But if you take six, and all of them are different by more than the error margin in their specs and you can rule out that at least five out of six were defective, there must have been some flaw in the test methodologies.Originally Posted by bglick
But if you take six, and all of them are different by more than the error margin in their specs and you can rule out that at least five out of six were defective, there must have been some flaw in the test methodologies.
Hi bglick,
why are some numbers in brackets?
sanchi
Intersting exercise, but a couple of observations if I may;
I don't thing it's important that my meter is ultra-accurate, regardless of the standard measured against. It just needs to be consistant.
I'm happy to work to the nearest 1/2 stop, and make adjustments at the processing stage having seen the first sheet. There are too many variables throughout the image making process to be any more accurate.
bglick has only done a comparison test and not dealt with absolute accuracy. I find the results interesting but not unexpected. As has been pointed out, additional exposure variability results from film response and shutter speed variability.
How did bglick increment the color temperature steps - black body, tungsten filament?
Read by one of the color meters I presume.
My take would be sort of like Gittings comment. I don't care how accurate my meters are as long as they are consistent. Zone tests determine the densities of chromes or negs as a function of meter readings. Grey card exposure gives a chrome at logD 0.9 and negative at logD 0.9 for normal development. Of course experimental stuff is apt to be logD all over the place. I just check meter calibration periodically using bright sun on clear day and grey card to read EV 16.3 and use a paste on arrow on the Pentax meters.
Nate Potter, Harrington ME.
> But if you take six, and all of them are different by more than the error margin in their specs and you can rule out that at least five out of six were defective, there must have been some flaw in the test methodologies.
The accuracy of the Sekonic meters are +/- .1 EV. You mention "flaw in the test methodology". You are missing the point. If a meter is calibrated at one color temp, it should be relatively accurate at that color temp. Which I would suggest they are. But how do they perform at different color temps? No specifications on this....of course not.... cause if the makers provided specifications on how far the meters are off at different color temps, do you think this would help sales? The avg. photog knows very little about the nitty gritty of this stuff, and meter makers want to keep it that way.... hence why i shared this info, which I am starting to regret....
> why are some numbers in brackets?
brackets always indicate negative values. <.5 > indicates the meter read .5 stops BELOW what the Gossen 3 read. From that, you can see the errors between the meters.....from my post above...
the meters themselves vary tremendously from each other... often a full stop variance....you can read across the lines, -4 to +6 = 1.0 stop
variance.
> brackets always indicate negative values...<
Ok,thanks.
I once tested a broncolor FCM and a Gossen lightmeter.
There was only a difference of 0.1 stops.
And I compared 25 - 30 times under different lighting conditions.
sanchi
Bill, as mentioned by a previous poster, the F5 that I also used for years had an incredibly accurate color metering system - MUCH, MUCH superior to the 5 and 6 series Sekonics that I use.
Shooting E6 with the F5 was an enjoyable and productive experience.
Perhaps I am not as stupid as I thought - my meters are partially to blame.
Anyway, is there a particular color meter that is considered state of the art?
Thanks for your research and posts!
>How did bglick increment the color temperature steps
As mentioned, color temp came from the Gossen 3 meter.
> I don't care how accurate my meters are as long as they are consistent.
Most meters are consistent..... they will be consistently inaccurate under different color temps. An exhaustive film test under controlled lighting would be an ideal test to determine not only the exposure error, but how the film reacts at these color temperatures. Considering most of the list found my 15 minute test absurd / uncalled for, I can't fathom how the list would react to the sharing of information for an exhaustive film test and the cost of using such expensive controlled lighting and metering equipment. But yes, this would be ideal.
My guess is though, (please note the word "guess"), using a color meter as an exposure meter (assuming the color meter reads Lux) will get you close enough to an accurate exposure reading, at any color temp. This assumes films responses does not alter such. This would explain the excellent success the NIkon had with its built in color meter for exposure....
Sekonic has a new color meter.... it has a film and digital setting for reading color temp, as both react differently. If I were to buy a new color meter, I would gamble on that one, as the color meter takes into consideration (somewhat) how chrome film reacts. Contact Sekonic for more information, and report back if you do....
> Perhaps I am not as stupid as I thought - my meters are partially to blame.
hey Rob, I guess its just me and you who do NOT have PERFECT exposure with chrome film 100% of the time. Most of the list seems to have never have a problem, it appears we stand alone. After 30 years at this, I am embarrassed that everyone else has had perfect success. Oh well, maybe practice does not make perfect.....or, maybe, I sometimes have been fed erroneous data from my meter....nah, impossible...$100 - $500 meters should give perfect readings under all circumstances, even if they are only calibrated and tested at one color temp.... Yeah, so it's my technique always at fault....
We don't know. Or rather, your test does not tell us - it only subjects several light meters to multiple variations in lighting, and assumes that among them only color temperature had any influence on the results.
You did not even maintain a constant change in EV - the figures are jumping all over the place, at a much greater magnitude than the change in K. For a clean test, only the tested parameter would be variable - a constant growth in EV would be the absolute minimum for an amateurish approximative test.
Posting Permissions
Reply With Quote
Bookmarks