As is always a bit of fun, I’ve gone through and pitted these two cameras against each other to see if I could tell much in the way of a difference in the file quality. DXO says there’s a 4 point difference in the performance between these two sensors (in the GX7’s favor) yet they seemingly employ the same sensor and processor… Seems a little weird, and while I feel DXO does a good job at giving all of us a great resource from which to compare different cameras, it didn’t seem to add up on paper. C’mon in, We’ll have a look at files from each of these micro 4/3 cameras…
Let me start by saying that I’m not refuting DXO’s findings. Quite the opposite really. I am curious to see what a 4 point difference looks like as I have both of these cameras in my possession, and a need for a project for the blog.
Let’s get right into it shall we?
We’ll start with high ISO noise performance.
Below, all shots were recorded at exactly replicated exposure values at the listed ISO values, with the tungsten white balance set in camera, shooting in RAW, converted by default profiles in Aperture 3.5.1 without any noise reduction applied using the Pana-Leica 25mm f/1.4 Summilux lens.
To get some range in the images and try to capture all the way through clipped shadows, I side lit the entire scene by one 60w bulb above camera to the left, and to gain a more focused specular highlight, I aimed a small maglite at the glass on the left (which explains the shadow). Click on any to see them larger, or any of the 100% crops to see it full sized.
Alright, no huge surprises here to my eye. I’m quite impressed by the GM1’s more true tungsten white balance preset. Both do pretty well to keep color consistent throughout the ISO range, even in the stratospheric 5 digit values.
To my eye, the GX7 handles the noise better up to about ISO 6400, but at each ISO setting up through 3200 the GM1, while a little noisier, is sharper. At ISO 6400 they both seem to more or less even out on the noise and sharpness and from there it just starts to get a little messy. Again, props to both for maintaining color fidelity through the entire range.
At ISO 12,800 and 25,600 they’re both noisy as all get out and I’d probably stick to my previous threshold with the GX7 at ISO 6400 unless absolutely necessary, but as you may have seen on my blog in the past with these high ISO comparisons, running one of these high ISO files through Topaz DeNoise 5 works some type of digital sorcery. Sure we don’t magically regain resolution that wasn’t there to begin with, but it can all but eliminate the noise while doing a fine job at retaining whatever detail might have been there. See below, and please click to see larger:
ISO 25,600 after a run through Topaz DeNoise 5:
Compare those to the unedited ISO 25,600 crops directly before. I’m always impressed with what DeNoise can do to make a messy file very usable. Not that this is a Topaz plug necessarily, but click HERE if you’re interested in downloading a full free trial of Topaz DeNoise 5 to give it a run through on your own high ISO files. While the two crops are not tack sharp, the noise is pretty much eliminated while the edges on the letters and numbers are entirely legible, the colors don’t suffer as badly from the noise reduction smear as I see when using other noise reduction software, and remember these are 100% crops.
How about dynamic range?
I know it’s boring to look at more of the above shots, but it was set up to intentionally capture the entire luminosity scale capable on these sensors as recorded by way of the histogram with both the shadows and highlights clipping, or up against their respective ends of the scale. Below are shots from the same setup, but this time at base ISO. We’ll start with the original files which were identically exposed, based on spot metering off of the grey card to establish midtone.
First, the original shots with their histograms:
And after increasing the exposure and compressing the highlight info by identical values, I was able to get these:
I was able to push the shadows up by a stop and bring the highlights back by way of the highlight recovery slider to retain all of the highlight info while gaining quite a bit on the lower end of the register.
Below is the GX7’s adjusted histogram overlaid on top of the GM1’s adjusted histogram. The peaks and valleys are similar, and both shots were captured and adjusted identically, but there are a few different values, primarily in the mid tones. Not a huge difference, and not one that is particularly perceptible, but interesting to see the histograms nonetheless.
It’s difficult to compare one camera to another in a dynamic scene outdoors, and I had a couple setups trying to compare dynamic range, but feel it’s best to keep it to the more boring indoor, controlled static setups.
One thing I have been noticing though is that under (close to) identical shots outdoors, with the same exact exposure and in camera settings, same lens, etc, the GM1 tends to push info to the left, seemingly underexposing by comparison. See below, and again click on any to see much larger versions.
To see it a little better, here is a shot of the two unadjusted shots side by side. Click to see the full sized screen shot.
The GM1 has compacted the entire range to the left comparatively according to the unadjusted file’s histograms, and the green channel on the GX7 is obviously pushed further into the mid tone range comparatively.
Again, I know it’s impossible to get identical shots when there are things like wind, clouds, reflection, atmospheric differences, ghosts and stuff around, but these two shots were taken within 20 seconds of each other or so, from a tripod in open shade on a sunny, cloudless day. This could also be the way that RAW files are converted via the default profiles in Aperture too, so really we’re talking about many moving parts, but for me, I find it interesting that two cameras with the same sensor and same processor, being used with the same exact lens at the same exact exposure settings seem to “see” things a little differently. Firmware differences perhaps, firmware. Interesting is all.
Detail and resolution?
Well, looking at fine detail in two identical shots, I’d say their ability to resolve fine detail is near identical too. Click the 100% crops to see full sized. First, the GM1 followed by the GX7 file captured identically.
As far as any other differences, I’ve not been able to see them. Both sensors, while tending to see color a little differently when set to the same white balance preset under identical settings, and the exposure shift to the left on the part of the GM1, all when shooting RAW files, seem to resolve similarly and perform near identically as far as noise goes to my eye.
So, I guess in conclusion, I have to assume that a 4 point DXO difference is a result of how these cameras seem to be calibrated in concert with their processors when it comes to their performance at differing ISOs. The noise seems a little different as does the overall resolution at these various ISO settings, but at near negligible levels as far as I’m concerned.
I’d say that while they have very slight differences in image quality, these two cameras are on the same level and I wouldn’t choose which camera based on these slight differences, but rather the other features each camera has versus the other. IQ aside, when looking at features they are vastly different machines, both providing a different set of pros and cons.
The GX7 is still my favorite micro 4/3 camera to date. It has all the features I want while providing a comfortable ergonomic grip, faster than average max sync speed in 1/320 (the GM1 has a very strange 1/50sec) and small, but not tiny profile. The GM1 goes the other way in that it has removed things like the IBIS, tilting screen, viewfinder, grip and extra control and function buttons, but offers a tiny package with tricks of its own like a 1/16,000 second max shutter speed (vs a 1/8,000 max for the GX7) and the ability to fit this level of image quality into small pockets.
I think that any differences in IQ would go largely unnoticed without pixel peeping (and have in real world shooting for me). As for the exposure shift, that is very easily correctible in post or by way of exposure compensation in camera and I am very happy with both of these cameras and what they provide me in my compact system.
You can see all the gear used in this review via the links below.
You may have noticed the little add on grip for the GM1 in the images at the beginning of the article. I was contemplating a variety of different add on accessories, and decided to stay as small and light as possible. I chose the very inexpensive FlipBac G2 grip ($10 at B&H here) and have been very happy with it. With a larger lens like the Oly 75 or PL25, it is a tight fit between lens and finger, but then anything that might fit on the camera there will suffer from the same issue. All in all $10 was a steal for something that has made this little camera ergonomically superior to what it was prior.
If interested, here are a couple other comparison and review articles involving the GX7, GM1 and the EM5:
Stay tuned as I’m in the process of reviewing a “legacy” FD 55mm f/1.2 lens after I’ve switched the original FD mount for an EF mount and have adapted it to the micro 4/3 cams as well. Not to spoil anything, but it’s a pretty special lens and on the GX7 with focus peaking and the tilting LCD has been providing me with a cool, waist level finder type shooting experience that has been a lot of fun. It is a very fun time to be a photographer.
Thanks as always for the read, feel free to fire off any questions via the comments or through the Contact Page and I’ll be happy to try and help answer them. Sign up to receive new posts via email at the top right of this page, and if you’ve not yet, find me on Facebook, Twitter and Flickr and let’s keep this party going.
Cheers and happy shooting,