DPReview Canon 5D mark III noise comparison

[dswatson83]

http://www.dpreview.com/previews/canoneos5dmarkiii/10

compared to the 5D mark II and D800, the D800 raw files suffer noise even more than the 5D mark II! The mark III seems to be about .5-1 stop better as far as raw files go from the Mark II. Everything seems sharp.

 

[dswatson83]

I'm actually a little disappointed from these noise tests with raw files. It beats the D800, 7D, and 5D mark II but not by much. Looks like the Mark 3 is the best by about .5 stop (and only at high ISO), then the mark II, then the D800, then the 7D which is about a stop below the mark 3. The biggest differences seem to be a usable 12,800 iso vs the other cameras that are not...but on the raw files, the difference isn't huge. This is no Nikon D3S but then again it is 22MP which is still very large.

 

[Astro]

i saw this in the comments.

compare it to the 550D sample:

http://i.imgur.com/AYQoL.jpg

mhm.... :

http://i.imgur.com/KHNeA.jpg


after having a close look at the samples i have to say while the 5D MK3 is sure a great camera the maximum IQ has not much (if at all) improved for canon over the last years.
i don´t speak about HIGH ISO.. i speak about ISO 100-400 for maximum IQ.

i don´t realy need high iso.
i checked my database and only 5% of my 43542 images are shot with an ISO higher then 800.
most of my pictures are landscapes, still life studio shots or portraits.

 

[skitron]

All I can say is I sure am glad I bit on the 5D2 for $1999 bundled with about $350 street value of software that I actually use. 5D3 looks slightly better in RAW but not even close to twice as much $ better. AF isn't much a problem so far with 5D2, nothing to write home about but useable for me. I'd say the 5D3's biggest problem is price point.

 

[awinphoto]

I'm actually a little disappointed from these noise tests with raw files. It beats the D800, 7D, and 5D mark II but not by much.

To be honest, i've mentioned and so have others that with these cameras now-a-days, things are getting so tight, so competitive, so close that from this point forward, I doubt you will ever really see, in raw terms, any clear hands down superior camera given similar generations of camera... things are so close and so minimal, the difference between 2 already very spectacular sensors are probably going to get smaller and smaller as we progress with digital cameras and photography.

 

[Canon-F1]

compare it to the 550D sample:

http://i.imgur.com/AYQoL.jpg

yep that does not look good.

 

[helpful]

The Canon 5D Mark III looks better than the D4 even and much better and sharper than the D800 and 5d II.

Look at this:

http://www.dpreview.com/reviews/studiocomparefullscreen.asp#baseDir=%2Freviews_data&cameraDataSubdir=boxshot&indexFileName=boxshotindex.xml&presetsFileName=boxshotpresets.xml&showDescriptions=true&headerTitle=Canon%20EOS%205D%20Mark%20III&headerSubTitle=Standard%20studio%20scene%20comparison&masterCamera=canon_eos5dmkiii&masterSample=5d3_5199&masterDisableCameraSelection=true&slotsCount=4&slot0Camera=canon_eos5dmkiii&slot0Sample=5d3_5199&slot0DisableCameraSelection=true&slot0DisableSampleSelection=true&slot0LinkWithMaster=true&slot1Camera=canon_eos5dmkii&slot1Sample=img_0038&slot2Camera=nikon_d800&slot2Sample=dsc_8318&slot3Camera=sony_dslra900&slot3Sample=dsc00025&x=-0.85652383245065&y=-0.9085714285714286

All the other images are getting mushy at ISO 6,400 but the 5D III is super sharp and the black is pure black with no noise. Amazing!

 

[SpartanWarrior]

Guy's don't forget though the 5D II also has a slower shutter speed than the 5D III, I kind of think the 5D III has about 0.5-1 stop advantage to the 5D II.

 

[VirtualRain]

I'm actually a little disappointed from these noise tests with raw files. It beats the D800, 7D, and 5D mark II but not by much. Looks like the Mark 3 is the best by about .5 stop (and only at high ISO), then the mark II, then the D800, then the 7D which is about a stop below the mark 3. The biggest differences seem to be a usable 12,800 iso vs the other cameras that are not...but on the raw files, the difference isn't huge. This is no Nikon D3S but then again it is 22MP which is still very large.

Are we looking at the same images? Looking at their 100% crop of the coins... 5D3 at 25K looks about the same as the 7D at 3200. And, the 5D3 at 12.8K looks much better than the 7D at 3200.

In general, I've noticed that looking at 100% crops of the 5D3 reveals obvious noise at high ISOs, but it's of a much more palatable variety and maintains much better contrast and sharpness than any other camera, especially my 7D.

I see myself using ISO 12,800 regularly and 25K in a pinch on the 5D3 where I am not happy with the noise at 3200 on my 7D and try to stick to max of 1600 ISO there. That's an effective 3 stop improvement for me.

 

[helpful]

Chroma noise (color) at ISO 6,400. Also way better:

http://www.dpreview.com/reviews/studiocomparefullscreen.asp#baseDir=%2Freviews_data&cameraDataSubdir=boxshot&indexFileName=boxshotindex.xml&presetsFileName=boxshotpresets.xml&showDescriptions=true&headerTitle=Canon%20EOS%205D%20Mark%20III&headerSubTitle=Standard%20studio%20scene%20comparison&masterCamera=canon_eos5dmkiii&masterSample=5d3_5199&masterDisableCameraSelection=true&slotsCount=4&slot0Camera=canon_eos5dmkiii&slot0Sample=5d3_5199&slot0DisableCameraSelection=true&slot0DisableSampleSelection=true&slot0LinkWithMaster=true&slot1Camera=canon_eos5dmkii&slot1Sample=img_0038&slot2Camera=nikon_d800&slot2Sample=dsc_8318&slot3Camera=sony_dslra900&slot3Sample=dsc00025&x=-0.85652383245065&y=-0.9085714285714286

 

[shizam1]

Unless you shoot JPEG, make sure you select RAW to compare the two cameras.

With RAW selected, and ISO 800, the 5DII actually looks sharper than the 5DIII... I wonder what method they used to do the conversion?

 

[helpful]

I'm actually a little disappointed from these noise tests with raw files. It beats the D800, 7D, and 5D mark II but not by much. Looks like the Mark 3 is the best by about .5 stop (and only at high ISO), then the mark II, then the D800, then the 7D which is about a stop below the mark 3. The biggest differences seem to be a usable 12,800 iso vs the other cameras that are not...but on the raw files, the difference isn't huge. This is no Nikon D3S but then again it is 22MP which is still very large.

Are we looking at the same images? Looking at their 100% crop of the coins... 5D3 at 25K looks about the same as the 7D at 3200. And, the 5D3 at 12.8K looks much better than the 7D at 3200.

In general, I've noticed that looking at 100% crops of the 5D3 reveals obvious noise at high ISOs, but it's of a much more palatable variety and maintains much better contrast and sharpness than any other camera, especially my 7D.

I see myself using ISO 12,800 regularly and 25K in a pinch on the 5D3 where I am not happy with the noise at 3200 on my 7D and try to stick to max of 1600 ISO there. That's an effective 3 stop improvement for me.

I agree. I think that there is some confusion about sharpness vs. noise. Sharpness is merely a matter of lenses, focus, and image software processing. The AA filter is not going to play that big of a role in the D800 vs. 5D3 except when it is completely removed. All these images could be made to appear much sharper if someone simply used Lightroom to sharpen for screen display. When strictly talking about noise, someone needs to look at the shadow and the blacks, and at colors for chroma noise (odd splotches of color in areas of a single color, ideally looking at multiple patches of different colors in the image). Sensors always have more green in a Bayer array, so it is better to look at red or blue to see chroma noise. Noise always tends to go down in brighter parts of an image (even at high ISO everything is blown out white with a bright enough exposure, so the noise goes away).

This is a center crop of an image at ISO 4,000 from a Nikon D7000 (night softball game). It is simply to demonstrate that ISO 4,000 of the D7000 is the same as ISO 1,600 of the Canon 7D (basketball game photo). Both of the images are mine--I have obscured the watermark only to keep my privacy.

I have looked at all the samples and as near as I can tell the 5D3 does have at least 2 stops lower noise than the 5D2, and at ISO 12,800 should perform about the same as the Nikon D7000 at ISO 4,000.

For any Canon 7D shooters out there, I would recommend the 5D3 has a must-have upgrade. Anyone wondering whether to buy the 7D or the 5D3 should absolutely go with the 5D3 if you can afford it.

The Canon 7D was simply unbelievable in almost every aspect, and it has filled in just as good as a full-frame camera in many situations for me. It is pretty amazing that another camera came out that was 1 1/3 stops better (Nikon D7000), and it is like a miracle that another camera exists which is 3 full stops better than the 7D (3 stops is eight times better, for those who don't know the vernacular). Don't hesitate to purchase the Canon 5D Mark III!

 

[helpful]

Unless you shoot JPEG, make sure you select RAW to compare the two cameras.

With RAW selected, and ISO 800, the 5DII actually looks sharper than the 5DIII... I wonder what method they used to do the conversion?

All the images posted to the internet are actually JPEG images (unless the original RAW is downloaded).

So the RAW images in the DP Review comparison tool are actually JPEGs that were processed at unknown settings by the staff of DP Review. The sharpness should not be looked at when evaluating noise.

The sharpness of the RAW images actually depends completely on what sharpness was selected when the images were processed to JPEG. And we just don't know. If you look at the JPEG generated by the camera, then that is at least an apples-to-apples comparison, unless they tweaked the default camera settings, which DP Review claims not to have done.

 

[shizam1]

Unless you shoot JPEG, make sure you select RAW to compare the two cameras.

With RAW selected, and ISO 800, the 5DII actually looks sharper than the 5DIII... I wonder what method they used to do the conversion?

All the images posted to the internet are actually JPEG images (unless the original RAW is downloaded).

So the RAW images in the DP Review comparison tool are actually JPEGs that were processed at unknown settings by the staff of DP Review. The sharpness should not be looked at when evaluating noise.

The sharpness of the RAW images actually depends completely on what sharpness was selected when the images were processed to JPEG. And we just don't know. If you look at the JPEG generated by the camera, then that is at least an apples-to-apples comparison, unless they tweaked the default camera settings, which DP Review claims not to have done.

Yeah, I know all that. And you can't compare the JPEG's generated by the cameras, because most people who want quality already know to shoot RAW and have their computer comvert to JPEG. Comparing in-camera generated JPEG's is not even close to comparing apples to apples if you want to gauge how the new sensor does at differen't ISO's, as the new Digic 5+ processor is some 30X more powerful than the one in the 5DII, and thus has time to do good conversions of the RAW data.

 

[helpful]

Yeah, I know all that. And you can't compare the JPEG's generated by the cameras, because most people who want quality already know to shoot RAW and have their computer comvert to JPEG. Comparing in-camera generated JPEG's is not even close to comparing apples to apples if you want to gauge how the new sensor does at differen't ISO's, as the new Digic 5+ processor is some 30X more powerful than the one in the 5DII, and thus has time to do good conversions of the RAW data.

Sorry, I should have been more positive. You are totally right that RAW is the way to make real comparisons.

The truth that I am trying to communicate is that the sharpness of "RAW" images when posted as JPEG images really depends on the processing done to them. The ultimate maximum level of sharpness depends on the sensor and AA filter of course, but I have seen scores of messed-up comparisons where a "better/sharper" camera's RAW images were just over-sharpened, and then compared to under-sharpened images from another camera.

The classic example of dishonesty with sharpening RAW images is to "prove" to newbies that they should "always shoot RAW." They give super sharp screen images produced by Lightroom from RAW images that are sharpened for viewing on a monitor, side-by-side with unsharpened (or even blurred) JPEG images from the camera at the lowest sharpening level. And then they crow about their conclusion that shooting RAW is the only pure way to take pictures, when it was all cooked.

And the very first word of my post was that noise isn't about sharpness.

I'm not sure what point you are trying to get across, anyway. I am defending the 5D3 based on accurate testing done by DP Review, and you are bringing up an irrelevant controversy about whether experienced shooters know how to shoot RAW vs JPEG and the power of the new Digic 5+ processor.

 

[shizam1]

I'm not sure what point you are trying to get across, anyway. I am defending the 5D3 based on accurate testing done by DP Review, and you are bringing up an irrelevant controversy about whether experienced shooters know how to shoot RAW vs JPEG and the power of the new Digic 5+ processor.

Ok, my point is I'm looking at the DPReview comparisons and actually seeing a better looking image with the 5DII at ISO 800. And we don't know what they used to do the comparisons, or what settings.

I have a 5DIII sitting at home waiting for me, so I'll do my own test tonight against my 5DII using the ACR RC and Lightroom to evaluate. Hopefully I get better results than what I'm seeing on DPReview!

 

[helpful]

Ok, my point is I'm looking at the DPReview comparisons and actually seeing a better looking image with the 5DII at ISO 800. And we don't know what they used to do the comparisons, or what settings.

I have a 5DIII sitting at home waiting for me, so I'll do my own test tonight against my 5DII using the ACR RC and Lightroom to evaluate. Hopefully I get better results than what I'm seeing on DPReview!

Hey, I'm your friend! My camera has arrived today as well! My box arrived at home and I am looking forward to doing the same thing. Hope I can get off work early...

I am sorry as well. I think I was a little too harsh, so my apologies to you.

 

[LetTheRightLensIn]

http://www.dpreview.com/previews/canoneos5dmarkiii/10

compared to the 5D mark II and D800, the D800 raw files suffer noise even more than the 5D mark II! The mark III seems to be about .5-1 stop better as far as raw files go from the Mark II. Everything seems sharp.

Looking at the jpg versions of their RAW ACR conversions:

It looks barely better than the 5D2 as is, however you need to note that they exposed the 5D3 1/3 stops less and the D800 2/3rds stops less than the 5D2 due to different sensor gain (talking ISO3200-6400).

So going by this it seems like the 5D3 is probably only 1/2 stop better than the 5D2, although we can still hope DxO will say 2/3 stop better, but I'm a bit more worried that might not be the case.

The D800 seems to be maybe 1/3 stop better than the 5D3 (again note that they exposed it for 1/3 stop less time and that you need to normalize the image sizes) at ISO6400. A bit hard to say.

Still waiting for DxO, we don't know that ACR applies the same NR or sharpening under the hood to each body and DxO should be able to better balance out for sensor gain differences.

 

[LetTheRightLensIn]

I'm actually a little disappointed from these noise tests with raw files. It beats the D800, 7D, and 5D mark II but not by much.

To be honest, i've mentioned and so have others that with these cameras now-a-days, things are getting so tight, so competitive, so close that from this point forward, I doubt you will ever really see, in raw terms, any clear hands down superior camera given similar generations of camera... things are so close and so minimal, the difference between 2 already very spectacular sensors are probably going to get smaller and smaller as we progress with digital cameras and photography.

At high ISO, barring a radically new technology they will probably all end up around within 1/3 stop of each other from now it seems, more or less the same, but Nikon is way ahead maximum dynamic range for now though by a few stops and even a stop plus ahead with their non-exmor sensors in that regard, they have the MP crown for now although that could certainly fall at any time back to Canon. I do greatly wonder if Canon can catch up for dynamic range and it would take $$$ bodies or years to happen. I sure hope not.

 

[shizam1]

Ok, my point is I'm looking at the DPReview comparisons and actually seeing a better looking image with the 5DII at ISO 800. And we don't know what they used to do the comparisons, or what settings.

I have a 5DIII sitting at home waiting for me, so I'll do my own test tonight against my 5DII using the ACR RC and Lightroom to evaluate. Hopefully I get better results than what I'm seeing on DPReview!

Hey, I'm your friend! My camera has arrived today as well! My box arrived at home and I am looking forward to doing the same thing. Hope I can get off work early...

I am sorry as well. I think I was a little too harsh, so my apologies to you.

Hey, I'll apologize as well, mis-interpreted your post! It's easy to come across harsh on forums too. I'm about to do some ISO tests, wish me luck!