Compare 5D2 vs 5D3 vs 7D dark noise shots, 100% crops online w +4 EV shadow push

[Aglet]

Here's a teaser, there's a lot more to come but I figured it would be interesting to see some results from my raw file collection of dark shots put together to compare.

On this page

http://a2bart.com/tech/5d2vs5d3.htm

You'll find a link to see a 5D Mark II compared with a 5D Mark III, a 7D, and a little consumer Nikon D5100.

All shots taken in manual mode, Daylight WB, 1/200s, f/16, all in-camera noise reduction disabled, all in-camera tone-curve tweaking disabled. lens cap on, in a dark room, viewfinder covered, LCD display off.

Raw files all processed with ACR using black clipping level set to 0, +4 EV exposure comp to make the noise levels more visible (and this is a range of shadow pushing that is possible). Adobe 2010 Standard image processing chosen for all bodies to hopefully normalize color response.

We can now get a better idea of low ISO DR performance of these 2 Canon icons at least at the black end of the test.
And then there's that bandy little 7D...

Now I need some sleep.

Anyone have a D800 or D4, D700, D3, D3s, D3x, etc., they can knock off some raw files for me to add to the mix? If so, send me a message.

2012-04-03-1310mst update - the error with the 3200 iso 5D3 tile should now be fixed. Thanks for the sharp eyes out there.

Meanwhile, anyone new to the thread can have fun with the 200 ISO guess-who below.
http://a2bart.com/techimg/dknz/200isoguesswho.jpg

 

[foobar]

Thanks for the comparison. It clearly shows my main gripe with the 7D: Lots of shadow noise even at ISO 100. It's almost impossible to get something useful out of the darker areas of an image because of all the noise. From my experience, this wasn't as much of a problem with older Canon APS-C Sensors.

Too bad that the 5D2 doesn't fare much better in that regard.
In overall noise performance, the 5D2 beats the 7D by ~1,5 stops, though.

 

[DavidRiesenberg]

That's a really interesting test. Is there any reason to shoot with a lens attached? Why not with just the body cap on? Also, in ACR, did you zero the brightness and contrast sliders or did you leave them at the default 50/20?

 

[nightbreath]

Hi Aglet, it'd be great if you took crops of an area corresponding 5% of the image to get relative noise performance of the whole sensor, not on pixel level. Of course if you didn't do that in your test

 

[prestonpalmer]

This doesn't tell us anything useful...

 

[docsmith]

Nice tests. Thanks for putting it together.

 

[thure1982]

This doesn't tell us anything useful...

Of course it does!
It tells us that people who have to much time on there hands should be outside shooting pictures but are instead shooting black walls inside a closet.

 

[thure1982]

All shots taken in manual mode, Daylight WB, 1/200s, f/16, all in-camera noise reduction disabled, all in-camera tone-curve tweaking disabled. lens cap on, in a dark room, viewfinder covered, LCD display off.

The lenscap is on?
Why f/16? Is there something special about you choosing this?
Why even a lens?

 

[12mm]

This tells us everything about the sensor in a camera, if you know how to read it. And in this case it tells us that canon tries to sell us the same old shit in its new 5D3 like in the 5D2. Its hardly improved by anything. Except the horizontal banding is gone. But still theres vertical banding. This test also shows us how clean the Nikon sensor is at lower ISOs. So please go ahead and tell us again that this isnt usefull at all for actual photograhy. For the people who know how to read these tests and try to evaluate if its worth upgrading from 5D2 to 5D3 this is of substantial interest! Thanks for these pictures. I was waiting for something like that before making my decision.

 

[Jettatore]

Somewhat related so I'm not starting a new thread.

http://www.dpreview.com/previews/canoneos5dmarkiii/12

I can't save it in the link but I switched the Sony in the lower right to the canon 7D. That makes it a comparison between the D800 - 5DIII, 5DIII and 7D (all for RAW shooting).

I gotta be honest, as far as this DPReview test alone goes, I can't see any real-world difference between any of the cameras. There is nothing about any of the minute differences at the same settings (doesn't really matter which settings either) and for real world usage they all seem nearly the same.

D800 looks exactly to me what a 5DII or 5DIII file would look like if you resized it to 44 Megapixels in the digital editing stage, and vice versa, the 5DII and 5DIII shots look exactly like what I imaging a high-megapixel D800 shot would look like if you scaled it down to a lower Megapixel in digital editing.

Bear in mind this is only what my eyes see, looking at this test/tool. I have a feeling the more important issues have to do with how you like the camera controls and ergonomics, the FPS, viewfinder, digital layouts and so on and so forth. They all look the same to me, and going by the galleries upon galleries of digital images I have studied, the same thing seems to hold true over and over again, having reasonably good glass, any modern digital body, good photography skills and good editing skills and wallah, you have amazing results.

 

[dafrank]

Congratulations! This means we can look forward to more "shooting the inside of my lenscap" scientism from those relentless fellows determined to show that almost all Nikons (and Sonys, if necessary to make a point), no matter how lowly, are obviously superior, in the most relevant and important way (sarc), to any Canon, especially the new 5DIII. Pleeeease, let's have some moderation here before this thing gets out of hand, like some eternal Pong game, and turns into a miniature version of DPR's Canon forum. I can barely stand to glance at that forum anymore, what with the seemingly unstoppable Nikon fanboy infection and inevitable bad reaction from the Canon faithful. Even now, the moderators there are attempting to tamp this insanity down. Please don't let it happen here.

 

[TW]

Please don't let it happen here.

Too late.

Far, far, too late.

<Sigh>

 

[kbmelb]

I appreciate the the test an the effort. I believe the 5DIII 3200 ISO shot is the same as the 7D.

It tripped me out a little. I couldn't figure out why 3200 was so bad on the 5DIII the I looked at the 7D.

 

[akclimber]

Thanks for the tests. Very interesting. Any plans on showing results from less extreme pushing, say +2 stops? That seems like a more reasonable, real world sort of test. My 5D3 should arrive tomorrow and out of curiosity I'll probably do the same test with a 5D2 & 7D and even a IR converted T3i if it's be useful.

And to the folks who are critical of anyone posting this type of info - some of us are interested in sensor performance pixel peeping and find it informative, whether or not it has any real world relevance. Personally, if I'm going to drop $3500 on a camera (or even just $100), I want as much info about the camera's performance as possible. If you don't, well maybe consider just moving on, that way it won't turn into a DPR type debacle.

Cheers!

 

[Loswr]

I believe the 5DIII 3200 ISO shot is the same as the 7D.

No, it's not. It's just that the completely random noise in both shots happens to match up perfectly. It's like that one monkey with a typewriter among an infinite number of them that just happens to bang out the complete works of Shakespeare.

Well spotted, BTW. And if a completely accidental error happens to make the 5DIII look worse, oh well.

 

[madmailman]

Hi guys,

A complete n00b here (<- First post ->) but I took a quick look at the link (http://www.dpreview.com/previews/canoneos5dmarkiii/12 posted above. As I am currently a 7D shooter I changed on of the cameras to the 7D, crancked up the iso to 3200 on the 7D, D800 and Sony A900. Then moved the wee square onto a black portion of the image so all the samples showed some shadow noise. Please could somebody else try this and tell me if I need new glasses or not. I don't want to post what I see until somebody else does so I can't be accused of missleading somebody with my observations.

Thanks
MMM

 

[straub]

No, it's not. It's just that the completely random noise in both shots happens to match up perfectly. It's like that one monkey with a typewriter among an infinite number of them that just happens to bang out the complete works of Shakespeare.

Actually it is, they're both "7Diso3200c.jpg", there is no "5d3iso3200c.jpg"

 

[Loswr]

No, it's not. It's just that the completely random noise in both shots happens to match up perfectly. It's like that one monkey with a typewriter among an infinite number of them that just happens to bang out the complete works of Shakespeare.

Actually it is, they're both "7Diso3200c.jpg", there is no "5d3iso3200c.jpg"

Silly me, did I forget the <sarcasm> tags?

 

[straub]

Silly me, did I forget the <sarcasm> tags?

Yes, you did.

 

[nitsujwalker]

I believe the 5DIII 3200 ISO shot is the same as the 7D.

No, it's not. It's just that the completely random noise in both shots happens to match up perfectly. It's like that one monkey with a typewriter among an infinite number of them that just happens to bang out the complete works of Shakespeare.

Well spotted, BTW. And if a completely accidental error happens to make the 5DIII look worse, oh well.

hahaha. That freaked me out.. The 5diii was looking pretty good then WHAM ISO 3200. Then I glanced at the 7D and it is the same. Oops.