Dxo mark SNR graph for a7,a7s,a7r

enerlevel · Sep 20, 2016

Hi all,
So I looked at the dxo SNR graph comparison between a7 and the a7s. And according to it, the a7 noise is same as the a7s till 20,000 iso. After that the a7s iso keeps on expanding where as a7 stops at 25600 max.
So keeping that in mind, I ended up buying the a7.
Now I happened to have some old pics taken from a7s in a particular hotel and yesterday I went there again to test my new camera.
Shots taken at same iso (20,000) are way different than what dxo mark graph suggests...
A7s shots were sharper , much controlled colours and less noise.
Obviously the subject and the shots were different but the lighting and iso were the same on both the cameras..
So is dxo mark graph given us false information ? If SNR is same on both the cameras then why is there so much difference in their output out high iso?
Also I am down sampling the a7 to match the 12megapixel of the a7s but still it cannot match the output...

KmH · Sep 21, 2016

Take a look where DXO Mark explains their testing protocols.

To eliminate setup bias, settings, instrument calibration, light levels, cleanliness (including that of the camera and lens), and all other parameters are checked and rechecked before each camera/lens is tested to ensure that all cameras and lenses are tested under exactly the same conditions.

Your test used different subjects and shots, so it wasn't a test made under exactly the same conditions.

With different subjects and different framing it's pretty much impossible to have the same lighting, because the camera records reflected light and each different scene usually has a different reflectance.

enerlevel · Sep 21, 2016

I have tried taking pics of many different subjects. Maybe the noise pattern is same but image sharpness is definitely different.
For example the a7 @20,000 is giving me false colours and the sharpness is compromised.
So after removing noise from both images, a7s is definitely sharper and more detailed.

If the dxo mark graph is OK, then the a7r should also give similar results @20,000 iso... but anyone who has used a7r can easily tell you that this is not true...