DxOmark Publishes Panasonic G3 Sensor Data - Does the G3 Deliver?

[iliakoltsov]

Art i totally agree with you, and on top of that you put M4/3 with next to no high-end lenses and a D5100 ( which is announced around 600 to 700 euros) now the customer unless it is a specific buyer that knows the advantages of the M4/3 mount (not many of them) the customer will go with best choice hence Nikon in this case. I personally think of moving to the Alpha 580 or maybe SLT 55 and have the G1 as the manual digital body.

 

[soundimageplus]

Interesting! It seems that even with improvements this 16mp sensor only matches that of the EP-2. So there seems little point in using something that does not better the existing sensor.

Again, as Amin says, this shows some of the problems of these DxO "tests". For example I have a GH2, and since its the nearest I can get to the forthcoming G3 sensor, it to me is in almost every way superior to the sensor on my recently departed E-P2.

One of the problems I have with DxO is I don't have any information about what lenses they use, or indeed if they use a lens at all.

In terms of how many MP's, thats a question of personal preference. Firstly if a 16MP sensor can match a 12MP then doesn't that show that the 16MP sensor is actually very good, since its more populated, with no loss of "quality"? If an increase in MP is not important to you then fine, but it is to me and many other people.

I remember people saying 6MP was enough, then 10MP was enough, now its 12MP thats enough. Well again, thats down to what you want.

Personally I'm very happy that cameras continually improve, and that we get more resolution with less noise. I'm always a bit puzzled why others seem to regard it as the work of the devil!

 

[RT_Panther]

The High-ISO performance difference between the G3 & GH2 is so minimal, it's not even worth getting worked up over :smile:

 

[Hikari]

Again, as Amin says, this shows some of the problems of these DxO "tests". For example I have a GH2, and since its the nearest I can get to the forthcoming G3 sensor, it to me is in almost every way superior to the sensor on my recently departed E-P2.

But the DXO test says the GH2 is better than the E-P2. So where is the problem?

 

[Hikari]

One of the problems I have with DxO is I don't have any information about what lenses they use, or indeed if they use a lens at all.

They are measuring the sensor output. The lens is irrelevant. They also have a section of their methodology. Basic, but enough to answer your questions.

 

[soundimageplus]

They are measuring the sensor output. The lens is irrelevant.

Well it might be to you, but I like a lens on my cameras. Helps me take pictures.

But the DXO test says the GH2 is better than the E-P2. So where is the problem?

Exactly. A 16MP m4/3 sensor is "better" than (in DxO terms) a 12MP m4/3 sensor. Not a problem for me.

It was suggested that because the G3 and E-P2 have identical results, why get the G3? I was indicating why.

 

[soundimageplus]

They also have a section of their methodology. Basic, but enough to answer your questions.

So, you'll be in a position to tell me what lens (or not) they used? I am unable to find anything that gives me any information on that.

 

[Hikari]

Well it might be to you, but I like a lens on my cameras. Helps me take pictures.

Then why bother going to DxO if you want systemic data? And why bother bashing DxO when you don't understand nor care about their tests?

 

[Hikari]

So, you'll be in a position to tell me what lens (or not) they used? I am unable to find anything that gives me any information on that.

I just explained the lens was irrelevant. Would you care about the tripod they used if any? This is simply a sensor response to an input. This allows for direct comparison of sensors in cameras regardless of optics. And that is the point. And that is really useful, perhaps not to you, but to many folks it is.

 

[soundimageplus]

Then why bother going to DxO if you want systemic data? And why bother bashing DxO when you don't understand nor care about their tests?

I don't. But thats how the thread started. If you noticed I take very little regard of what they say, but others do, and seek to come up with some notion that the G3 therefore isn't worth bothering with.

Plus you didn't answer my question. Can you explain to me why I should care about their tests? I'm always open to be convinced. Do they use a lens? If so which lens? I may be stupid but somewhere I have this idea that if they tested the G3 with a 20mm f/1.7 lens and the E-P2 with a 14-42mm then the results might be different. Or is that not the case? Can you explain to me how they get an objective, meaningful result between two sensors? Do they use the same lens or no lens at all? You say its there somewhere but I can't find it. Is that an unreasonable request?

I'm not DxO bashing, I've just never had somebody explain to me why I should take their results seriously. If you can do so then I may well become a convert. Again is that an unreasonable request?

 

[Hikari]

Let me chime in here with objective testing. The point of objective testing is to show how something acts. That is it. Objective testing does not say which is "better." That would be subjective testing.

 

[soundimageplus]

Let me chime in here with objective testing. The point of objective testing is to show how something acts. That is it. Objective testing does not say which is "better." That would be subjective testing.

I'm still no closer to understanding exactly how and what DxO are testing. You seem reluctant to give me even a clue as to how they carry this out.

If you won't do it, then at least give me a link as to where I might find it.

 

[Hikari]

I don't. But thats how the thread started. If you noticed I take very little regard of what they say, but others do, and seek to come up with some notion that the G3 therefore isn't worth bothering with.

Plus you didn't answer my question. Can you explain to me why I should care about their tests? I'm always open to be convinced. Do they use a lens? If so which lens? I may be stupid but somewhere I have this idea that if they tested the G3 with a 20mm f/1.7 lens and the E-P2 with a 14-42mm then the results might be different. Or is that not the case? Can you explain to me how they get an objective, meaningful result between two sensors? Do they use the same lens or no lens at all? You say its there somewhere but I can't find it. Is that an unreasonable request?

I'm not DxO bashing, I've just never had somebody explain to me why I should take their results seriously. If you can do so then then I may well become a convert. Again is that an unreasonable request?

You don't have to pay attention to or care about their tests at all. Just like MTF, the results will make little difference to how you take pictures. You will simply adjust to take limitations of your camera accordingly. And like MTF, it will not describe subjective criteria in you image.

Again, the lens is irrelevant because it has no influence on the result. I have no idea whether they test on an optical bench with standardized optics or not, but that does not matter. All they care about is standardized illumination of the sensor. Then they use the RAW output to determine response. That is it--sensor input/output.

Now, for others that have a need for objective measurements of sensor response, the data is useful. For some who may be trying to understand the response of a sensor to explain things in their images, this can offer clues to an answer. For a qualitative measure of quality ("better"), you are using the wrong tests.

 

[Hikari]

I'm still no closer to understanding exactly how and what DxO are testing. You seem reluctant to give me even a clue as to how they carry this out.

If you won't do it, then at least give me a link as to where I might find it.

I posted this before your previous response. This is a general statement.

 

[soundimageplus]

You don't have to pay attention to or care about their tests at all. Just like MTF, the results will make little difference to how you take pictures.
Now, for others that have a need for objective measurements of sensor response, the data is useful. For some who may be trying to understand the response of a sensor to explain things in their images, this can offer clues to an answer. For a qualitative measure of quality ("better"), you are using the wrong tests.

Well thank you for telling me what I don't have to do and what I'm doing wrong. I was attempting to try and understand something. However it seems I won't get any help here.

 

[Hikari]

Well thank you for telling me what I don't have to do and what I'm doing wrong. I was attempting to try and understand something. However it seems I won't get any help here.

So you don't understand the plots? The type on data? What don't you understand? Signal and noise? Dynamic range? Sensitivity?

 

[Amin]

But the DXO test says the GH2 is better than the E-P2. So where is the problem?

Actually, DxOmark gave the same overall score to the G3 and E-P2 because the overall score is based on base ISO testing. Looking at the primary data, it is clear that the G3 is testing better than the E-P2 (especially in dynamic range) at higher ISO values.

David, DxOmark has some information about their testing methods published on their site. There is no lens on the camera in their sensor tests, and sensor resolution does not affect their results.

 

[soundimageplus]

So you don't understand the plots? The type on data? What don't you understand? Signal and noise? Dynamic range? Sensitivity?

How do they get their results? What do they actually do with the sensor? Do they use a lens? Do they click the shutter and take a picture? Do they produce a image? Do they then process it?

 

[soundimageplus]

Actually, DxOmark gave the same overall score to the G3 and E-P2 because the overall score is based on base ISO testing. Looking at the primary data, it is clear that the G3 is testing better than the E-P2 (especially in dynamic range) at higher ISO values.

David, DxOmark has some information about their testing methods published on their site. There is no lens on the camera in their sensor tests, and sensor resolution does not affect their results.

Thanks for that. I had trouble finding it though. Do you have a link?

 

[Hikari]

How do they get their results? What do they actually do with the sensor? Do they use a lens? Do they click the shutter and take a picture? Do they produce a image? Do they then process it?

It is unprocessed RAW data from the sensor. Just as they state on the site. They are not testing spatial attributes, so why are you interested in optics and lenses?

Try this for the details:

DxOMark - Detailed computation of DxOMark Sensor normalization

This might help as well:

http://www.dxomark.com/index.php/Our-publications/DxOMark-Insights/Canon-500D-T1i-vs.-Nikon-D5000

In fact they have an insight page:

http://www.dxomark.com/index.php/en/Our-publications/DxOMark-Insights