# OMD now on DPreview's studio scene comparison

Started Apr 25, 2012 | Discussions thread
Shop cameras & lenses ▾
Re: OMD now on DPreview's studio scene comparison

Anders W wrote:

Timur Born wrote:

Detail Man wrote:

The Shutter Time on the EM5 was longer than on the GH2 by a full 2/3 EV ("stop"), but have look for yourself. The recorded peak and average RAW signal levels are (nevertheless) significantly lower in the case of the EM5 relative to the GH2. No less than 1.537 EV less. Hard to ignore ???

I am not blaming you Timur, but in the interest of not having evidently false information repeatedly disseminated on the forum, let me point out that Detail Man's claim that the signal levels of the E-M5 and GH2 studio scene RAWs differ by 1.537 EV is clearly wrong. The correct value is about 0.84 EV. In the following, I show that such is the case and explain why Detail Man reaches another conclusion.

The average RAW signal levels that Detail Man uses for his calculation, i.e., those for the ISO 200 E-M5 and GH2 RAWs (after subtraction of the black-level offset) are as follows:

E-M5
R 202.1
G1 372.4
B 149.7
G2 377.0

GH2
R 365.6
G1 725.0
B 271.1
G2 720.1

where R is the red channel, B, the blue channel, and G1 and G2 the first and second green channel.

In the post in which Detail Man originally calculates his 1.537 EV difference, which can be found here

he chooses to base his calculations on the unweighted sum of the four channels and then relates the resulting value to the max recordable level. I prefer instead to base my calculations on the weighted average of the channel averages, i.e., (R + B + 0.5G1 + 0.5G2) / 3, but this difference between us is of little practical consequence for the results.

Right. But no compelling reason not to do that. Had thought about doing the same - but the numerical differences were close enough to insignificant that I did not bother. Sounds good to me.

With a weighted average for the E-M5 of

(202.1 + 149.7 + 0.5*372.4 + 0.5*377.0) / 3 = 242.17

and for the GH2 of

(365.6 + 271.1 + 0.5*725.0 + 0.5*720.1) / 3 = 453.08

we have a difference as expressed in EV of

log2((453.08 / 3974) / (242.17 / 3811)) = 0.84.

The central mistake of Detail Man is to include in his calculations the shutter speeds used by DPR for the two cameras. First, these speeds are clearly irrelevant if we talk simply about the signal levels in the two files, as Detail Man does in the passage quoted above. They are also irrelevant for the purpose of indirectly trying to determine how much light the two cameras were effectively given since, as reported by Andy Westlake, the lighting conditions for the studio samples are not strictly controlled and may vary from one camera to another.

Yes, I do at this point agree with you on that (as you have approached it, and for the reasons that you offered - then and now), and better understand your position on that. Fair enough. Getting to what is important in making comparisons is what is truly important. Good show ! ...

First, I have already tried repeatedly and at considerable length to explain to Detail Man why he is wrong about his 1.537 EV estimate for the purposes he uses it. Since, after doing this, I was ultimately met by a series of rather strong personal insults, as shown here

I will not respond to any post of his.

I know, and understand. I apologize for my curtness expressed. I simply became frustrated with the tedium of our conversation surrounding the the "Shutter Speed" bit - but your position on that is (technically) reasonable. Somehow, simply seeing "spell out" your position in numbers enabled me to better "see" your position. That's more about how my own mind operates. Peace.

However, I am happy to respond to anyone else having questions about the matter discussed.

Second, what I have said above should not be taken to mean that I am out to defend DPR or that I am personally convinced that the studio samples for the two cameras at issue are indeed fully comparable. Rather, what I have said should only be taken to mean that my factual basis for considering the comparability issue differs from that of Detail Man.

Complain
Post ()
Keyboard shortcuts: