E-M5 or GH3? You tell me

Now that we're starting to get some comparable studio test shots taken with the GH3, it's interesting to have a close look at the output from Panasonic's flagship m43 model in comparison with output from the E-M5, Olympus' flagship model.

Several posters on this board have expressed their preferences for the image quality output by one or the other camera, often citing things like, and I paraphrase, 'very subtle differences', 'smoother / more gradual tonal gradation' and 'more of a 3-d feel'.

You'll see I'm trying to be neutral here. I'm not identifying any specific posters, and I'm also not saying which cameras I have seen the above, paraphrased statements applied to.

Without further adieu, here are two images from image-resource.com - one from their review of the E-M5, the other from their preview of the GH3. Have a look at both, and see whether you can tell which is which, and whether you find that one, in your subjective opinion, looks better than the other (in terms of 'pure' image quality, alone).

Note: In order to keep this interesting, I'd ask you to state your opinions in reply bodies, and not in the subject line (so as to not influence the opinions of other viewers). Also, I'd ask you to respond with an opinion before reading anyone else's response, again, to remove unnecessary bias from this exercise :-)

Full Size Version of image #1

Full Size Version of image #2

Some info about how I prepared these images:

I took two RAW files, one taken with the E-M5 and one with the GH3, both of the same studio scene and taken with the exact same lens, exposure, f-number and ISO settings.

I converted both files to DNG using the Adobe DNG Converter.

I opened both files in RawTherapee 4.0.8.10 and made the following changes:

1. I set the Postprocessing Profiles of both files to 'neutral-UnchangedTransforms'
   
2. I set the Color Management Input Profile of both to 'none'
3. I equalized the white balance settings in both images
4. I then adjusted Exposure Compensation on the E-M5 file by plus 0.58 EV
5. I cropped both images so they would have the exact same dimensions

I made changes #s one through three in order to get as close to the actual RAW data as I knew how, discarding any instructions from the camera to the RAW processor regarding levels adjustments, color and white balance settings.

I made change #4 in order to (as closely as quickly possible) match up the brightness of both images, since change #1 made the E-M5 image darker than the GH3 image.
I saved both files as 100 quality JPEGs from RawTherapee and uploaded them to my gallery.
My opinions will follow in a reply to this post.
Have fun!
--tex

I don't see any real difference. I'd go with either picture.

--
Inkheart

I've gone past the correction window of 15 minutes now, so I have to add this important correction as a reply here:

I swapped the links to the full-sized files on accident. The link under the first picture in the original post actually goes to the full-size version of the second picture, and vice-versa.

As you can probably deduce from that mistake, I'm having a VERY difficult time telling these two images apart, especially if I'm not looking at them zoomed 100%

Sorry for the confusion. Again, the links to the full-size images are swapped in the original post. Please keep that in mind when coming to your own conclusions.

tex

I think that the {RAW}  image quality of both cameras is similar enough to have some assume the same sensor. So , the differentiating  factors between each camera and what one would chose as “the best” is very much down to one’s own personal usage and opinions. I very much like the feature set of the E-M5 but for me the ergonomics {button placement, size, menus etc.} are all wrong. I am keen on video making the GH3 a better choice for me and I very much prefer a fully articulated LCD. Combine this with the fact that I was able to buy my GH3 bodies with the limited 30% discount offer from Lumix Lifestyle in the UK which made them cheaper than the E-M5 and it was a no brainer.

Jim

Since I know which file is which ( or do I? ), I won't yet give the answer, but I will offer a few observations I have, after comparing these two files pretty closely.

---

The E-M5 file is has noticeably finer definition than the GH3 file

This could be down to a focusing issue in the GH3 photo, but when I view at 100%, I'm unable to find any part of the GH3 photo that is as finely defined as most of the E-M5 file, and since both photos were made with the same lens at f/8.0, it may very well be that differences in fine definition are down to the cameras.

optyczne.pl , in their recent review of the GH3, mentioned that the GH3's output is noticeably less well-defined than that of the E-M5, likely due (at least in part) to the fact that the E-M5 has a weaker anti-aliasing filter than the GH3.

Question: Is it possible that a file with finer definition might appear to have harsher gradation between some tones, since the transitions between two tones will, indeed, be more finely defined?

---

The GH3 RAW appears to have clipped highlights, the E-M5 RAW doesn't

Note, I say 'appears' here, because I'm going off of what I see in the histograms of both files, as shown by RawTherapee. Both of the sample photos were taken at +1.0EV, and DxOMark indicates that their 'measured ISO' for both cameras is divergent in such a way that the E-M5 RAW files tend to underexpose compared to the GH3 RAW files.

This may not be interesting, except as a confirmation that the GH3 exposes further to the right of the E-M5, per default.

---

The RAW histograms look very similar (to my untrained eyes)

The E-M5 histogram is pushed to the left some in comparison to the GH3 (after setting the Post Processing Profile of both in RawTherapee to 'neutral-UnchangedTransforms'). Unfortunately, the scene framing does not match with 100% accuracy, so it's hard to tell which slight histogram differences could be due to variable framing, rather than to actual sensor or in-camera processing differences.

---

The E-M5 image appears warmer (to my untrained eyes)

Even after resetting everything I could in RawTherapee and matching up the white balance settings in both files, the E-M5 file seems to have a slightly warmer or yellower tone to it than the GH3 file.

---

In Closing

Does one look better to me than the other. Well, that's not really a fair question, since I am biased toward one of the cameras (I own an E-M5 and am very happy with it), and I know which file was taken by which camera.

The most striking and undeniable difference, to me, is the comparable lack of fine definition in the GH3 file. I prefer to have the extra definition that the E-M5 file offers, since I can always remove or 'blunt' that in post-processing, but I can't add it back where it doesn't exist from the start.

To me, this would suggest that the E-M5 is the more capable camera, from an image-quality standpoint, since it is able (apparently) to deliver files with a higher definition than the GH3, and it is also able to deliver files with a 'degraded' definition that matches the definition provided by the GH3.

One could argue that the stronger AA filter of the GH3 brings its own advantages, but I've taken almost 13,000 photos with my E-M5, and haven't had more than a handful that showed ill-effects that may have been made worse because of the weak AA filter. Actually, exactly one image comes to my mind out of 13,000, where I ended up with visible moire that was easily removed in LightRoom.

So, for me, I'll take the finer detail if the tradeoff is easily-removable moire in one out 13,000 images.

Other than that, I think these images are remarkably similar - very little substantive difference to my eyes.

tex

I agree that the IQ is so close as to be practically a wash. What I find sort of funny is that I've seen a number of GH3 owners claim that the IQ of the GH3 is better than that of the E-M5.

On the other hand, I don't recall having seen any E-M5 owners claim that the E-M5 IQ is noticeably better than that of the GH3, and my close comparison backs that up for me, with the one exception of finer detail in the E-M5 file (fully unsurprising due to the E-M5's weaker AA filter).

As you know, I prefer the featureset of the E-M5, but I agree with you that the choice between the two cameras should really be based on the respective featuresets, since, in my personal opinion, any perceived substantive differences in image quality (w/r/t RAW files, at least) is more in the head of the one doing the perceiving, rather than in objective reality.

tex

I'm comparing these two images in my iPad screen. I find the second image having a bit more detail, especially in the shadow(model's arm).

Also the white shirt seems whiter on the second image.

--
My nickel, since the penny is being discontinued...
Jeff.

Based on what you said, I'd say the first pic is GH3, and the second is OM-D.

--
Inkheart

--
________________
www.acahaya.com

Moire is a much more interesting topic for video than for stills. Since the Panasonics are real video stars it might just be better this way.

Apart from that those two cameras are so different from the handling perspective. This, video and maybe IBIS will surely be the more defining factor to decide between those two. I guess therefore this thread is great to show that there is not too much between them regarding image quality.

You can disregard the previous post now - Aleo contacted me and was nice enough to offer to correct the links for me.

Seriously, I lost track of which image what which so many times during this exercise

Unless I'm peeping at 100%, it's very hard for me to tell them apart on-the-fly with any consistency.

Thanks Aleo!

tex

Makes sense in view of the DxO data on color response. While the two cameras have very similar color response on top of all other similarities on the sensor side, the E-M5's red and blue channels are both stronger (relative to the dominant green channel) than the corresponding channels on the GH3. On the blue side, the difference is trifling but on the red side, it is probably big enough to be noticeable, which would explain why the E-M5 image appears warmer unless steps are taken in PP to balance out the difference (e.g., by means of a profile suited to the camera).

The two cameras well may have different CFAs even if the silicon is the same, which would explain the slight difference in color response.

That's clear, and I've said nothing to the contrary

Agreed, 100%. I actively disagree with those who claim that one of these cameras offers noticeably better stills quality than the other, with the exception, of the fine detail advantage (if, indeed, the two sample files are indicative) delivered by the E-M5.

The decision comes down to other factors, completely, in my mind.

tex

You are welcome. I also like to have the threads well done with no errors.

---

--
Simplicity is the ultimate sophistication.
Pure logic is the ruin of the spirit.
If plan A doesn't work, the alphabet has 25 more letters, keep calm.
Imagination is more important than knowledge.
God is the tangential point between zero and infinity.
Aleo Photo Site

I have noticed that too yesterday when comparing at Image Resource the OMD to the GH3, maybe Olympus use a different AA filter or they apply less NR, the GH3 seems to apply a bit more of NR.

Anyway both cameras did very well, IQ is great.

---

--
Simplicity is the ultimate sophistication.
Pure logic is the ruin of the spirit.
If plan A doesn't work, the alphabet has 25 more letters, keep calm.
Imagination is more important than knowledge.
God is the tangential point between zero and infinity.
Aleo Photo Site

Yes, I'm pretty sure the GH3 uses a stronger AA filter in order to combat moire in video. I'd call that a value-neutral decision. If video is important to you, you'll want an AA filter that cuts down on moire. If video is less important to you, you'd probably prefer a weaker AA filter and the improved fine definition that comes along with it (in stills).

Agreed - I think they're super close. I'm glad m43 has two flagship models (plus Oly's other 2 models with this sensor) that offer such outstanding image quality while catering to different niche markets. It makes the system, as a whole, stronger.

I think if you're trying to choose between an E-M5 and a GH3, stills image quality shouldn't be your #1 decision criteria. These two cameras are just too close

tex

Your processed images have been re-encoded by the DPReview display system using 2x2 Chroma Sub-sampling and only a 90% Quality Factor. They may have also been down-sampled by the DPReview system (which, from my testing may be Lanczos-3) over an above any (possible) down-sampling that you may have previously performed using RAW Therapee (you must like the older version better, and not the more recently released Version 4.0.9.50 ?).

Out of curiosity, which of the de-mosaicing options did you use (for both RAWs, I presume) ?

I can see (at 700% view) that the first image displayed appears that it may well have more re-sampling artifacts than the second image displayed. Could this be related to your cropping of one of the images in order to try to disguise the differences in Field of View between the test-shots ?

The first displayed image also had noticably lower amounts of Red relative to Green in the RGB histogram of your DPReview Gallery images provided (most prominent in the mid-tones comprising the bulk if the image-data). So (I suppose) that means that the first displayed image is the GH3 image ?

To make any kind of observations or resulting statements about "resolution", "definition", etc, you really should upload JPGs that have not been down-sampled at (or, if one of them has been re-sampled slightly to disguise differences in Field of View, consider that to be a source of distortions, particularly if down-sampling, as opposed to up-sampling, is performed) ...

... and enable the downloading of "Originals" from your DPReview Image Gallery (in DPReview profile Settings section).

Observations or judgements based upon DPReview display system degradations are questionable.

Such uploaded JPGs would likely exceed the 20 Mbyte limit (but if you use 2x2 Chroma Sub-sampling when encoding the JPGs, the file size can be brought under the 20 Mbyte limit).

.

No particular investment in the outcome - but I think that the method of presentation is inadequate.

I run a full frame system plus mft and had quite some others over the years. To me there is no need to struggle for a 1, 2 or 5% difference, especially if there are systems in the marked delivering MUCH more. Any difference below 20% is to me rather secondary for all practical means.

As a passionate ultrawide fan I had many discussions in the Canon forums and between Canon and mft as well to find out who is leading by how much in this field. I think this is rather in vain since buying a D800e with a 14-24 simply is so much better for that application than both.

In the end all or choices are compromises one way or the other and it helps a lot to understand how big those compromises are. There should be no reason to fight over it.

So again, nicely done.

Not as funny as some of the E-M5 owners who claimed the E-M5 is as good as the D800 oops I forgot you don't find that as funny as  I do

I have not been on the forum much recently but other than possibly one rather keen GH3 fan I cant say I have seen many claim the GH3 is superior, and certainly nothing compared to the hyperbole that surrounds the E-M5. I see for some reason that imaging resource does not provide a GH3 sample with NR at 0 .I do not use RAWtherape but looking at  the RAW studio shot  files from IMAGING RESOURCE  even with the NR1 option compared to the NR0 on the E-M5 the differences if any are tiny and certainly not observable at 100% , even in areas of very fine detail such as the fabric 200% crop below differences are very small and could be down to a number of factors . RAW files all  NR turned off in ACR default sharpening

E-M5GH3 BASE

If the E-M5 had been just a wee bit larger with a better [ for me } arrangement of buttons and a fully articulated LCD I would more than likely be using the E-M5 as any gains in the GH3 video quality are matched by the excellent IS system of the E-M5 . In fact I have bought the Olympus 50-200 SWD for use on a potential future O-MD with full AF support for FT lenses. Differences this small will have little if any real life impact on final output and as you say what camera you chose is solely down to ones own subjective "feeling" . The GH3 is pretty pricey in the UK and had I not been directed to the special discount offers , I am not honestly sure if I would have opted for the GH3.

Jim

Ah, but I do - that's yet another ridiculous comparison, but one that I haven't seen made in a very long time. Months, actually. Although you still harp on it every chance you get. I wonder whether you'll still be harping on it a decade after the last misinformed E-M5 fan makes that obviously silly assertion. I wouldn't be surprised, the slightest.

Heck, I've even seen you harping on this in OTHER forums at DPReview in the past couple of weeks. Care to provide a link to the last instance of someone saying the IQ of the E-M5 is equal to that of the D800 that you've seen here?

And, lastly, that's a bit of a red herring, really. I'm interested in comparing the GH3 to the E-M5 here. If you'd like to compare the E-M5 to the D800, I'd invite you to do your own work and start your own thread - feel free!

Surrounds? Present-tense?  OK 

I know of at least three GH3 owners who've claimed in the past week or so on this forum that the GH3 ahs better stills image quality than the E-M5 (one of them just today). I know of exactly 0 E-M5 owners who've made similar statements in the past week on this forum.

Learn to let the past go, Jim. It gets old hearing someone complain about wrongs they heard months ago. And it's an immature tactic, anyway. I'd prefer to out all incorrect claims, whether they be about the E-M5 or the GH3.

Wouldn't you? Or do you just prefer to correct incorrect claims about the E-M5 and let incorrect claims about the GH3 (and other Panasonic cameras) stand unchallenged? Certainly seems that way. Two wrongs don't make a right

Sorry, I have no interest in comparing the JPEG files here. If you'd like to do that, feel free to start your own thread. This one's talking about the differences between RAW files.

RAW Therapee is free to download and easy to install - give it a shot!

And on this we agree - the differences aren't in the stills image quality, I'd say. The differences are in the other features and attributes of each camera.

tex