DxOMark: Panasonic GH4 Has The Best Micro Four Thirds Sensor Yet!

And then you drill into their graphs and realize that you're never, ever going to tell the difference between the GH4 and the EM1/GH3/GX7/GM1/EM5/EM10 unless you're staring at a test diagram. About the only useful bit of info from the DXO test was that the GH3 has legit ISO100, even though it's labelled as ISO125. GH4 labeled its ISO100 properly.

Another point of contention: why link to that site, which doesn't appear to even link to the source? Is that your site? You should probably add this link: http://www.dxomark.com/Cameras/Panasonic/Lumix-DMC-GH4

I'm a bit of an academic. Proper citation, even as it exists on the web, should be observed.

 

Thanks Bob. The GH4 looks like it will be the king of m43 for a couple months at least, and the DxO crown champion... leading to much joy and much angst from those who pay attention to DxO

 

...by a measurable but immaterial increment?

 

I've never put a lot of stock in DXO, especially the summary numbers. The reality is that any sensor differences between those cameras are so small they're be overwhelmed by any and all processing differences: color balance, sharpening, you name it. Unless you routinely shoot at ISO 6400, make 30 x 40 enlargements that you expect to view from 24" or less, or need razor thin DOF, any of today's current m43 cameras will produce stellar results, IF THE PHOTOGRAPHER KNOWS WHAT S/HE IS DOING.

 

I strongly agree on the summary number DXO comes up with. Even the major category numbers are not all that useful. It's a lot more meaningful to look at the results of their individual tests, but in most cases today what you find is that the differences are really minor. Sensors are simply not what distinguishes front-rank cameras today.

 

Essentially same as the others. Variance will give you small difference like you see.

 

Precisely. And when you consider that the GH4 (and for that matter the E-M1) are basically the same as their predecessors, it's actually rather disappointing. More than two years since the E-M5, we still have yet to see a newer sensor offer a significant improvement for images. Video yes, but stills no.

 

Just incremental improvements until E-M5 II, I bet you. It's just how the industry works. Sensor technology are on longer cycles than the annual body refreshes.

I am interested in how the GH4 auto focus system works in real life. Sounds like a cool implementation. Since it's not a phase detection system there are no missing pixels.

[video=youtube;HwarLGh75OA]https://www.youtube.com/watch?v=HwarLGh75OA[/video]

 

Okay, maybe I should have said 'a significant improvement for still image quality'. AF performance is really orthogonal - poor sensors can have good AF (think E-P3) and good ones poor AF (Fuji's X-Pro 1 comes to mind). But none of these newer sensors lets me get the same quality shot I can take with the E-M5 at ISO 1600 at ISO 3200, or anything like that. That's an example of what I'd consider real progress.

 

Dhazeghi was speaking of image quality, and the differences that DXO reports between the GH4 and EM-1 are probably within the realm of pure sample variation (0.2 bits color depth, 0.1 EV dynamic range, 791 vs. 757 ISO). Even greater difference simply do not show up in practical photography, even if printed at large scale - there are too many real world variables that consume the theoretical advantage. The camera wars, in the short term, are not going to be won on sensor performance and image quality.

Faster focusing speed, better C-AF performance, better UI - these are the things that are going to allow camera makers to differentiate their product, at least until there is some significant improvement in sensor technology.

 

Ditto here!

That's good to hear. Have to shot moving subjects enough to get a feel for it's performance with C-AF nailing subjects moving towards or away from the camera?

 

Speed is still very important to computing, if not all users. However, what you're trying to get at is called the Law of Diminishing Marginal Returns, where improvements are cheap and significant early on. As time goes on, improvements get slower or more expensive (or both). Finally there comes a point where the marginal return is so low, it just makes more sense to plow your resources into a something entirely new.

It's called a law because it applies to R&D, resource extraction (Peak Oil Hypothesis is just a narrow (and slightly misapplied) LoDMR), the size of organizations, and project planning/management. Even agriculture. I mean, picking the low-hanging fruit is easy and bountiful. The last apple at the top of the tree? Not so much.

Camera sensors are already pretty darn efficient. It's unlikely we'll be able to squeeze any more than a couple of more stops out of them without changing them dramatically.

 

Ah, the end of science? Speak to you in 10 yrs :-D

Sent from my C6503 using Tapatalk

 

I want to know how DXO figures the camera ISO scores?
They seem to be wrong to me.

GH4 lower than GH3?
D4s only slightly higher than D4?
D5300 only slightly higher than D5200 and D7100 when the usable ISO is twice as high?
When I am able to take so much better Pictures and Video in poor light at higher ISO, how can they rate them this way?

 

Which is exactly why you should be carefull with DxO. They do very precise measurements, trust these. But the measurements are based on preconditions of what to measure and these do not nessecarily have to relevant for your shooting style...

Sent from my C6503 using Tapatalk