they had a lower low instead of the poor SOS feature.

Click to expand...

+1. Hope they see this poll.

afahmic said:

+1. Hope they see this poll.

Click to expand...

+2 man +2!

asdalton:
Scale A (linear): 100, 80.5, 61, 41.5, 22, 2.5
Scale B (logarithmic): 100, 50, 25, 12.5, 6.25, 3.125

Click to expand...

one small clarification - the scale described as "logarithmic" is actually a geometric scale (say, as in the example quoted above, it doubles every step).

what is going on is that the eye perceives light logarithmically, so you get a geometric increase in light intensity by design, to counterbalance that and result in a perceived increase in light that is linear!

When I first got my P1D CE, I thought something was wrong with it, because the change from medium to high wasn't as dramatic as I thought it should be. That was my only gripe with an otherwise great light. Same with the P2D I now carry. I'd prefer smaller intervals at the low end, and bigger intervals at the high end.

I recently received my Liteflux LF2, and once again, the fact that the user adjustable mode moves linearly is my only beef. The light level seems to change really fast at the low end, but toward the high end it is like watching a sunrise. I have to wait for the blink to know when I've reached maximum. Other than that, it's a fantastic light.

Yeah, I know, I basically just repeated everything asdalton said. I guess that's my long-winded way of saying I think he's right.

Oh yeah, it should be obvious, but I voted for logarithmic, or geometric, or whatever we decide the word is.

Logarithmic for me too. At the high-end, precise control is not that important. However, at the low end, where too much light could be critical, you want that extra control.

I like the linear ranges in asdalton's example. They are less than 100% jumps in lumen output. I must admit though that my Chameleon does range in 100% jumps from level 2 to level 5 in each CT. CT1 = 350mA's to led with four log jumps level 2-5(350mA's to led, CT2 = 500mA's with four log jumps level 2-5(500mA's to led), etc up to 1.2 amps to led (CT5). I have no real issues with Chameleon, but linear would be good, though there would have to be more levels in eact CT to reach the max lumens for each CT.

Bill

In my opinion the HDS EDC was the gold standard for multi level lights. It had adjustable levels, but it came with some well thought out defaults.

The default lumen levels were from lowest to highest:

0.33 lumens
2.6 lumens
10 lumens

The highest level was 42 lumens, 60 lumens, or 85 lumens depending on the model you had.

What really impresses me is that they kept the lower three levels the same for all three lights. If they had stuck with a pure logarithmic scheme they would have raised the lumens on the lower three levels on the U60 and U85. I would not have liked that. The lower three levels were just right the way they were. Why change them just to keep the levels totally logarithmic?

Calina said:

As an example, the Fenixes LxD and PxD would be better lights if they had a lower low instead of the poor SOS feature.

Click to expand...

Yes. I hope Fenix notice this request.
Not a fancy "programmable" interface, just a low of less than 3 lumens.

I don't know whether the P3D is on medium or high until i;'ve clicked through them and got to the SOS. :laughing:

Art Vandelay said:

In my opinion the HDS EDC was the gold standard for multi level lights. It had adjustable levels, but it came with some well thought out defaults.

The default lumen levels were from lowest to highest:

0.33 lumens
2.6 lumens
10 lumens

The highest level was 42 lumens, 60 lumens, or 85 lumens depending on the model you had.

What really impresses me is that they kept the lower three levels the same for all three lights. If they had stuck with a pure logarithmic scheme they would have raised the lumens on the lower three levels on the U60 and U85. I would not have liked that. The lower three levels were just right the way they were. Why change them just to keep the levels totally logarithmic?

Click to expand...

for my uses those levels are not quite right; almost, but not quite:

so my second acquired 120P is

0.47lm (pri)
10lm (sec)
30lm (min) [sic]
85lm (max)

and the first acquired 120P

15lm (pri)
30lm (sec)
60lm (min) [sic]
120lm (max)


i find the second one i acquired more useful overall, and the first one better for use on the trail at night where, for me, less than 15lm while on the move is always insufficient output.

i EDC the second acquired 120P and use it daily (usually on the 0.47lm o.p. level)

Live on earth is usually best described with e-functions (logarithmic).
My eyes notice brightness in a logarithmic way,
so what the heck should I do with a linear torch?

I prefer a common-sense, usability approach rather than either linear or logarithmic. For example, if I were going to create a three-stage light, I would:

Set the high stage to 100% (duh)

Set the low stage by carrying some lights for a day and seeing what's most useful. Whatever that worked out to be, I'd set to low. Most people would probably be happy with about a lumen.

Repeat the above step for the middle stage, doing bigger jobs.

I think the biggest problem with bad stage setting is a ridiculous adherence to reason instead of usability. A foolish consistency, and all that. For a number of lights, I try to use the levels and wonder if anyone at the manufacturer ever actually carried the light for a week. I don't want these guys arguing about linear or algorithmic or percentages, I want them arguing about what level of light seems to be useful for broad classes of jobs, then deriving the percentages from that without regard to mathematical function, at least for lights with a small number of levels.

I voted log for if it's not one level. I prefer one level, though.