5DMk2 vs OM-D High ISO Test

A few weeks ago I posted a comparison of the 5D MK2 to the OM-D in good light.  You can find that thread here: http://forums.dpreview.com/forums/post/50428206, but to save you some reading, basically there was no meaningful difference that would be visible on any print.  That may be expected of any modern sensor camera, but what about at high ISO?  I decided to run a test at ISO 1600 to 6400 and see how the 2 camera systems compare.

There are 2 ways to do this comparison.  The things that are often important are depth-of-field and shutter speed.  Since the DoF of m4/3 is 2 stops greater than a FF sensor, the DoF of the OM-D at f5.6 is the same as the 5D at f11, then for equal DoF and shutter speed, then the ISO for the OM-D can be 2 stops less than the 5D.  So, in low light, for a given shutter speed and DoF, the OM-D can be shot at ISO 1600, while the 5D will have to use ISO 6400.

There are also times where you have to open the lens as far as it goes; the subject is a bit in the distance so DoF will be OK and the light is so dim that you have to maximize ISO.  In those instances, the performance of one camera at a specific ISO can be realistically compared to that of the other camera at the same ISO, because you'd use that high ISO in the field to keep shutter speed as short as possible.  So, I ran this comparison both ways: 1:  5D Mk2 at ISO 6400 vs OM-D at ISO 1600, to simulate the situation where you want the same shutter speed and DoF; and then with the images shot at the same ISO on both cameras for the 2nd scenario.

Here's my setup:

5D Mk2.  Manual Exposure.  Manual ISO.  Auto WB.  RAW.  24-105 f/4 L lens set to 50mm and f/11. Auto Focus using the center spot.  Processed in ACR with default settings.  The 5D2 image was downsized in PS CS6 to be the same width as the OM-D image.

OM-D.  Manual Exposure. Manual ISO.  Auto WB.  RAW.  Panasonic 12-35 f/2.8 set to 25mm and f/5.6. Auto Focus using the center spot.  Processed in ACR with default settings

Both cameras were mounted on the same tripod and remotely triggered.

For both cameras I shot a sequence of exposures at 1/3 shot increments and picked 2 that appeared to be the right exposure and equal densities.  The EVs may have been different, but I wasn't testing metering accuracy.  As it turned out, the EVs used for both cameras were the very close.

For a location, I wanted something where the light wouldn't be changing, so I picked a room in an old Civil War era fort that is unevenly lit by incandescent lamps.

I used auto-focus with both cameras and both cameras had no problem focusing.

Here's the overall scene, with the OM-D first:



OM-D.  ISO 1600.  Auto WB.  f/5.6

Now the 5D Mk2 Image:

5D Mk2.  ISO 1600 Auto WB f/11

The first thing I noticed when processing the photos is that the OM-D, set to AWB absolutely nailed the color balance.  It's perfect.  The Canon is WAY off.  Now, some may like the warmer tungsten balance produced by the Canon, but if I want that, I'll add it myself.  I'd rather the camera get the colors accurately, and the OM-D was absolutely spot on.  Because I didn't want to have color balance as a major issue when comparing images, I re-processed the Canon exposures in ACR using tungsten color balance.  Here's the re-processed Canon image:

5D Mk2 with ACR color balance set to tungsten.

It's still warmer than the OM-D image, but a lot closer than the original and, for this purpose, close enough. Both of the above images were shot at ISO 1600.

Both cameras were focused on the barrel at the left end of the middle row.  (By the way, they're empty).  Let's take a look at a tight crop from the above scene to compare IQ at the different ISOs.  First, in the case where the exposure value (EV) is the same, comparing the 5D at ISO 6400 to the OM-D at ISO 1600.  Here's the crop from the OM-D, below:

OM-D  ISO 1600  f/5.6

5D Mk2  ISO 6400 f/11

Both of the above images have been exposed to have the same EV and DoF.  There's no real comparison here:  The OM-D produced a much more noise-free image.  If you enlarge and look at the printing on the sign behind the barrel, you'll see that it's quite clear on the OM-D shot, and that noise makes it more difficult to read on the 5D image.

Therefore, in low light, when shooting so as to maintain a common depth of field and exposure value, the OM-D's 2 stop greater DoF enables you to shoot at a 2-stop lower ISO and the resulting image displays significantly less noise.  The DR of the OM-D image also appears to be better, but it may simply that the darker areas of the 5D image are obscured by noise.

The next comparison is where I compare the 2 cameras at the same ISO.  In this case, the OM-D would enable you to shoot at a higher shutter speed, but that isn't a factor in this comparison because the cameras were on a sturdy tripod and the barrels weren't moving.  So, here are both crops at ISO 1600. First, the OM-D image

OM-D ISO 1600

Now, the 5D image at ISO 1600

5D Mk2 ISO 1600

I'm going to have to score this one slightly in favor of the 5D.  There isn't a lot of difference in the noise, but what difference there is favors the 5D.  The 5D is also maintaining somewhat better shadow detail (notice the brick pattern in the shadow of the sign on the left).  If I took the time to color balance these 2 shots so that they're the same, and perhaps adjust EV slightly, I doubt that anyone could tell the difference between large prints made from both images.

Now, here's the same crops, but at ISO 3200.  First, the OM-D image:

OM-D ISO 3200

Now the 5D image:

5D Mk2 ISO 3200

There's a larger difference here in favor of the 5D.  The wood racks still look quite smooth in the 5D image, whereas there's quite a bit of noise noticeable, especially in darker areas, on the OM-D.  The printing on the sign behind the barrel still looks clearer on the OM-D image, so maybe there was a bit of a focus shift, but noise certainly favors the 5D.

Now, same thing at ISO 6400:  Again, the OM-D first:

OM-D ISO 6400

Now the 5D at ISO 6400:

5D Mk2 ISO 6400

At ISO 6400 is where the wheels are starting to come off the wagon for the OM-D.  While the image may be useable for small prints, the noise pattern is much more noticeable, and coarse, than that of the 5D at this ISO.

To my eye, the difference in noise is slightly less than one stop between the 2 cameras.  The OM-D at ISO 1600 is very slightly better than the 5D at ISO 3200.  I think that this is a remarkable performance for so small a sensor.  Then, consider the first scenario that I presented, above, which for me is a realistic one.  That is, I'm shooting the OM-D at f/5.6 and would have to shoot the 5D at f/11 to obtain the same DoF.  Therefore, I have the ability of shooting the OM-D at a 2 stop lower ISO and, since there's about a 1 stop difference in ISO performance, I gain a full stop of lower noise in that scenario.

My conclusion is that while the 5D MK2 is quite useable up to, and including ISO 6400, and I've done that many times with good results, the highest ISO where I feel that OM-D provides a good IQ is 3200.  I haven't tried to apply any NR to any of the images, and doing so might extend the useable ISO range of the OM-D, but that's another test.

My shooting typically doesn't demand a very shallow DoF.  I'm not often shooting portraits. So, the scenario that I presented first, where I compared the 5D at ISO 6400 to the OM-D at ISO 1600 is the most realistic for most of my shooting.  In that case, the OM-D will allow me to obtain a higher quality image in low light.  I'm impressed.

Nice work. Thanks.

I have found my D4 to be 2 stops better than my OMD.

-Bill

Fashion Meets Fighting

http://www.hatakeyamagallery.com/april-2012-bill-fulcher/

I am intrigued to see several posts on this forum comparing M43 with Full frame and showing very good performance from the latest M43 cams.

I have not done a direct comparison yet but, looking at a few files, my 'gut' says that there's about 1-2 stops between the OM-D and D700 above ISO 1600. For me, that's a pretty impressive performance from the m4/3 sensor. I think that this echoes your findings, though I haven't seen any files from the D4. How would you rate the D4 vs. the D700/D3?

--
Dave Sanders

This is the aspect that typical FF fanboys purely miss. When I shot full-frame it was a constant struggle to produce adequate depth of field without getting into heavy diffraction. And I'm just not into high ISO, so the shutter speeds required meant tripod time. Changes in sensor size move the DOF curve around, but enclose no more area underneath.

This is understandable aside from pixel-pitch as I believe 6400 is the start of the OMD's 'expanded' ISO range. 6400 is not on the 5DII. The same thing is noticeable when comparing the OMD's output to that of its APS-C Sony-sensor cousins in the K5 and D7000.

I can only stomach 1600 on the OMD, but hey, I can stand 1600. I capped things at 800 (strictly) on my DSLRs, both of which carried larger sensors.

And when you handhold in low light, the OMD is just dynamite when used with a short prime lens like the 14 or 20 Panasonic. I can usually stay at 800 in rather dank conditions due to an impressive 'totally sharp' hit rate at 1/8s or even less. Even the monopod stays in the car these days.

Yes, I'm not always shooting shallow DOF portraits so there are certainly times when more DOF is a boon. I like that I can get reasonable DOF with my 14/2.5 in a variety of situations. I'm still not convinced that this whole thing is an either/or situation. It feels more like an AND situation to me...I shoot Nikon AND Olympus m4/3 and both tools have different uses. It's a great time to be a photographer, if you ask me.

So far I agree as far as colour photography is concerned, but I like the OM-D at 3200 & 6400 in B & W. The noise has great grain-like texture and a detail softening reminiscent of Tri-X or Delta 3200, films which I really enjoyed for a certain 'look'.

Again, I think I can quite easily justify two systems at the moment: FX for  high mpix, high ISO and shallow DOF, m4/3 for size, weight, IBIS and increased DOF.

--
Dave Sanders

Thanks for the comparo.

I'm pretty happy with the OMD through iso 5000 in raw, but it all depends on what you want to use it for. I use it for family shots mostly, meant for not-so-large prints and photobooks. Outdoors, of course, I don't need such high iso.

--
John Krumm
Juneau, AK

Excellent test, and your conclusion is spot-on.  The EM5 has a more efficient sensor than the 5D2, so it will have an advantage for Equivalent photos (same DOF and shutter speed).  Combine that with the outstanding sensor IS of the EM5 that works with all lenses, and, well, the EM5 is the clear winner here.

Thank you, so much, for the work, effort and care taken into this excellent comparison, which is so useful to remind many (forum members) that photography is not always about shallow DOF...

Best regards,
Pedro

Thank you.  For many who shoot portraits, shallow DoF is a requirement and I understand that.  It's just not the type of photography that I enjoy.  So, m4/3 works well for me.

Is it really a direct comparison?

Does not seem so.

Very thoroughly planned and well-executed test.

Someone already mentioned it in this thread, but I think it's worth repeating that ISO 5000 is the last ISO on the E-M5 that is not 'extended.' For that reason, you might be interested in comparing the performance of the two cameras at ISO 5000 (for your own purposes, of course).

Also, according to measurements and calculations performed by some of the forum's more technically-minded (I'm thinking mostly of Anders W's calculations based on data from DXO, as well as from Great Bustard's measurements on his Digital Camera Sensor Data site  [see here for OM-D E-M5 Sensor Data]), it may make sense to set 800 or 1600 as your upper ISO limit, then push / dodge / brighten the photos as necessary in post processing. Depending on the RAW processor you're using, this approach may yield better-looking images than using on higher ISOs in the camera.

Might be another technique you'd like to test (for your own purposes), since you seem to be pretty interested in maximizing image quality.

tex

Thank you for the comparison as I am considering the Canon 5dmkii or the Olympus OMD, Which I do find a bit small, so I was pondering the Panasonic GH3 when it eventually appears here in OZ. There is not as much in it as I thought there would be.

Thank you for a very interesting and well done comparison. When my 40D went drowned and I was forced to replace it I first considered a 5DII. But as i could see on test shots the OM-D wasn't far behind in IQ (and much better than my 40D) and I had an old faiblesse for Olympus since my Pen FT and OM cameras, I decided to get one. Obviously not the wrong decision. I'm still amazed how good this little camera is. I can't really get it into me. It's good to see tests and comparisons like yours to confirming that I 'm not totally crazy using that little beast instead of a bigger "real" camera.:-)

The Canon 5D Mk 2, not one of the best performing FF camera for low-light/high ISO (see D800, D600, D3s and D4 which are almost a full 2 EV better in low-light/high ISO performance) is over 1 EV better than the EM5 in low-light (See link below).

http://www.dxomark.com/index.php/Cameras/Compare-Camera-Sensors/Compare-cameras-side-by-side/(appareil1)/793%7C0/(brand)/Olympus/(appareil2)/483%7C0/(brand2)/Canon

Canon 5D Mk II Low-light (sports) score 1815 ISO

Olympus EM-5 Low-light (sports) score    826 ISO

Nikon D800  Low-light (sports) score      2853 ISO

Nikon D3s    Low-light (sports) score      3253 ISO

Pentax K5 II Low-light (sports) score      1235 ISO

What you also need to remember is the 5D2 produces high resolution files, so normalizing both camera's files to 16 mp will show even less apparent noise in the 5D2 files.

I'm always puzzled by the EM-5 to FF comparisons since the EM-5 doesn't even get close to APS-C cameras like the D7000, K5 or Fuji X-Pro1 for low-light/high ISO performance.  Why the need to skip a format?  Is it to show that the EM-5 can play with the big boys?  Well, no offense but it can't.  At least not for high ISO, DR and color depth (again see the D800 on DxOMark Sensor Ratings).

Nice comparison, but strange conclusions re the 5D2 and EM-5 low-light abilities (sorry to be spoil the party. Yikes who invited Markus?).

All the best, Markus

1) read the explanations !

2) he shot both 1600 and 6400 ISOs and same ISOs

This is a very well made test, you just need to read the whole thing !

To the original poster : thank you for sharing this double and interesting test.

--
rrr_hhh

Of course you'right, but I still find it amazing that this kind of comparisons are possible to do without the not too old FF totally out-classing the newer m4/3. One stop difference to 5DII is not very much. Remember 5DII is still an actual camera and not long ago highly reviewed. But technology develops on all platforms and the last year we've seen quite a few new cameras with exciting performance pushing the borders forward. Given the fact that OM-D, D 800 and K5 obviously share sensor technology, the differences you list are exactly what you could expect regarding differences in sensor size. My conclusion, supported by the comparison, is that if you compromise only a little, you are well off with an OM-D compared to a 5DII, which was among the best you could get, all categories, only two years ago or so.

Well you've just seen the evidence and clearly, for all intents and purposes, and in certain scenario's, it clearly can. In some scenario's it's actually better and vice versa.

The conclusions are what you've just seen, I'm not sure what you missed?

--
Any problem on earth can be solved by a well aimed Pomegranate...
Tony
http://the-random-photographer.blogspot.com/

People test the cameras they own ! They can't test the ones they don't have. And photo enthusiasts are more likely to own a FF than the casual Canon 600D owners, they are more likely to get a second system too : they are the ones dragging around heavier bodies, the ones who want a smaller take everywhere body.

Once they get it, it is a normal reaction to test both bodies and see how much more they can get from the heavier body and whether that matters for them. Personnally, I found the test very well made and usefull for me? I own a 5d Mk I and decided not to update to the 5D3 (the fact it became heavier including improvements I don't need, made me angry). I'm still hesitating whether to upgrade to the lighter 6D : will I use it enough to justify the update ? So in this context, the test and the following discussions are very interesting. It has nothing to do with fanboyism.

Also, knowing that Nikon since a few years is offering better sensors than Canon doesn't help those who are heavily invested in Canon glass : you don't swap systems that easily. The test is useful for all canon users.

The conclusion is valid for who ever value deeper DOF rather than sallow DOF : the shallow DOF race is no better than the higher pixels number : each has to know when he/she needs it.

Just an ironic development : the Olympus marketing wanted to attract P&S upgraders to MFT.. But in fact they are attracting DSLRs downgraders

--
rrr_hhh

[snip]

Nothing in the original poster's conclusions is contradicted by any of the information you've provided here. Nothing. Zero, nada, niente, nichts und nocheinmal überhaupt nichts. It appears that your FF-fanboyism is negatively affecting your reading comprehension skills.

Well, you didn't spoil the party with data or evidence that refutes the original poster's conclusions. Rather, you spoiled it with your anti social personality, yet again.

I'd be surprised to learn that anyone invites Markus anywhere, except maybe his family, begrudgingly, out of a feeling of familiar duty. But that never stops Markus from showing up wherever he pleases.