GX7 Lens Review on DxO

Started Dec 15, 2013 | Discussions thread
Anders W Forum Pro • Posts: 21,468
Re: DxOMark promises to retest the GX7

Pictures shooter wrote:

One month passed and we didn't see any new revised results of GX7/lenses from DXO yet.

After some reflections I begin to believe that it's not an accidental solitary isolated failing of DXO test. No, apparently this case is just the next illustration of general trend of DXO tests quality.

So, some serious doubts begin come to my mind and generally it looks like for me that DXO tests are not trustworthy at all.

The simplest example: based on dpreview's GX7 review we can see that GX7's ISO125 value does not extend the camera dynamic range, quite the contrary ISO125 reduces its real DR. I myself face with an unpleasant overexposure in the bright lights if I set ISO125, and it happens more often than if I use ISO200 that is apparently GX7's native base ISO value. You can easily check that DXO's GX7 test shows absolutely different results. Looking at DXO pictures we get impressions that ISO125 is definitely better also concerning GX7's dynamic range value.

The explanation in this case is that DPR and DxO do not define DR in the same way. DPR's definition focuses on the shape of the tone curve of the OOC jpegs. DxO, on the other hand, defines DR as the max signal divided by the "noise floor" (the point in the deep shadows where the SNR equals one) using RAW files. This explains why they come to different results.

If you are a RAW shooter, the DxO DR value is the only one you have any reason to pay attention to. The shape of the tone curve is for you to decide when you convert the RAW file. And DxO is right that the DR of the GX7 is higher at ISO 125 than at ISO 200.

The reason why you get clipped highlights more easily at ISO 125 than at ISO 200 is that the camera uses another (steeper) tone curve at ISO 125 than at ISO 200. Why it does so is a bit mysterious since ISO 125 is "a real ISO" (i.e., uses a lower gain in the ADC process). Consequently, there would be no need to use another tone curve at ISO 125 than at ISO 200. But for reasons unknown, Panasonic decided to do so anyway.

Again, if you are a RAW shooter, the above is of no consequence and using ISO 125 is better than using ISO 200 when the light level permits. If you are shooting OOC jpegs, by contrast, you may have to modify your exposure practice when shooting at ISO 125 or abstain from using any ISO lower than ISO 200.

Next example, I'm the owner both Pany G5 and GX7 and based on my personal experience I can say that G5 provides much worst level of noise in shadows at high ISO than GX7 5 does, and so I could expect that any serious test in any well equipped laboratory should clearly show this GX7 obvious advantage, otherwise it's rather doubtful value of such testings... But I can't find such significant difference looking at DXO test at all, we can see just quite moderate 1...2 dB difference, no more...

As already indicated by Steen Bay, the DxO measure of primary significance with regard to shadow noise is the DR value. If you compare the DR curves for the G5 and the GX7, you will see that the GX7 has better DR at any ISO, just as one would expect given your personal experience with the two cameras.

We could find similar doubtful cases with other cameras too. Can we go on to trust DXO tests?

I think we can generally trust their sensor measurements provided that they are correctly interpreted. I do not rely on their sensor scores since they are rather arbitrary and do not necessarily reflect the measurements I find most important.

When it comes to their lens tests, things get more complicated since DxO have yet to publish a technical definition of their measures of sharpness (P-MPix and acutance). Until they do (they have promised that more technical information will become available during the current month, January 2014), it is somewhat difficult to have a clear opinion of what those measures are good for.

There are also some problems with regard to comparability between the results for lenses that are designed for software correction of distortion and lateral CA (such as many MFT lenses) and those that are not. Since DxO uses the RAW files for all measurements, the software corrections are not applied when the measurements are taken, which implies worse results for distortion, vignetting (which is affected by distortion correction), and lateral CA than if the software correction for which the lenses are designed would be applied.

Like the sensor scores, those for lenses are rather arbitrary so I don't look at those.

The results for the lens tests on the GX7 specifically remain a mystery right now but hopefully we will have a clearer view when DxO performs the retest using another GX7 body that they have promised.

My personal feelings such tests look like too commercially affiliated to push some brands. I do would like to be wrong here.

If there is any bias here, I don't think it takes the form of manipulating results for individual cameras or lenses. What they decide to measure, how they measure it, and how they design their scores is another matter. On the other hand, the measures are known (for the most part), we are free to critcize them as we see fit, and noone forces us to rely on their scores.

 Anders W's gear list:Anders W's gear list
Panasonic Lumix DMC-G1 Olympus OM-D E-M5 Olympus E-M1 Panasonic Lumix G Vario 14-45mm F3.5-5.6 ASPH OIS Panasonic Lumix G Vario 7-14mm F4 ASPH +28 more
Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow