bobn2

Nice illustration. I think the other point here is that it is wrong to assume that the visibility of noise is solely a function of SNR. We don't view SNR, we view a processed image, and that ...

I agree with all of that. I would judge the DR in the normal manner, largest signal level divided by smallest signal level. I would then point out that DR is probably not the best metric when it ...

I don't remember ever claiming that you cannot derive both the saturation level and read noise from the DxOMark DR measurements taken along with their 'measured ISO'. However you still can't do it ...

I think that we both agree on the 't', we just want to look at the it from a different direction.

I think what you're saying is that it hasn't done the job perfectly, which is indeed the case, but it has reacted to the actual photon distribution. If one could arrange to illuminate the sensor ...

I think that the 'measurable' is implicit in the Wikipedia definition. But most DR definitions that I've seen in most fields don't include the 'acceptable', because what is 'acceptable' is ...

Isn't that normalisation?

You'll see that is discussed in some detail elsewhere in the thread. Yes, you are exactly right, any metric needs to be designed with respect to which information it is trying to tell you. In this ...

It's not a matter of what I'd 'like' to see reported, it's just what is the established (so far as I know) definition of 'DR'. I have no objection to other metrics, but bundling them as 'DR' isn't ...

I haven't reproduced your post, for a couple of reasons. One is that Ken and others have covered much of the detail very thoroughly. The other is that maybe the 'I said - you said - he said' style ...

As I remember, I used the background of the Kodak chart, which Kodak says is 18%.

Isn't it 20 after normalisation? Presumably 'picture height', rather than 'half-diagonal', which is the usual 'picture height' in lens resolution, and also in CoC calculations. If so, then that ...

We can test this conclusion in a practical way, simply by taking cameras with similar DR and see if shadow noise is similar. Taking for example a Nikon D5 and Leica M10. Take your choice of DR ...

If they use a black frame then the amplitude signal is zero, and the SNR is 0. The SNNR (see Detail Man's earlier post) is 1. It's hard to know what DxO Analyzer is doing. If it's truly using a ...

What do I need to change to make it accurate?

Good information. Agreed

You can get a 'dumb adapter' from mFT to E-mount, for instance https://www.amazon.com/Haoge-Adapter-Olympus-Panasonic-NEX-VG10/dp/B07DGKJVCM/ref=sr_1_3?crid=2233G2O5K8GMG&dchild=1&keywords=micro+fo ...

I'm not saying that anyone should do anything in particular, I would just prefer that if they are doing something they would be clear what they are doing and what is the rationale for it. The ...

Interesting. Thanks. One interesting difference in terminology is 'electronic camera' as opposed to 'digital stills camera' in the later versions. I guess at that time the Sony Mavica could have ...

The distinction between SNR and SNNR is of importance when we get down to small values. In terms of image sensors, the base SNR of zero is not too difficult. Put the sensor in the dark, there is ...