How accurate are exposure meters? Not very....

[bglick]

> unless you are doing copy work, "accuracy" may be a subjective determination.


The first step in proper exposure is to get an accurate light meter reading.... if your starting point is 1 stop off, your starting point is off... more errors after this, (processing, flashes, etc) has no bearing on the fact you are starting with inaccurate data.



> In the real world you simply get used to your meter.


Do you ever notice how grass always looks green, regardless whether the color temperature of the lighting is 3000K or 10,000K? If grass always looks green, and colors always look true (our visual system compensates for such, drawing on its reference banks), then how do you "get used" to your meter? How do you know when your meter will produce a certain error, if you don't know yourself the condition exists? Our eyes are the worst judge of both color temp and light intensity, as we have too many involuntary compensations occurring. I have shot under many situations where I swear the color temp is near 6k, and yet, its 9k, been fooled hundreds of times.


Trying to notice a 1 stop change in light is not easy for our visual system. Sunny 16 is great, if you have NOTHING else to work with....but I have been out mid day and swore Sunny 16 would apply, and yet, I have had readings anywhere from 14 to 17 EV. This is the value of meters.... (preferably accurate ones)



> But to call them unreliable for color readings is absurd.


I marvel you can say this, after seeing the test results above. Unless of course you don't believe them, then I would fully understand your position, otherwise, I don't get it. These meters have large variances at different color / EV ranges. Even the manufacturers would probably admit to this.



> Color tranny film is fussy stuff;


hence why I posted this, now somehow, you are supporting my efforts?



> we'd all go broke if our meters were unintelligently designed


Go Broke? why? How about..... our exposures will sometimes "not be as accurate as we would like"..... this seems like a more reasonable assertion from the data I provided, vs. your exaggerated claim of..... "we'd all go broke?" ???


Unintelligently designed? No way, modern meters are very intelligently designed. Heck these companies have had 80 years to perfect the meters. The problem arises, when you try to use a meter that is calibrated at only one color temp, then attempt to use it at "many" color temps. The manufacturer offers no assurances of their accuracies at "non-design" color temps. This is not stated anywhere in the sales brochure, but its the reality of these low cost exposure meters. (low cost, vs. more scientific grade equipment)

[Drew Wiley]

Color temp has little to do with it, unless you're clear out in left field at the extremes.
Color temp varies all the time unless you're under studio conditions - that's why they
invented programmable color temp meters in the first place - but we rarely use them
except in commercial photog because we like what we see; otherwise we just add a
cc filter appropriately. Some photocells have a fairly symmetrical curve of response.
You can either plot this (nobody does) or get used to it. Once you place one of these
into a camera things get complicated, since cameras nowadays try to do everything
for you. With view cameras we generally take handheld readings, so get to
understand the meter's sensitivity per se better, at least until you run into things like
flare and bellows extension. Color trans film has very little lattitude. I shoot the stuff
in 8x10 - can't afford to be guessing. Half-stop error, no way! But for one scene I
read some grass, next shot in the desert -no grass - I read a brick red rock. How
come both readings work? Because the meter is predictable, reliable. I know through
experience or testing that it's more sensitive to certain colors than others. No
difference with an enlargement meter except that you can keep the color temp constant (with certain kinds of light sources). Color temp is a different variable than
color sensitivity. At a certain point these two variables transect; but most modern
meters aren't that fussy about a thousand degrees K here or there. Reading a bright
red versus a saturated blue, however, might drive one nuts, unless you already understand the bias. Spotmeters behave different than indicent meters, and so forth.
But there is NO ABSOLUTE REFERENCE POINT. There is no such thing as a "true"
reading because, in the field at least, conditions are always changing, and different
people use meters in a different manner. Light meters are not spectrophotometers!
Stick with the brand which works for you and calibrate accordingly.

[BrianShaw]

Brian, if you describe your K factor more clearly, I will try to respond.... and there is no industry standard for photo meters...

Sure. I can do that. It seems to show up in lots of conversations and publications regarding meter calibration/accuracy.

http://www.largeformatphotography.in...-meter-cal.pdf
see page 7

http://dpanswers.com/tech_kfactor.html
see Section 2.

http://photo.net/medium-format-photography-forum/007XcE
see Severi Salminen , Feb 27, 2004; 06:08 p.m.

Also see:
http://en.wikipedia.org/wiki/APEX_system

I know that my Weston meter states on the plate K=1 but I haven't seen such data about other meters, nor do I know if differences make any appreciable difference.

Re: photo meter standards... I know the photometer and goniophotometers I used way back in what is almost a prior life were calibrated in accordance with a NIST standard. I believe had something with the reflectance of calcium carbonate. I recall having a "calibration standard" that looked like white rock that we calibrated our instruments with prior to doing studies. Other than that, my memory is vague. I can still recall the calibration sticker on the PhotoResearch light meter that assured traceability, and I still remember writing big checks for that calibration. How that relates to photgraphic light meters I can't say, hence the question.

[bglick]

> Color temp has little to do with it,


HUH? Color temp has everything to do with it.... did you see the results at near daylight vs. 9k, as I explained in my previous post above?




> Color temp varies all the time unless you're under studio conditions - that's why they
invented programmable color temp meters in the first place - but we rarely use them
except in commercial photog because we like what we see;


It doesn't matter what we see, it matters what the film sees....




> With view cameras we generally take handheld readings, so get to
understand the meter's sensitivity per se better,


Can you explain how one gets to understand their meters sensitives? Our eyes are terrible at estimating EV and color temp readings...did you read my previous post?




> but most modern meters aren't that fussy about a thousand degrees K here or there.


did you see my findings in the original post ?? How do you draw this conclusion?? Based on what? At least I provided some test data to demonstrate otherwise...





> Reading a bright red versus a saturated blue, however, might drive one nuts, unless you already understand the bias.


What if you are reading the light source itself? How do you understand the meters bias, when your visual system can not read color temp?




> But there is NO ABSOLUTE REFERENCE POINT. There is no such thing as a "true" reading because, in the field at least, conditions are always changing,


There is a true light level reading at a given point in time - the moment you press the exposure meter button! IF there was no reference point, than explain how a meter is calibrated? Of course there is standards..argggg... Of course conditions are always changing, that's why you keep taking new readings?


sheeesh, I am ready to surrender.... looking for white flag....

[BrianShaw]

bglick, here is what I was trying to recall re: standards. I thought I saw them before and just found them again. Do you know anything about them; do they impact your discussion?

ANSI PH3.49-1971. American National Standard for general-purpose photographic exposure meters (photoelectric type). New York: American National Standards Institute. After several revisions, this standard was withdrawn in favor of ISO 2720:1974.


ASA PH2.12-1961. American Standard, General-Purpose Photographic Exposure Meters (photoelectric type). New York: American Standards Association. Superseded by ANSI PH3.49-1971.

[bglick]

Hi Brian, thanks for the clarification..... since the Adams days, the K factor term has been more commonly called "meter reflectance factor"... but I could be wrong.... I rarely see the K factor term used much these days...regardless, it's 6 of one, half dozen of the other....

First, I hope it was obvious to everyone, I took incident readings of the light source itself, the sun. I fired all 5 meters within a few seconds, then recorded the values after they were fired to avoid any delays in taking the readings. To be clear..... I WOULD HAVE NEVER POSTED MY FINDINGS IF THESE WERE SPOT METERS READINGS !


The reason is obvious, as stated by the some of the nice links you provided. Reflectance readings vary between meters, as each maker seems to have secretive reflectance values. Many years ago, I once tried to match up my Pentax spot and my Sekonic 778 Spot, they were more than a stop apart at all times. I researched this as hard as I could, and could not come up with the exact reflectance values each meter used. Without this information confirmed, you can not ascertain if a meter is not calibrated, or its reading is accurate, but it displays EV information based on a different "assumed" reflectance value of the target. This is why, I only use spot meters when in a jam... whenever possible, I always read the light source with incident readings...from there I can deduct what I need to know from the shadows through the spot meter (looking for delta EV values), disregarding the nominal readings.


As for the standards of the light meters, no, I am not familiar with the ANSI references you provided. For incident readings, a photographic light meter such as my 608 correlates very well with my foot-candle meters I use for my day job. This assumes, the color temp of the light source is ~ 5000K for the Sekonic. Photographic light intensity units, EV's, are directly related to Lux and Footcandles.


And to be clear, I am not trying to prove the accuracy of the meters in this test. They all proved relatively consistent when at the design color temp. The purpose was to determine how large the errors will be between meters under varying color temp conditions. It is quite obvious, they all react very differently. More than I would have liked to have seen. Again, at least for color chrome film....negative film can absorb these type errors. I strongly believe that Nikon shifted to a true color compensated exposure system as most pros shot chrome film, and they saw this as a valuable feature to better nail exposure on chrome film in all lighting conditions.


If I had the time and energy, and a light which I could easily control the color temp and light intensity (possible with filters).... i would run a film test with my favorite chrome films.... I would record the color temp, incident reading, Lux reading and then bracket the grey step chart.... then analyze the results, and create a more reliable protocol to correct for the incident meters spectral response. Or, it's possible the Color temp Lux values by themselves would be close enough to accurate, where it would serve as my new incident meter.... only testing would provide these answers...

Since my exposures are pretty damn good 90% of the time, its a lot of work for that 10% :-)

[Scott Knowles]

Scott, even if you eliminate the Gossen 3 as the "standard", look at the errors between the meters themselves, that was the point. If they all had very similar errors vs. the Gossen 3, then I would agree, but its obvious, the errors against each other are extreme IMO. This demonstrates at different EV levels and different color temps, the different makers of meters have errors that are not consistent with each other.

A more scientific test would be nice.... but if you take two new different makers meters, and they read 1 stop different than each other...... the evidence is obvious...they both can't be accurate. Even if the error falls right in the middle, thats .5 stops off per meter.... that's the best case scenario. That's the point.... no need to go to the next level unless you want to track each of the errors vs. a standard. This test identified the meters have much larger errors than most think, period.

But you also assume the other meters are representative of that brand and model, and they're really just a sample study, and not much more. I suspect there's considerable variation within each brand and model because it depends on the age and use of each one, and if/when they were checked and calibrated.

If all you're saying is they're all different, ok, point proven, they're all different including the Gossen. It still doesn't establish which if any are accurate and over what light conditions. That's where the reference standard is important.

In the end, it's an interesting study but only so far as the light meters you have and used in it, just a sample comparative study, but not one to draw conclusions outside of the obvious which you've proven, your light meters are all different.

Every year I take all (13) my old Minolta manual focus cameras out, add new batteries and do a similar study, comparing light meter readings with Sekonic L-308b and L-358 light meters. I'm only looking for anamolies, and almost every year they're all within 1/3-1/2 f-stop of each other and the Sekonics. That's the whole value of the test, find any errant cameras.

For a short term study at work (USGS) we planned to install some small (then $100) pyranometers (measuring insolation) at gages around Western Washington. The first thing we did was run all of them against a NBS standard pyranometer to establish the range of each and the range of all of them so we could understand the data.

Otherwise any data would be meaningless outside of a ballpark value. That's what you have with your study, a sample ballpark study and of little value outside itself for you and your light meters. Sorry, but that's what it seems to me.

[bglick]

> If all you're saying is they're all different, ok, point proven, they're all different including the Gossen. It still doesn't establish which if any are accurate and over what light conditions.


BINGO! That's all I have been saying, about 8x now :-)




> just a sample comparative study, but not one to draw conclusions outside of the obvious which you've proven, your light meters are all different.


Well, not only are they different, this demonstrates they all can NOT be accurate, that is the KEY point. yes, its a can of worms, but the purpose of this post was to show the can, the dirt, and the worms exist :-)



> That's what you have with your study, a sample ballpark study and of little value outside itself or you and your light meters.


Of little value? This simple test demonstrated that any one of these meters can be .8 - 1.5 stops off, under certain color temps / EV ranges. I have clearly identified this, something that was not known previously. Of course, I don't know which meters are accurate at all these different readings, that would require a more extensive testing.


But most studies start with the path of least resistance first......i.e. start with simplified tests to demonstrate a problem actually exists before going to more in-depth studies. This simple test demonstrated that exact point, so I would strongly disagree with your assertion there was little or no value to this finding. It was well worth the 15 minutes. But the 2 hours wasted on this thread I am starting to question my sanity on....

[Ivan J. Eberle]

No doubt that some motivated geek could've gutted a beater F5 to mount the meter and display in an Altoids tin for you in the same amount of time.

[bglick]

Ivan, that is actually a good idea..... potentially beater F5's may have a 2nd life after film :-) hmmmm........