I have written previously my assertion is based on extensive tests carried out by the U.K Professional Photography Magazine using a Minolta Autometer Vf with various DSLRs and types of film, the tests were extensive and the conclusion irrefutable.
Originally Posted by wiltw
The 'digital calibration' vs. 'film calibration' is a bunch of hooey!!!, and they used to believe the World was flat.
Picked up an 18% gray card yesterday and performed your test with direct sunlight ~7:15 a.m. and got these results:
Originally Posted by wiltw
E-P1 = 1/60 sec @f/16 (does not provide 1/3 stop detail)
SBC reflective = 1/60 sec @f/16 +2/3
SBC incident = 1/60 sec @f/16 +1/3
Given that noparking's test reflects what I said it would, is sufficient to demonstrate the earlier point that I made about comparing film vs. digital meters!
Ben, can you link us to the test you refer to?. The body to which you refer has either got it right and the fact is that wiltw's test and noparking's execution of it is simply luck or the body had got it wrong( seems unlikely) or there is a way of reconciling these apparent opposite conclusions which may be the most likely explanation.
I hope nobody thinks that a digital body will give random exposure values just because it has a digital sensor.
The exposure reading of a digital camera is not exactly matching the exposure in the FILM-ISO standard - OK - but it is not without correlation to light ;)
0.66 is not a small mistake and certainly it cannot be considered satisfactory for slide film use. But once one gathers - after a little experimentation - that his digital camera is yielding results which are let's say 0.66 EV "more" than what would be appropriate in the FILM-ISO standard, then one compensates and the reading is reasonably accurate along all its EV range.
The greatest source of exposure mismatch is probably the "matrix" or "pattern" camera metering as that will introduce some "guessing" by the camera regarding which part of the image is the one the photographer would - likely - consider the important one. So that is never comparable to an external light meter reading which averages the scene in front of him.
Folks need to consider that the len's RATED aperture may in fact NOT be true to its real aperture size. And the light transmission of the lens might be less efficient due to the use of less expensive coatings. So unless the lens is identical in all the testing, it would not be surprising that a meter reading does not match between two different brands of cameras, both with supposedly f/1.8 lenses. Magazines like Modern Photography and Popular Photography used to pose the ACTUAL max aperture (vs. the manufacturer's rating), for example the Olympus 100mm f/2 might really measure as f/2.1, and pass less light to the meter.
I have taken two different f/2.8 lenses for my digital camera, and I have seen about 1/3 EV difference in the level of light seen by the meter, at the so-called same f/2.8 setting when both lenses were set to same FL to capture the same field of view!
Add to the meter comparison issue the fact that the ISO equation also allows variability due to the manufacturer chosen C value and K value for incident and for reflected meter calibration. So even if the same lens is mounted on two different camera bodies, the readings from their meters may not match.
So in practice, while 0.33EV might be the standard of variability allowed within the ISO equation itself, the fact that the lens passes different levels of light due to actual vs. rated aperture and due to transmission variability, adds more range to the variability of results.
Although digital, color negative film, color transparency film, B&W film, polaroid film all should be exposed differerently, the meters on all these cameras should give the same readings. Or else the meters are wrong.
For the record, the test above was done with the camera in aperture priority mode (f/16) and a Tiffen 40.5mm uv filter attached. I just re-did the test without the uv filter
Originally Posted by noparking
E-P1 Auto exposure mode
E-P1 = 1/320 sec @f/10
SBC-R ~ 1/320 sec @f/8
SBC-I ~ 1/320 sec @f/9
E-P1 Aperture priority mode
E-P1 = 1/125 sec @f/16
SBC-R = 1/125 sec @f/16 +1/3
SBC-I = 1/125 sec @f/16 +1/3 (this time reflective and incident readings matched - I have no explanation why)
All my Gossen meters read 1/3 to 2/3 stop more light than my other meters. Judging from your numbers the Luna Pro is just fine. Take it out and shoot a roll; you'll see it's okay. You may have to make a slight adjustment in your development for the tiny bit of underexposure but I doubt you'll be able to detect it.
Noparking, you should find in your E-P1 manual the way to choose the metering pattern: something like "average", "center-weighted", or "matrix" (where "matrix" can be called "pattern" or something to that effect).
Then you should choose "center-weighted" or "average". The default choice is probably "matrix".
Auto-exposure (full auto, that is normally called "Program" mode or whatever P in PASM means) or Aperture priority do not have any influence in how the light reading is performed so there should not be any difference in exposure between the two automatic modes (or any other exposure mode).
When reading the light reflected by the bloody grey card a small different in the angle between card and light meter (or camera) matters. In your first example you got a 1/3 EV of difference (f/8 and f/9) while in the second example you had the same reading between incident and reflected light reading. I suppose the first time you did not have the exact same angle, or maybe you were projecting some shadow by mistake.
Be wary when doing this kind of experiments of the shadow you, the photographer, project on the grey card or on the light meter. It's easy to make this mistake. You'll see that if you repeat your measures with great care, you'll probably obtain very coherent results.
I don't use grey cards any more. The "caveats" in their use are many: angle between light source, camera, and light meter, mistakes due to shadows, mistakes due to "mirror effects" (reflectivity of the card at certain angles) and general clumsiness and difficulty in use. Incident metering works MUCH better and is MUCH faster :)