Dynamic Range Comparison: Sony RX100, Olympus E-M5, Panasonic GX1

Amin

Hall of Famer
DxOMark recently published data about the Sony RX100 and Olympus E-M5 sensors, prompting some to recognize that the RX100 actually has a tiny bit more measured dynamic range (DR) than the E-M5 while both have a fair bit more DR than the Panasonic GX1:

8092199616_7d92470719_o.png


Since I have all three of these cameras on hand, I thought this would be a good time to show what I consider to be the value as well as the limitations of these test scores. Are these DR differences observable in practice, and are these numbers really representative?

In order to carry out this test, I chose a scene in my home with consistent lighting and a range of very bright to very dark, shooting each camera at base ISO with a matched angle of view and matched DOF throughout a variety of exposures until I obtained one RAW file from each camera with a similar degree of recoverable highlight detail.

Here was the test scene as seen by the GX1 (left) and RX100 (right):

8092193645_de30317bcf_o.png

In the following 100% crop comparisons, the RX100 is at a slight disadvantage because I have not matched the output size and the RX100 has a higher megapixel count, but I think it will be clear that the small difference in megapixels has little influence on some fairly obvious differences.

Here are the highlight regions from the files at 100%, "pulled" to reveal that recoverable highlight details are fairly well matched:

8092199928_0d01001243_o.png


When you look at the pushed shadow regions from the same files, it's obvious that the RX100 can capture details across a wider swath of contrasting shadows and highlights:

8092199834_a76a4ee651_o.png


Based on the "eye test", the measured DxOmark DR result of 10.6 (GX1) and 12.4 (RX100) stops seems believable.

Now let's look at the E-M5 (left) compared with the RX100 (right).

First the highlights showing that the files were selected to roughly match in terms of recoverable highlight detail:

8092193509_7079eeabba_o.png


Next the "pushed" shadows:

8092193377_4b8fc33e8d_o.png


In this case, the practical results don't seem to bear out the measured results of 12.3 (E-M5) vs 12.4 (RX100) stops.

DxOmark has the following to say about their DR measurements as pertains to landscape photography (source):

What is paramount is dynamic range, especially because photographers will often aim for detail in high-contrast settings, juxtaposing bright sky with shadowy foliage, mountains, etc. Ideally, the dynamic range of the camera should be greater than the dynamic range of the scene, otherwise details in shadows are lost or highlights are burned.

Dynamic range falls rapidly with higher ISO settings, as any analog or digital amplification performed will increase the noise in the darker areas, making it harder to distinguish between fine levels of contrast.


Maximum dynamic range is the greatest possible amplitude between light and dark details a given sensor can record, and is expressed in EVs (exposure values) or f-stops, with each increase of 1 EV (or one stop) corresponding to twice the amount of light.

Dynamic range corresponds to the ratio between the highest brightness a camera can capture (saturation) and the lowest brightness it can capture (typically when noise becomes more important than the signal, i.e., a signal-to-noise ratio below 0 dB).

A value of 12 EV is excellent, with differences below 0.5 EV usually not noticeable.
This scale is open, as incoming light is not a bounded quantity.


In summary, DxOmark results are useful but limited. I'm hoping to see one of the big review sites adopt DxOmark's rigorous testing methods and use them to produce test results which can be evaluated with the "eye test".
 
On a separate note, DR is far from the be-all end-all to me. The GX1 has never disappointed me from an image quality standpoint, and I would love to get the kind of high ISO performance the GX1 gives me from the RX100.
 
forgive my n00bness. All the testing numbers shown look comparable between RX100 and E-M5 except for the "sports" low light ISO. Is that what we are seeing in your results here?

Also on a totally unrelated note, I like shadows :D
 
What we're seeing here is the visual analogue to what DxOmark is rating as "Landscape (Dynamic Range)". I was testing plentiful light, low ISO, high DR. In contrast, the "Sports (Low-Light ISO)" rating deals with low light, high ISO signal/noise performance.
 
Hi Amin.
Are you sure about the equivalence of exposure/highlights topping? The GX1 file needs much less recovering than the E-M5 one in the highlights (of course, it's more underexposed). I mean: lowering the exposure till I obtain a 255 topping white, it results in the E-M5 needing a -3,35, while the GX1 is at -2,15.
This way, the very bad result in shadow recovery by the GX1 is obvious.

Highlight recovery in ACR may be misleading: it goes on putting a plain gray in areas missin any information by the sensor, as you can easily see in your crops. And "near white" detail in the E-M5 (and perhaps in sony sensors in general) is obtained using a smoother curve/transition in the highlights.

But where it tops, it tops. So, it appears that the GX1 shot should have been exposed a stop more...

Am I wrong somewhere?

[edit] I re-read your line: "until I obtained one RAW file from each camera with a similar degree of recoverable highlight detail". Indeed, the amount of detail is similar, with pure white appearing more or less at the values I've given above (-2,20; -3,35). Is the fully blown-out area of the GX1 increasing (a bit/a lot) in size at longer exposures? I'd gladly take a look at a GX1 raw file with a +1 stop exposure if you could kindly upload it... :)

Best
Alessandro
 
Amin,

Thanks for doing that.

I think your test, besides pointing out that the DXO measurements -- useful as they are -- are only a guide to relative performance, raises another question that all photographers eventually need to answer: what are you going for with your photographs?

For me, the answer to that question has two components. First, I want to capture whatever it was that my eye saw that prompted me to take the photo in the first place. Yeah, I know, there are lots of essays that say, in essence, that is impossible because of the differences between the human sensing system and a camera. Nevertheless, press a gun to my head and demand that I 'fess up about what my goal is, and my answer will be: "Shoot what I saw."

The second part of the answer would be: shoot what I saw in such a way that it captures the elements that I responded to emotionally and caused me to want to shoot the photograph. It might be the sunlight playing in the autumn leaves or the gloom of a forest trail or a cloudscape. It is the contrast between light and dark that often brings the drama and the interest to a photograph.

And that brings us to the point of this post. In my view, the eye does not capture high dynamic range. In the scene that you shot, if you concentrated on the brightest part of the scene, your iris would contract, reducing the amount of light and allowing you to see more detail in the bright areas and losing detail in the dark areas. If you concentrated on the dark areas, your eye would open up, allowing you to see more in the darker areas, but losing detail in the bright areas.

In Outdoor Photographer magazine, a company has been advertising a product that allows post processing of scenes so that, in the end, everything looks like it has had the same level of lighting and detail, even when the original scene had high dynamic range. It has the effect of bringing up detail while flattening contrast. The result, to my eye, is bizarre. The photo has a look of unreality as if it were a painting or a computer generated image, but it certainly doesn't look like a scene that has been captured "in the wild."

So thank you again for your post. I'm with the other poster: "I like shadows" and highlights too!
 
Hi Amin.
Are you sure about the equivalence of exposure/highlights topping? The GX1 file needs much less recovering than the E-M5 one in the highlights (of course, it's more underexposed). I mean: lowering the exposure till I obtain a 255 topping white, it results in the E-M5 needing a -3,35, while the GX1 is at -2,15.
This way, the very bad result in shadow recovery by the GX1 is obvious.

Highlight recovery in ACR may be misleading: it goes on putting a plain gray in areas missin any information by the sensor, as you can easily see in your crops. And "near white" detail in the E-M5 (and perhaps in sony sensors in general) is obtained using a smoother curve/transition in the highlights.

But where it tops, it tops. So, it appears that the GX1 shot should have been exposed a stop more...

Am I wrong somewhere?

[edit] I re-read your line: "until I obtained one RAW file from each camera with a similar degree of recoverable highlight detail". Indeed, the amount of detail is similar, with pure white appearing more or less at the values I've given above (-2,20; -3,35). Is the fully blown-out area of the GX1 increasing (a bit/a lot) in size at longer exposures? I'd gladly take a look at a GX1 raw file with a +1 stop exposure if you could kindly upload it... :)

Best
Alessandro


Hi Alessandro,

It's not that the GX1 is underexposed or that the the E-M5 intrinsically has a smoother curve/transition in the highlights. Rather, what you are seeing is a result of the normal tone curve applied by LR to E-M5 files, which relative to GX1 files are underexposed (relative to highlight clipping) and "pushed" by ACR or the in-camera JPEGs.

I did a similar test here where the GX1 fared better in the shadow comparison because I failed to match the highlights (where the GX1 did worse) because I didn't take into account the way that GX1 files are exposed more to the right (than the E-M5) relative to highlight clipping. You can read more about these issues here.

At any rate, I'm attaching the one-stop more exposed GX1 RAW per your request. If you compare it to the E-M5 file in any RAW converter (not just ACR/LR), you'll see that there is noticeably less recoverable highlight detail as well as shadow detail in this GX1 file.
 
... And that brings us to the point of this post. In my view, the eye does not capture high dynamic range. In the scene that you shot, if you concentrated on the brightest part of the scene, your iris would contract, reducing the amount of light and allowing you to see more detail in the bright areas and losing detail in the dark areas. If you concentrated on the dark areas, your eye would open up, allowing you to see more in the darker areas, but losing detail in the bright areas....

Very true, and that's one reason why I have never been let down by GX1 image quality. I tend not to "push" shadows very much during postprocessing, so these differences don't come out. However to play devil's advocate, the same argument can be made re: sharpness - If you concentrate the eye on anything, only the central area of your vision is remotely sharp and everything in the periphery is blurry. We only have the sense of a sharp reality because our eyes scan around and the brain integrates signals. In the same way, when you scan around a high DR scene, the brain puts together a high DR image.
 
Fine, thank you very much. The GX1 file exposed as the E-M5 one has indeed a huge loss in the highlight details, and still less recoverability in the shadows. The sony sensor looks one generation ahead...
Nonetheless, the GX1 is a fine and perhaps underrated camera, at least at the price tag it's got now.
I've got one - while giving away the E-M5 for financial reasons - and found the usability I enjoyed with the GF1, a bit more immediate and friendly than the Oly. Sure, the E-M5 is a better camera overall, but in my case it's been a € 300 vs € 800 deal...

Alessandro
 
Hi Alessandro, I agree with everything you said and in many respects prefer my GX1 to the E-M5. This comparison was not meant to establish one as better than the other, only to show a visual version of the DxOmark results.

Sent from my Galaxy Nexus using S.C. App
 
The difference in dynamic range between the GX1 and E-M5 is very real, but not something that you will always notice every time you press the shutter. Not every scene you face will require 12+ EV to record, nor look better by doing so. In practice a higher DR camera has a larger working range in difficult lighting conditions.
 
I just noticed that DxOmark shows higher DR for the RX100 at ISO 80 than for ISO 100, and they don't even have a data point for manufacturer ISO 125. This is very odd because the camera will default to ISO 125 even in auto ISO mode and bright noon sunlight. In fact, the RX100 manual specifically states: "When [ISO] is set to a value lower than ISO 125, the range for recordable subject brightness (dynamic range) may decrease", which strongly suggests that ISO 125 is the true base ISO for this camera.
 
I noticed that as well, but the same is the case for other cameras with a base ISO and the option to use a pushed lower ISO. Look at the Nikon D5000 for instance. I'm guessing it's something to do with the way they test dynamic range, but it doesn't really inspire confidence that the rest of the measurements are useful.

The DPReview dynamic range measurements for the RX100 clearly shows a lower DR for ISO 80 and 100.
 
forgive my n00bness. All the testing numbers shown look comparable between RX100 and E-M5 except for the "sports" low light ISO. Is that what we are seeing in your results here?

Also on a totally unrelated note, I like shadows :D

Actually I was shocked when I saw the low light performance score of dxomark. My own impression is that the RX100 is doing better than even my old APS-C Canon 40D. Looking at the pushed shadow areas it seems to me there is significantly less grain on the RX100 compared to the GX1 and it´s only slightly worse than the OM-D. The only thing not to like about the RX100 ISO behavior is the "cheating" that ISO 3200 in reality is only ISO 2000. Admittedly a lot of manufacturers do the same.

Wolfgang
 
The DPReview dynamic range measurements for the RX100 clearly shows a lower DR for ISO 80 and 100.

It shouldn't be lower for those; it should be the same. All of them are applying the same analog gain, so at the RAW level they should all look the same. On a camera with base ISO 200, shooting at nominal ISO 100 should give you the same RAW data as shooting at nominal ISO 200 with +1EV exposure compensation.
 
Back
Top