Dilemma with my upcoming test - metering

I am doing a 1Ds Mk.II, 1D Mk.II and a 5D test in the next 2-3 days. Doing some initial setup I'm having a hard time reconciling a proper 3200 exposure.

Both cameras 1Ds/5D are set to M with constant light. Both have the same metering modes. Same lens even. But I'm getting two different exposures because the meter is dialed to the sensor.

I have always been a proponent of comparing camera to camera on the same settings but I'm now thinking that that was false. As long as the metering mode is the same you have to match to what the person operating would do.

I'd see a scene and meter 2/3 over at ISO 3200, pulling it back to 0 ev in post. If the light and the scene are the exact same then it should be a real world application. If a scene demands 50 and f/2.8 from a 5d and the EXACT scene with a 1Ds demands 40 and f/2.8 then that's a real world test.

I am open to some suggestions here because my initial results are showing underexposure from one camera and it is increasing the noise, nullifying the test.

Am I being clear?

I had a similar result when I was comparing a XT to the XTi to the 40D. They all metered just a bit differently. I choose to make all the settings identical and you could clearly see the XT over expose just a bit, the XTi under expose just a bit and the 40D spot on. I wish I had gone back and done it the way your proposing. It seems to make sense to me.

I want as many nay sayers to be happy. I'm going %100 crop with pixels on target. The subjects in every scene will fill the frame different with resolutions and crop factors varying. I want it to be real world pixel - no resizing, dead on exposure from the meter.

Then showing the same with +2/3 ev pull back in post to see what one reacts best to "shooting to the right".

So I got one thumbs up from Gjl. Anyone else?

Do you have access to an incident light meter? You could meter the scene separately then set the exposure on each camera to the incident meter. You can't get much more accurate than that.

I do and that seems the right thing to do but if the meter tells me 1/50th and f/4 then that's the same as me setting the cameras to M and doing them as I have found.

One is underexposed with both cameras using the same setting, same as what I'd do with a handheld meter.

Then, I agree with Gjl. Just shoot them as you suggested and show the differences.

Mark_Cohran wrote in post #4668681
Do you have access to an incident light meter? You could meter the scene separately then set the exposure on each camera to the incident meter. You can't get much more accurate than that.

Wouldn't that assume that the digital sensors between the different bodies are all equally sensitive to light and all behave the exact same way given a fixed illuminance level. That was very true with film but I have a suspicion that each make of digital sensors have slightly different characteristics even when the illuminance levels are constant and the camera's meter is adjusted to compensate. A good experiment would be to compare the cameras metered settings to an external light meter.

I'll preface the reason for not shooting at the exact light meter suggestions and at the same lens settings to defer to the "real word" user who looks at their meter.

This demands though that I use constant light and measure the light every shot with the hand held meter as a control for the test.

I'll reference back to this thread as well in the test.

gjl711 wrote in post #4668746
Wouldn't that assume that the digital sensors between the different bodies are all equally sensitive to light and all behave the exact same way given a fixed illuminance level. That was very true with film but I have a suspicion that each make of digital sensors have slightly different characteristics even when the illuminance levels are constant and the camera's meter is adjusted to compensate. A good experiment would be to compare the cameras metered settings to an external light meter.

If the cameras don't meter light the same way, i.e. they measure differently than a calibrated incident light meter, then I think there's an issue. But, you know, in the film days we often rated some films at more or less than their ISO designated sensitivity simply because in practice we felt the film was closer to our settings than what the manufacturer stated. For example, some films with a 400 ISO manufacturer's rating were often used at 320 ISO. Maybe there's something similar occurring here. I wouldn't expect the sensors to be very far apart though.

Between the 1Ds Mk.II and the 5D with a high contrast scene I am seeing a 2/3 stop differential. I was pretty amazed.

cosworth wrote in post #4668830
Between the 1Ds Mk.II and the 5D with a high contrast scene I am seeing a 2/3 stop differential. I was pretty amazed.

You've got me curious. I think this weekend I'll meter a gray card with several different bodies and compare them (as well as an incident reading) to see what the variation will be from sensor to sensor.

Hey Jason! You are kinda complicating things but I think its the right way to go. How about?
1. Using a couple external meters to verify the light level
2. Then shoot the cameras at the external meter settings.
3. Use your experience/chimp to get the correct ETTR

You could then show the differences between cameras and the value of ETTR in low light conditions.

Mark_Cohran wrote in post #4668850
You've got me curious. I think this weekend I'll meter a gray card with several different bodies and compare them (as well as an incident reading) to see what the variation will be from sensor to sensor.

I would be very interested in seeing the results. I would think that they should be close but there is much more in the digital world that can throw in variations that with film. You have the gain of the amplifiers, the number of stages of amplification, capacitance in the chips themselves, resistance and loss, the addition of noise reduction circuitry and the A/D conversion. All this stuff could affect the sensitivity of the sensor and who knows, maybe the manufacturers just adjust the metering system to compensate.

cosworth wrote in post #4668830
Between the 1Ds Mk.II and the 5D with a high contrast scene I am seeing a 2/3 stop differential. I was pretty amazed.

You specifically stated with a high contrast scene....so they begin to agree with low contrast? I guess what I'm asking is would they all meter a grey card the same?

It's hard to believe there is as much as a 2/3 stop difference. That's pretty big.

Oh, and I think I caught it in your earlier setting but are you suggesting the 1D is 2/3 stop less sensitive to light?

The complaint you will get is, if I need f/2.8 and 1/50 for the shot I'll set the 1D 1/3 stop higher in ISO to get it.

DDan wrote in post #4668888
3. Use your experience/chimp to get the correct ETTR

I'm trying to show the subtle differences between high ISO performance. Chimping won't cut it. Exposure is VERY critical to noise control.