5D MarkIII Low ISO performance in shadows

[bdunbar79]

I plan on testing my 5D Mark III tonight downtown doing night macro shots.  Yes, I know I'm insane, but at least it'll give us concrete shadow perfomance noise and/or cleanup post-processing. 

[Alexandros]

I plan on testing my 5D Mark III tonight downtown doing night macro shots.  Yes, I know I'm insane, but at least it'll give us concrete shadow perfomance noise and/or cleanup post-processing.

Please do so, and update us when you have some news !

[Ivar]

No comparison is bad or good, they just serve different purposes. Downsizing a bigger image compares how a camera reacts to the *same* amount of light - which wouldn't be the case without downsizing (assuming the same (area) size sensors and shot setup). At the same time looking resolving power, a smaller image has to be uppsized because downsizing would just throw the extra info away - detail wise there is no difference between two 600x400 pixels image one being 80MP MF the other G1X. All comparison have meanings done and interpreted properly.

Yes, turn my 36mp body into a 22mp body in order to make it look better.  Thats how DXO rationalizes the high per pixel noise into something usable.  Maybe if they resized it to one pixel it would be perfect.

The point is that there is no holy grail, and all the posting being done by those who just read articles and have zero experience with either camera is not very helpful.

Obviously, my D800 can take some amazing images at ISO 100 - 400, but just like the OP notes with his 5D MK III, as soon as you get into shadows and the ISO gets up over 400, you have to be more careful with exposures, and at very high ISO's, the 5D MK III has more DR than the D800.

[jaduffy007]

i still dont have any sort of low iso noise anything like this on any of my bodies 5dmk3 5dmk2s or 1Dmk3

Would it be possible for you to post a link to a raw where the OP and all of us can see how much better your bodies are?

[bigmag13]

Another example of 5D MKIII shadow noise.

You can see the LR metadata.

Is that shadow noise acceptable?

I will have some 8x10's printed tomorrow to see how this noise looks when printed.

well first off is that i think you should have tried to pull more info from that shot. looking at the histogram and looking at the shot tells me that there is too much difference between the lights and dark's in the image to expect much detail from the shadows. its just not a properly exposed pic ( as Siliconvoid pointed out). The histogram should be a lil more to the right because the DR is great in that shot. its not high-key or low-key so you should have pulled for more detail in that fur if you wanted it exposed properly. 

I knew that the nose of my dog and the white fur were to far apart in range so had to expose as so, which meant popping the flash or using a tripod to get enough info.
I popped the flash.
-shot with a 50D and a Sigma 30 1.4

[David Hull]

This guy ran through this same thing over on DPR a week or so ago.  Quite a few people tried to help him but apparently he didn't hear what he wanted to here there so he has trotted this back out over here.  Regardless, one look at his RAW files for the two images he put up in his OP will show that they are under exposed by 2 and 3 stops respectively.  A simple application of ETTR will solve his issue (for these two images anyway).

[Alexandros]

This guy ran through this same thing over on DPR a week or so ago.  Quite a few people tried to help him but apparently he didn't hear what he wanted to here there so he has trotted this back out over here.  Regardless, one look at his RAW files for the two images he put up in his OP will show that they are under exposed by 2 and 3 stops respectively.  A simple application of ETTR will solve his issue (for these two images anyway).

Dear David,

This guy has a name, Alexandros. Thanks for bringing the report back from DPR forums. You have done an excellent job as a forum patrol/scout.

[wockawocka]

Am I the only one to think that 5d MK3 has very poor performance in shadows, even in low isos like 100 or 200 ? High color noise, weird noise patterns, vertical bands/stripes all without any pushing!!!
Straight out of the box the images look terrible in the shadow areas.
Why is that? I am very disappointed. VERY disappointed ...
Even my poor old 350d did better in that domain...
Is there something wrong with my copy or is it supposed to be like this ?

I think with any comparison the original raw files need to be put on a server for individuals to download.

Otherwise nobody can spot other reasons for poor performance. (Like badly exposed images - which cause noise). The images need to be from the same spot, in the same lighting too. A dog in amber lighting will be totally different to a flower with a dark background behind it.

[David Hull]

This guy ran through this same thing over on DPR a week or so ago.  Quite a few people tried to help him but apparently he didn't hear what he wanted to here there so he has trotted this back out over here.  Regardless, one look at his RAW files for the two images he put up in his OP will show that they are under exposed by 2 and 3 stops respectively.  A simple application of ETTR will solve his issue (for these two images anyway).

Dear David,

This guy has a name, Alexandros. Thanks for bringing the report back from DPR forums. You have done an excellent job as a forum patrol/scout.

My pleasure, thanks.

Now, if you were to re-shoot the shots in your OP with a proper exposure instead of two to three stops under, your result would be better.  IMO, the noise you are seeing is normal for that sort of exposure scenario with that camera based on what I have seen from other images I have looked at.
 

[revup67]

I've posted this before but if you click this link http://www.fredmiranda.com/5DIII-D800/index_controlled-tests.html scroll almost to the bottom and note the paragraph that begins with:  "I know this is disappointing for Canon shooters.." for a potential work around to the reduced DN.

Also, I think as Mt. Spokane pointed out earlier..there's no Holy Grail.  I recently read and saw an in depth article with video clips on how difficult it is to get the D800 with proper WB and the reviewer was attempting to decide between the 5DM3 or D800.  He claims if was far more difficult to adjust for proper WB than deal with noisy DN..he chose the 5DM3.  If interested here's that link as well:  http://www.cinema5d.com/news/?p=11652

[n0iZe]

Because do you really think those samples are properly exposed?  A great camera cannot compensate for poor photography.

Instead of bashing could you please keep your mouth shut. I tried to be polite before but I am losing my temper. Or t least think before you type. I don't have the right to have a dark underexposed looking image? The shadows should be clean in such situation. I didn't want criticism regarding my images or my technique I just wanted to figure out the noise issue.

Dude, you really take a bit of critics personal, don't you?
You have to prepare for critics if you upload your images to a website - it's always the same. It's web 2.0 - share your opinion on everything, even if no one would be interested in your opinion, you still share it don't you?
So just let the critics talk as well, since we also let you talk.

That said, I'll talk now.
After taking a look at your pictures, I suppose you want to create a dreamy look on your photos. However, focus on the flower photo is badly set, also is the exposure. Why did you take it at 1/2000sec at aperture 2.8? Where is the point in this?
Seriously, I have to defend Canon on this: take a look at that black. A camera is not supposed to fix your incorrect exposure! If you need dark images like this one, it would probably work out better if you take shots in RAW and afterwards adjust brightness.
I think that your banding would then be reduced by an awful lot.

Also imho the second shot doesn't look as bad as the first one regarding the banding and noise. I would guess it's because that shot isn't quite as under-exposed as the first one.


Just a suggestion:
People should sometimes practice more before complaining. Cameras don't fix pictures, they just take them.

[Marsu42]

Cameras don't fix pictures, they just take them.

While it the shortcomings of the op's shots are obvious and don't need to be discussed further, I disagree with you on this one: Of *course* I want my camera to fix my pictures = enabling me to fix errors in post!

Exposing correctly might not be an issue while shooting landscapes, but if you shoot events then you're quite busy with the things around you, so a lenient camera (good af, much dynamic range beyond the scene dr as a fix for wrong exposure, low noise at high shutter speeds to fix my shaking hands while holding heavy gear for hours).

I'm sure future camera bodies will cover up photog's lack of experience even more in the future, so as a pro it's less and less about exposing correctly by running around with a light meter...

Not too long ago I even saw a sensor concept enabling you to shift the *focus* in post, so you can just shoot away and wonder what to focus later on. But for the moment, this was a proof of concept and made for mobile phones rather then dslrs.

[n0iZe]

Cameras don't fix pictures, they just take them.

While it the shortcomings of the op's shots are obvious and don't need to be discussed further, I disagree with you on this one: Of *course* I want my camera to fix my pictures = enabling me to fix errors in post!

Exposing correctly might not be an issue while shooting landscapes, but if you shoot events then you're quite busy with the things around you, so a lenient camera (good af, much dynamic range beyond the scene dr as a fix for wrong exposure, low noise at high shutter speeds to fix my shaking hands while holding heavy gear for hours).

I'm sure future camera bodies will cover up photog's lack of experience even more in the future, so as a pro it's less and less about exposing correctly by running around with a light meter...

Not too long ago I even saw a sensor concept enabling you to shift the *focus* in post, so you can just shoot away and wonder what to focus later on. But for the moment, this was a proof of concept and made for mobile phones rather then dslrs.

Of course you do want them to do fix the pictures. However, they are just not capable of this at this point.

I do not question that it's not always an easy job. However I feel free to question it when it comes to still photography - afaik neither flowerpots nor flowers move a lot. So you have all the time you want for exposing, taking several photos at different exposures if necessary. Digital photography gives you way more space for trying out things. It won't harm your CF card if you take three different shots when not sure about exposition.
I catch myself doing this sometimes aswell, since the display on my Canon EOS 1Ds Mark II is so poor you can't actually judge the photos in daylight. Indoor, it's okay. But never try to judge the photos outside, lol. It doesn't work well.

Regarding shifting the focus after taking the actual picture, I think you refer to the Lytro. However I'm not sure how great the photos will come out: check out the "Play with the picture" function on the right of the main frame on their website lytro.com . I think sharpness is really poor. Also this is an easy picture since you can't really play with focus - you have two choices; either you focus on the girl or on the guy / wall. No matter what you focus on the girl, ring, tea cup or face - it always looks the same to me.
Don't know if this is a problem of illustration or if it's actually that bad in real life use.