5D mk II vs mk III real world comparison: portraits, wildlife, landscapes, night

[Tony Northrup]

I spent the last couple of days documenting a rather exhausting comparison of the 5D2 and 5D3. Here's the pixel-peeping video:

http://youtu.be/D3Ums46kKsE

That page has links to the JPG and RAW files (in a Lightroom catalog) so you can compare them yourself.

In a nutshell:

• Low-light Portraits: The 5D3's AF helps a bit, but even at ISO 25,600, the 5D2's image quality is at least as good as the 5D3's.
• Walking, running, and flying subjects: The 5D3 has much improved AF and a faster shutter, doubling or better the number of in-focus shots of moving subjects.
• Still wildlife: The AF and faster shutter produce a higher ratio of usable shots. No improvement in actual image quality.
• Night photography: Live View sensitivity is better, making it easier to frame pictures. AF is good enough to focus in dark situations where you'd have to manually focus with the 5D2. No improvement in image quality.
• Landscapes: The electronic level is nice, but if you're a landscape photographer, just buy a bubble level and save your money for a nice lens, because there's no improvement in image quality.
• Studio: No improvement in image quality, but small touches improve workflow: the RATE button, easier image deletion, easier AF selection, quicker image review.
• HDR: Expanded bracketing is important, but I'd just recommend getting a 5D2 and installing Magic Lantern. I can't imagine wanting to use the in-camera HDR feature.

The biggest shocker: No noticeable improvement in raw image quality, even at high ISOs. It's baffling to me considering how Canon (and just about every review I've seen) is bragging about the new sensor. The video walks you through my testing and analysis, and while it's non-scientific, I found the same results in every scenario.

Here are tight crops from the studio pictures (the second pictures are from the 5D3). In the video I zoom into 8:1 and find them both equally noisy. I've spent many hours fixing noisy 5D2 pictures that got rejected by the stock sites, and later accepted after I manually reduced the noise, so I'm rather good at spotting noise:

















I think ISO/noise wars are the new megapixel wars, and Canon couldn't manage to produce a better sensor than the 5D2's. Yet, they knew that image quality is the #1 selling point, so they couldn't market it with, "Same image quality with better AF and lots of tweaks." Instead of making improvements to the sensor hardware, they increased the noise reduction in JPEG processing and allowed two stops higher ISO--equivalent to pushing a 5D2's "H2". And software changes don't show up in raw files, and don't matter to those of us shooting raw (and at this price point, nobody should shoot JPG).

Based on my results, this statement (attributed to a Canon tech) seems to be a lie: "The 5D Mark III has a 22.3-megapixel sensor, redesigned for dramatically improved low-light performance".

Thoughts? What's your interpretation?

 

[Marcus Antonius]

Good to see confirmed what i was already suspecting.

Marcus

I spent the last couple of days documenting a rather exhausting comparison of the 5D2 and 5D3. Here's the pixel-peeping video:

http://youtu.be/D3Ums46kKsE

That page has links to the JPG and RAW files (in a Lightroom catalog) so you can compare them yourself.

In a nutshell:

• Low-light Portraits: The 5D3's AF helps a bit, but even at ISO 25,600, the 5D2's image quality is at least as good as the 5D3's.
• Walking, running, and flying subjects: The 5D3 has much improved AF and a faster shutter, doubling or better the number of in-focus shots of moving subjects.
• Still wildlife: The AF and faster shutter produce a higher ratio of usable shots. No improvement in actual image quality.
• Night photography: Live View sensitivity is better, making it easier to frame pictures. AF is good enough to focus in dark situations where you'd have to manually focus with the 5D2. No improvement in image quality.
• Landscapes: The electronic level is nice, but if you're a landscape photographer, just buy a bubble level and save your money for a nice lens, because there's no improvement in image quality.
• Studio: No improvement in image quality, but small touches improve workflow: the RATE button, easier image deletion, easier AF selection, quicker image review.
• HDR: Expanded bracketing is important, but I'd just recommend getting a 5D2 and installing Magic Lantern. I can't imagine wanting to use the in-camera HDR feature.

The biggest shocker: No noticeable improvement in raw image quality, even at high ISOs. It's baffling to me considering how Canon (and just about every review I've seen) is bragging about the new sensor. The video walks you through my testing and analysis, and while it's non-scientific, I found the same results in every scenario.

Here are tight crops from the studio pictures (the second pictures are from the 5D3). In the video I zoom into 8:1 and find them both equally noisy. I've spent many hours fixing noisy 5D2 pictures that got rejected by the stock sites, and later accepted after I manually reduced the noise, so I'm rather good at spotting noise:

















I think ISO/noise wars are the new megapixel wars, and Canon couldn't manage to produce a better sensor than the 5D2's. Yet, they knew that image quality is the #1 selling point, so they couldn't market it with, "Same image quality with better AF and lots of tweaks." Instead of making improvements to the sensor hardware, they increased the noise reduction in JPEG processing and allowed two stops higher ISO--equivalent to pushing a 5D2's "H2". And software changes don't show up in raw files, and don't matter to those of us shooting raw (and at this price point, nobody should shoot JPG).

Based on my results, this statement (attributed to a Canon tech) seems to be a lie: "The 5D Mark III has a 22.3-megapixel sensor, redesigned for dramatically improved low-light performance".

Thoughts? What's your interpretation?

--
Making a moment last forever

 

[jindrich]

Thanx for the comparison. It certainly confirms what I was fearing, that is, that's Canon's MARKETING Department and NOT the Engineering Department the one making progress.

Canon has apparently lost the edge and technological advantge it previously had and by playing very conservatively thinking it "will be enough", it has been outplayed by the competition.

It has happened with the recent C300 vs new Sony products in the video field and it's happening against Nikon in the Photo field.

I'm getting closer and closer everyday to get a D800/D4+lenses and forget about Canon for the next 5 years.

 

[Tony Northrup]

Good to see confirmed what i was already suspecting.

So, it's not just me. I'm happy with the 5D3 regardless, but this seems like it could become a class-action situation.

 

[Teosax]

Hello tony,
thanks for your review,

there is one thing though that i am not convinced about.

When you show the first comparison of the eye of chelsea, and state that at 800 the 5d2 has less noise
I think your test, in this case, is not fair at all
the 5d2 has had around one more stop of light that the mrk3

maybe they have the same exposure time, but nevertheless, the 5d3 is way underexposed compared to the 5d2,

I think the only way to compare noise in 2 different cameras is giving to the sensor the exact same amount of light
honestly i think the noise comparison does not show useful results

i think maybe you should or re shoot it, or just overexposed the 5d2 file until it matches the brightness of the 5d3, and at that point draw noise conclusions

matteo
ps. u got some killing shots in your website

I spent the last couple of days documenting a rather exhausting comparison of the 5D2 and 5D3. Here's the pixel-peeping video:

http://youtu.be/D3Ums46kKsE

That page has links to the JPG and RAW files (in a Lightroom catalog) so you can compare them yourself.

In a nutshell:

• Low-light Portraits: The 5D3's AF helps a bit, but even at ISO 25,600, the 5D2's image quality is at least as good as the 5D3's.
• Walking, running, and flying subjects: The 5D3 has much improved AF and a faster shutter, doubling or better the number of in-focus shots of moving subjects.
• Still wildlife: The AF and faster shutter produce a higher ratio of usable shots. No improvement in actual image quality.
• Night photography: Live View sensitivity is better, making it easier to frame pictures. AF is good enough to focus in dark situations where you'd have to manually focus with the 5D2. No improvement in image quality.
• Landscapes: The electronic level is nice, but if you're a landscape photographer, just buy a bubble level and save your money for a nice lens, because there's no improvement in image quality.
• Studio: No improvement in image quality, but small touches improve workflow: the RATE button, easier image deletion, easier AF selection, quicker image review.
• HDR: Expanded bracketing is important, but I'd just recommend getting a 5D2 and installing Magic Lantern. I can't imagine wanting to use the in-camera HDR feature.

The biggest shocker: No noticeable improvement in raw image quality, even at high ISOs. It's baffling to me considering how Canon (and just about every review I've seen) is bragging about the new sensor. The video walks you through my testing and analysis, and while it's non-scientific, I found the same results in every scenario.

Here are tight crops from the studio pictures (the second pictures are from the 5D3). In the video I zoom into 8:1 and find them both equally noisy. I've spent many hours fixing noisy 5D2 pictures that got rejected by the stock sites, and later accepted after I manually reduced the noise, so I'm rather good at spotting noise:

















I think ISO/noise wars are the new megapixel wars, and Canon couldn't manage to produce a better sensor than the 5D2's. Yet, they knew that image quality is the #1 selling point, so they couldn't market it with, "Same image quality with better AF and lots of tweaks." Instead of making improvements to the sensor hardware, they increased the noise reduction in JPEG processing and allowed two stops higher ISO--equivalent to pushing a 5D2's "H2". And software changes don't show up in raw files, and don't matter to those of us shooting raw (and at this price point, nobody should shoot JPG).

Based on my results, this statement (attributed to a Canon tech) seems to be a lie: "The 5D Mark III has a 22.3-megapixel sensor, redesigned for dramatically improved low-light performance".

Thoughts? What's your interpretation?

 

[altenae]

Canon has apparently lost the edge and technological advantge it previously had > and by playing very conservatively thinking it "will be enough", it has been > outplayed by the competition.

I Disagree !!!

Take your time and read all the older topics about the 5D Mark II

We ASKED FOR :

-The 5D Mark II sensor, because we all said it was so good
-Better weathersealing
-No plastic card door
-Less Shutterlag
-More FPS
-Better AF
-etc.

We said...it's about time Canon listened to their customers.
So CANON dit EXACTLY this !!!!

I really wonder how al this whining would be if the Nikon D800 was not released.
Also the 5D Mark III AF is outstanding (one of the best)

For me a superb AF is much more important then maybe 2 stops more DR !!!

If there is not much improvement over de 5D mark II then it will save you a lot of money.

Canon has not lost the edge, because of what ???

The 5D mark III is a superb DSLR with superb AF and is in the same class as the D800 or D4 and 1D Mark IV.

The AF of the 5D Mark III gives me more keepers of BIF then the 1D Mark IV

Canon lost the EDGE ----I don't think so.

--
Enjoy Photography
http://www.wildlife-photos.net

 

[Tony Northrup]

Hello tony,
thanks for your review,

there is one thing though that i am not convinced about.

When you show the first comparison of the eye of chelsea, and state that at 800 the 5d2 has less noise
I think your test, in this case, is not fair at all
the 5d2 has had around one more stop of light that the mrk3

maybe they have the same exposure time, but nevertheless, the 5d3 is way underexposed compared to the 5d2,

I think the only way to compare noise in 2 different cameras is giving to the sensor the exact same amount of light
honestly i think the noise comparison does not show useful results

i think maybe you should or re shoot it, or just overexposed the 5d2 file until it matches the brightness of the 5d3, and at that point draw noise conclusions

matteo
ps. u got some killing shots in your website

Thanks for the compliment, Matteo, and good point. I'll say I know there are plenty of lab tests out there, and duplicating that effort wasn't my goal.

The low-light portrait example at ISO 6400 was shot at the same shutter speed (1/90th at f/2.8), but the 5D3 looks a bit darker. Maybe the sun went behind a cloud, but don't throw the baby out with the bathwater.

While the 5D2 looked better in that example, it was just one of a suite of tests, and my overall conclusion is that the 5D2 and 5D3 image quality are equal. So, check out the studio shots, the songbirds, the landscapes, and the night photography. They all had very consistent lighting and no result showed better noise in the 5D3.

 

[cs hauser]

...but even at ISO 25,600, the 5D2's image quality is at least as good as the 5D3's.

No, it is not.

 

[Ah Pek]

Class action? Lol! You need to get a life. Raws only have incremental improvements. The improvements have been in processing software and jpeg engines. Ever compared raws between 1ds and 5d? 1d2 and 1d3?

Good to see confirmed what i was already suspecting.

So, it's not just me. I'm happy with the 5D3 regardless, but this seems like it could become a class-action situation.

--
'The majesticness of that duck is overwhelming!' - Bulbol

 

[Tony Northrup]

We ASKED FOR :

-The 5D Mark II sensor, because we all said it was so good
-Better weathersealing
-No plastic card door
-Less Shutterlag
-More FPS
-Better AF

Well, those things are all great, but I personally also asked for a sensor with lower noise so I wouldn't have to spend so much time cleaning up my stock images. Canon's marketing says the 5D3 has that. It doesn't, and it's important that buyers understand the truth.

It's a great camera, and much improved, but the 5D2 has basically the same image quality for $1300 cheaper.

 

[Tony Northrup]

...but even at ISO 25,600, the 5D2's image quality is at least as good as the 5D3's.

No, it is not.

OK, you have to give me something more to go on. I spent two days shooting six different types of photography with both cameras, compiled the results nicely, and provided everyone with the raw files to do their own analysis.

I would be thrilled to figure out how to get cleaner images with the 5D3, so please tell me how you're doing it, and show me the pictures.

 

[jindrich]

Canon has lost the edge. Yes.

For years, it was the most advanced manufacturer in the Photo field. This is not the case anymore.

Canon has FIXED everything that was wrong on the 5D2, a $3k camera which was essentially a 20D with a (great) FF sensor. Fixed issues at last, and it makes the 5D3 a great camera.

But besides the fixes, what's the revolutionary feature that takes the mark III beyond the mark II? The much higher price perphaps?

Nikon's recent D800 on the other hand, represents a VERY BOLD move. It's means Nikon has reduced the price of the D3X from $8k to $3k and it has even improve the camera further: MF-like detail and DR and Video as good or better as the best Canon.

And compared to the 5D3 it also brings a better and faster shutter, better X-Sync, inbuilt Flash Commander mode, HDMI out that's not crippled, 3D metering as the flagship, Crop modes,... etc.
And €500 cheaper.

Anyway back to the OP, Tony thanks again for your great comparison. Loved the rubbish high-iso video footage with INMACULATE sound, which ultimately sells the shot. This to remind us again how audio is just as important as video.

 

[oscarvdvelde]

I Disagree !!!
We ASKED FOR :

-The 5D Mark II sensor, because we all said it was so good
... ... ...
We said...it's about time Canon listened to their customers.
So CANON dit EXACTLY this !!!!

Yes, but then they forgot about updating the sensor to make it future-proof also for the next 3.5 years. So from 2008-2015 the Canon 5D-series image quality stood still.

Reasons mentioned for the desire to stay around 20 megapixels were increased dynamic range and higher ISO performance. Now it turns out that nothing was won (in RAW) by staying at 22 MP, while the other sensors made in the year 2012 showed that high ISO performance and two stops more exposure latency can be had, along with more resolution.

 

[Anant Dabholkar]

...but even at ISO 25,600, the 5D2's image quality is at least as good as the 5D3's.

No, it is not.

OK, you have to give me something more to go on. I spent two days shooting six different types of photography with both cameras, compiled the results nicely, and provided everyone with the raw files to do their own analysis.

I would be thrilled to figure out how to get cleaner images with the 5D3, so please tell me how you're doing it, and show me the pictures.

Comparison of specs between 5DII/III
http://www.imaging-resource.com/CAMDB/compare_results.php

Compare Images at different ISOs for the II and III

http://www.imaging-resource.com/IMCOMP/COMPS01.HTM

Upto about 800 ISO, they are almost equal, beyond that the III has an edge 1/2 stop or more.

 

[altenae]

I Disagree !!!
We ASKED FOR :

-The 5D Mark II sensor, because we all said it was so good
... ... ...
We said...it's about time Canon listened to their customers.
So CANON dit EXACTLY this !!!!

Yes, but then they forgot about updating the sensor to make it future-proof also for the next 3.5 years. So from 2008-2015 the Canon 5D-series image quality stood still.

Reasons mentioned for the desire to stay around 20 megapixels were increased dynamic range and higher ISO performance. Now it turns out that nothing was won (in RAW) by staying at 22 MP, while the other sensors made in the year 2012 showed that high ISO performance and two stops more exposure latency can be had, along with more resolution.

The DR of the 5D mark III will be better in high ISO.
So it really depends what you want to capture.

The Nikon D800 is a super camera but in real world not better then the 5D mark III
Both very good.
You want to make me believe that the Nikon D800 is out of this world...
Well NO.

Some tests with DXO Analizer.
These are only numbers abcourse.

http://forums.dpreview.com/forums/readflat.asp?forum=1032&message=41135303

Until now it seems that the 5D mark III and the Nikon D800 are very close depending on what you capture and how.

There is more then portrait photography !!!!

--
Enjoy Photography
http://www.wildlife-photos.net

 

[Tony Northrup]

Anyway back to the OP, Tony thanks again for your great comparison. Loved the rubbish high-iso video footage with INMACULATE sound, which ultimately sells the shot. This to remind us again how audio is just as important as video.

Thanks. My book (Stunning Digital Photography) has video lessons integrated into it, so I have created many videos but i'm still relatively new to videography. I used a Sennheiser ew 100 G3 wireless lav mic tucked inside my jacket to cut wind noise. The receiver was mounted to the 5D3's hot shoe and plugged directly into the mic jack. This was the first time I used a DSLR to record the audio; I've been recording the audio off-camera into an H4n and then syncing in post. The audio was good enough that I'll be leaving the H4N in the studio from now on.

After reading reviews from other people, I expected the low-light video to be much better than it was. Natively it was much darker; I had to jack up the brightness in Premiere Pro. In fairness, it was really dark outside; even two minute exposures at ISO 100 and f/4 were a bit dark--I should have done 4 minutes.

 

[Invertalon]

I disagree with your claim that the RAW files are not improved at all.

I see a difference between the 5D2 and 5D3 at high ISO. At 12,800 the 5D3 wins, no competition. At 25,600 the difference is huge. I will even use 51,200 without a problem if I need to.

I clearly see a difference in the quality of the noise, the contrast and lack of banding and other bad artifacts at high ISO. At low ISO the IQ is really close, but the 5D3 has the edge at the high end.

I owned a 5D2 for a year prior to the 5D3 and have processed thousands of files, many at really high ISO. ISO 6400 on the 5D2 was great, but 12,800 fell down fast.

 

[Tony Northrup]

Class action? Lol! You need to get a life. Raws only have incremental improvements. The improvements have been in processing software and jpeg engines. Ever compared raws between 1ds and 5d? 1d2 and 1d3?

I think you're saying you agree with my assessment that the 5D3's sensor isn't any better, but you think it's an obvious conclusion.

I don't think it's obvious at all. Canon's marketing discusses the many improvements they've made to the sensor and the results they have on image quality, which I've not been able to find, and many people on this forum believe the 5D3 creates much better images.

While I haven't compared the raws from the 5s and the 1s, I shot the 5D classic and 5D2 side-by-side since the 5D2 came out, and the 5D2 produces MUCH better raw images than the original 5D. Marketing aside, the historical improvements would lead me to believe the 5D3 would create better images.

 

[ukat123]

Raws only have incremental improvements. The improvements have been in processing software and jpeg engines.

Yes, I agree. It is only software tweaking that gives a perception of better noise performance.. Raw files tell the truth. 5d2 and 5d3 files are nearly identical, and in some cases, 5d2 files are even better. It has been confirmed by many tests. I think it is ok as many wedding photographers and kind of point and shoot photographers are not looking after the best quality. They will shoot jpegs to save time and processing so will benefit from the software tweaking. The biggest benefit of the 5d3 is its AF, which is mainly what 5d2 owners wanted, so Canon listened, but juts took an easy route. Saying that, they cannot develop a groundbreaking camera with each iteration. So perhaps the next one will be groundbreaking.

 

[Tony Northrup]

The DR of the 5D mark III will be better in high ISO.
So it really depends what you want to capture.
...
There is more then portrait photography !!!!

Well, if you watch the video linked in the original post, you'll see I tested the cameras in just about every photographic scenario. I was specifically interested in DR, and tested it by recovering both the highlights and shadows of the raw files, at a variety of different ISOs. I could find no improvement.