There is something wrong with "equivalent aperture" comparison.
Q7 has 1/1.7'' type sensor, multiplier is 4.55.
2.8 x 4.55 = 12.74 (and not f/17 as chart shows)
Smallest sensor, fullHD. I am skeptical.
Sergi Gabriel: If max resolution = 4608 x 3456 (4:3 ratio), then Effective pixels
= 4608 x 3456 = 15 925 248 Pixels = 15 925 248/1024/1024 = 15.1875 MegaPixel.
For 3:2 ratio max resolution = 4608x3072 = 14.156 MegaPixelsFor 16:9 ratio max resolution = 4608x2592 = 11.39 MegaPixels
Not 16.1 megapixels like in specification. It is narketing game from vendor.
Or an inattention of dpreview. There is a native 4:3 sensor in the GH3, no real multi-aspect.
ogo: Why can't people just admit that a 4/3 sensor can be good at high ISO, and instead of that just say that Olympus cheated on high iso ?
Now think about this :- R. Butler clearly stated that you can't use those studio samples to judge ISO scale- for those shots E-M5 exposure times are the same that Panasonic GX1. Did anyone accuse Panasonic to cheat here ? Just look at DXOMark of GX1 (DXOmark being the "reference" for those accusing Oly to cheat). GX1 is right on scale- you can't compare exposure for 4/3 and other sensors because the aspect ratio is not the same (3:2 vs 4:3). 3:2 sensors have more white on the sides to expose, thus making the comparison impossible- you ignore light transmission differences between lenses (T-stops) which can have a big impact- studio samples on another site (focus-numerique.com) show same exposure times for NEX-5N and E-M5, and still the noise results are very comparable with what is shown here.
After you wrote technical supidities and you could not demostrate any of them the only weapon that resmains for you is scoffing and indulging in personalities.
@ Antony John:
Quote:"@ Manmachine, quick look, Panny 45-175 5.2 TStop.And it doesn't have "15 to 20 elements""
So what we are talking about?
This sentence shows that you simply do not know what stops and aperture means.
"I give up with you"
It is a best thing from anybody who continously mix up the ttl and the external metering.
@ Antony John
You could not show any example from DXO where difference in transmission more than 0.1 EV with similar lenses, but you talk about 0.7EV effect.
So which statement was wrong?
The only thing I can do is repeat myself: Your examples show that you simply do not understand the simplest basics of optics, like aperture etc.
Do you really know what that word "aperture" means?
Panasonic 45-175 is an 1:4.0 - 5.6 lensThe TStops: 4.3 - 5.7, the 5.2 is an average value.
Click on "measurements" and click "transmission" after.Look at the bottom diagram called "T-Stop difference to manufacturer"All that you see is the effect of entrance pupil/focal length ratio, not the effect of reflection on surfaces.
And now, do the same thing with a prime lens.
"More white on image means more incoming light"
There is no connection between histogram and a cropped image where we try to find details and define the noise.
Download two images, load them into an image editor software, and measure the pixels at the same area. They are very similar, but exposure values show 2/3 EV difference.
DXO Mark results:
Canon 50mm 1.4 : 1.6TStopsNikon 50mm 1.4: 1.6TStopsZeiss 50mm 1.4: 1.6TStopsSigma 50mm 1.4: 1.7 TStopsSony 85mm 1.4: 1.6 TStopsSigma 85mm 1.4: 1.6 TStopsNikon 35mm 1.4: 1.7TStops
So where is that huge difference?
The difference between real and nominal aperture is much significant, that difference is transmission. Look at the lens patents. The real aperture of an 1.4 lens is about 1.35 to 1.45.
"3:2 sensors have more white on the sides to expose, thus making the comparison impossible"
I hope you have heard about method of measure the incoming light. It is only 50 years old at least.
" light transmission differences between lenses (T-stops) which can have a big impact"
It is a misbelief. The difference between transmissions is only a few percent, especially in these lenses which do not contain 15..20 elements. Light loss at surfaces with modern multi coating is 0.2 - 0.3%. We are in year 2012, not in 1930.
Twenty surfaces can produce 5-6% loss, the difference between transmissions of two prime lenses is only 2-3%.
manmachine242: ISO3200:NEX5n: 1/800, f87D: 1/640, f9E-M5: 1/800 f6.3
ISO200:NEX5n: 1/50 f87D: 1/40 f9E-M5: 1/50 f6.3G3: 1/80 f6.3
The difference is consistently 2/3 EV.
real sensitivity of E-M5 at ISO200 = ISO125 on 7D, NEX5n, G3real sensitivity of E-M5 at ISO3200 = ISO2000 on 7D, NEX5n
citizenlouie: You have misinterpreted the meaning of exposure value and ISO.
There is a problem with all dpreview test samples if the compared cameras have different aspect ratios.
The constant height of object plane is invented to measure a special resolution parameter, the linepair / picture height.
But now, we try to judge the noise and noise filtering performance by means of details. In these samples the optical, lateral magnification is different with 3:2 APS-C and 4:3 mFT sensors. The same object shows different amount of pixels on different sample photos even if the cameras have same number of pixels.
It makes the whole comparing a bit meaningless. The 3:2 and 4:3 crops of object plane should have same diagonal size, not same height. I would not like to link that french site which uses the correct technique (only from this point of view), but dpreview should have rethink this.
There is an extremly big problem if the difference between transmissions causes 2/3EV loss...
Just Having Fun: Richard said, "I wouldn't try to draw conclusions about ISO accuracy from this scene".
It is very possible the lighting was slightly different. The fact is we are only guessing, and do not have all the facts.
I hope you are right.
ISO3200:NEX5n: 1/800, f87D: 1/640, f9E-M5: 1/800 f6.3
manmachine242: "Sensor size: Four Thirds (18.7 x 14 mm)"
I think "Four Thirds" is a typo.
Sorry, but you are totally wrong.
FourThirds (4/3) is a sensor size standard, the name is deriving from the external size of analogue video camera tubes (Vidicon) as many others 2/3'', 1/2.5'' etc.
4:3 is the ratio of two sides.
There are FourThirds cameras with natively non 4:3 aspect ratio. (GH1 and GH2)