High ISO comparo: 5DIII vs. D800

[3kramd5]

I'm GLAD that Canon focused on ISO and camera features rather than just pumping out more megapixels...even though I had hoped for around 28mp rather than 22mp, I'm satisfied.

Can you say they focused on ISO and AF rather than more MP when the D800 has more DR and perhaps quite similar ISO and also advanced AF (although perhaps the 5D3/1DX AF will prove better than the D4/D800 AF???)?

Of course you can. What the Nikon ends up being better/worse at has no bearing on what Canon's focus was.

[meli]

I'm GLAD that Canon focused on ISO and camera features rather than just pumping out more megapixels...even though I had hoped for around 28mp rather than 22mp, I'm satisfied.

Can you say they focused on ISO and AF rather than more MP when the D800 has more DR and perhaps quite similar ISO and also advanced AF (although perhaps the 5D3/1DX AF will prove better than the D4/D800 AF???)?

Of course you can. What the Nikon ends up being better/worse at has no bearing on what Canon's focus was.

Doubt you can. Its not that Canon operates inside a bubble, they do have to take into account what the competition is offering or where its competition R&D is heading to.
Its another thing though to what percentage this affects their decisions. It seems that the notion of having such a huge userbase in 2 different fields (stills /video) tied with their products, affects more their decisionmaking.

I hope this pattern doesn't continue, cause it starts to remind me Sony in the 90s right before the downfall.

[caMARYnon]

Did you read this http://imaging.nikon.com/lineup/dslr/d800/compatibility02.htm ? and In particular note 5 ...
So, let's wait for professional reviews and comparisons before judge

[Maui5150]

Some newer High ISO from the 5D MK III

http://www.engadget.com/2012/03/12/canon-eos-5d-mark-iii-high-iso-sample-images/

Also note, this is a pre-prod version of the body, so I expect production versions to be even sharper

[altenae]

Some newer High ISO from the 5D MK III

http://www.engadget.com/2012/03/12/canon-eos-5d-mark-iii-high-iso-sample-images/

Also note, this is a pre-prod version of the body, so I expect production versions to be even sharper

And again jpeg directly out of the camera. 

[V8Beast]

I would REALLY like to hear from a broad set of editors and art directors to know if that line has even a scrap of truth in it. I don't think DR or even noise are anywhere near the top things on an editors mind when they are critiquing photographs for publication.For one, no one can even gauge the dynamic range of a photo by eyeballing it, and even if they did measure it...what are they measuring? The DR capability of the camera you used to take the shot, or your fully post-processed image that has a myriad of exposure tweaks, curve adjustments, color tweaks, noise reduction, and sharpening applied?

Of course they're not sitting there measuring the DR or noise of final edited that images submitted to them. They judge an image just like anyone else based on the immediate visual, emotional, and artistic value it captures. That said, if there are overt technical deficiencies in the image, be it excessive noise, clipped highlights, or lack of shadow detail, they're going to notice. I'd say this is the photographer's fault rather than the equipment's fault in most instances, as it's the photographer's job to know his equipment and work around its limitations.   

Assuming you actually did capture a photo with 14 stops of dynamic range, does that even matter a wit for the final presentation format...in this case print?

The short answer is yes, it absolutely matters. Unfortunately, you have to accept the fact that the image you capture isn't going to reproduce in print nearly as nicely on paper as it does in it's original digital glory. Rather than say, "oh well, it's not going to reproduce anyway" and put in a half-ass effort, it means you put in even more effort to get your digital captures as good as humanly possible. 

If your photos are so noisy or have such atrocious DR that an editor dumps them, then the problem is far more likely that you aren't exposing or lighting your scene properly than the fact that the camera shows a minor amount of banding noise in the lower few bits of the 14 available.

Did I ever imply that this is the case? If you can't expose an image properly, you're not going to work professionally. Case closed.

As for IQ...well, even the crummy samples Canon has offered demonstrate that the 5D III will take photos with stunning IQ when they are exposed properly (and that really is the goal).

5DII? Please. I've taken images with my 20D and 1DsIII that are indistinguishable from each other. With enough extra effort in the field and post production, you can get stunning results from lesser gear. That doesn't change the fact that spending hours of additional time in post production isn't cost effective. 

There are SO many other things that make a photograph, and many more that make it art. Whether your camera is a stop or two less capable than the competitions is not going to cost you your job with that fancy magazine.

I think we're actually in agreement. My point was that regardless of how an image is used in print, the impact the digital files makes on an editor or art director on a fancy monitor is very important. A stop or two of DR or noise isn't going to be the difference between paying your bills or going broke, but to say it doesn't matter because it won't show up in print is ridiculous. The more latitude you have in your files, the greater the potential to save you time in the field and deliver a better product after the post production process.

[briansquibb]

A stop or two of DR or noise isn't going to be the difference between paying your bills or going broke, but to say it doesn't matter because it won't show up in print is ridiculous. The more latitude you have in your files, the greater the potential to save you time in the field and deliver a better product after the post production process.

+1 .... which is why I keep the iso as low as possible and why I have become a flash fanatic

[jrista]

A stop or two of DR or noise isn't going to be the difference between paying your bills or going broke, but to say it doesn't matter because it won't show up in print is ridiculous. The more latitude you have in your files, the greater the potential to save you time in the field and deliver a better product after the post production process.

+1 .... which is why I keep the iso as low as possible and why I have become a flash fanatic

You guys are missing my point. Assuming the worst case scenario, the 5D III hasn't changed. Its not better, but it also isn't worse. Millions of photographers have used the 5D II as well as cameras with much worse read noise and the same DR as offered by Canon cameras for years, and its never been a problem. Look at read noise levels for digital MF cameras, touted as offering FAR better quality than any lesser camera by professionals who use them every day for publication work. Digital MF has relatively poor QE (15-25%), high to very high read noise (15-30 electrons), limited maximum saturation relative to the likes of any current Canon or Nikon/Sony (less than half as much in more cases than not), and they all top out at around 11.5 stops of DR or less.

The Leica M9, also considered one of the best professional grade cameras on the market, has consistent read noise of about 15.5 e-, maximum saturation at lowest ISO of 30000, and maximum DR of 11.1 stops. There have been reports of banding issues with several Leica sensor designs as far back as the M7, and the M8 had particularly bad banding...but it was still considered a better camera than anything from Canon or Nikon...since banding only ever exhibited in shadows, and was relatively easily mitigated in post.

You can make the argument that better DR may make your life easier. If you regularly find yourself dragging up the shadows, then you might as well jump ship and head over to Nikon where the grass is greener. Or you could ETTR, utilize the sensor DR better (Canon does seem to have a bit more highlight headroom than Nikon by about 1/2 a stop based on DPR charts), and correct exposure at the click of a button in post (or, in the case of LR, you could simply set a negative exposure bias in the default import profile for your cameras, and never actually have to worry about it again...no time wasted whatsoever.) But the simple fact of the matter is the cameras that are literally considered THE BEST on the market by most professionals who shoot for print and publication on a daily basis, the likes of Hasselblad, Phase One, Aptus, etc. are no better or worse than anything Canon is or has been putting out. Actually, with the 1D IV, 7D, and 5D II/III, Canon is better on a technical level...although we all know that still doesn't matter a wit when it comes to producing good photographs.

[briansquibb]

You guys are missing my point. Assuming the worst case scenario, the 5D III hasn't changed. Its not better, but it also isn't worse. Millions of photographers have used the 5D II as well as cameras with much worse read noise and the same DR as offered by Canon cameras for years, and its never been a problem. Look at read noise levels for digital MF cameras, touted as offering FAR better quality than any lesser camera by professionals who use them every day for publication work. Digital MF has relatively poor QE (15-25%), high to very high read noise (15-30 electrons), limited maximum saturation relative to the likes of any current Canon or Nikon/Sony (less than half as much in more cases than not), and they all top out at around 11.5 stops of DR or less.

The Leica M9, also considered one of the best professional grade cameras on the market, has consistent read noise of about 15.5 e-, maximum saturation at lowest ISO of 30000, and maximum DR of 11.1 stops. There have been reports of banding issues with several Leica sensor designs as far back as the M7, and the M8 had particularly bad banding...but it was still considered a better camera than anything from Canon or Nikon...since banding only ever exhibited in shadows, and was relatively easily mitigated in post.

You can make the argument that better DR may make your life easier. If you regularly find yourself dragging up the shadows, then you might as well jump ship and head over to Nikon where the grass is greener. Or you could ETTR, utilize the sensor DR better (Canon does seem to have a bit more highlight headroom than Nikon by about 1/2 a stop based on DPR charts), and correct exposure at the click of a button in post (or, in the case of LR, you could simply set a negative exposure bias in the default import profile for your cameras, and never actually have to worry about it again...no time wasted whatsoever.) But the simple fact of the matter is the cameras that are literally considered THE BEST on the market by most professionals who shoot for print and publication on a daily basis, the likes of Hasselblad, Phase One, Aptus, etc. are no better or worse than anything Canon is or has been putting out. Actually, with the 1D IV, 7D, and 5D II/III, Canon is better on a technical level...although we all know that still doesn't matter a wit when it comes to producing good photographs.

Perhaps we are talking at cross purpose here.

My point is to maximise the DR because that gives the best possible image with some lattitude to play with. Go for high ISO and you throw that DR away. High ISO looks very flat colourwise so why use it unless you HAVE to?

[LetTheRightLensIn]

I trust them because they have solid track records about being accurate and meticulous about their camera testing. The only track record I have for most of the DPR tech heads is that they like to tear up Canon raw files and complain about them...A LOT. They could claim (or literally) have 10 Ph.D's...that wouldn't change their track record.

That is interesting because the DPR tech heads have a solid track record of ISO 100 DR measurements matching up with DxO. While DPR does not.

They enjoy complaining about Canon DR, I won't fault them for doing what they enjoy...rather than getting out into the real world to enjoy photography...although I think they might be happier doing the latter.

Some of them actually have much more extensive galleries, from around the world, than the people telling them to go out and shoot.

That is pure assumption, and will be until the camera actually hits the streets and non-beta software is used to evaluate IQ.

Maybe production models will vary appreciable from the beta cams, but that has not been the case in the past.
Beta software is not being used for the DR tests. The only place people have used beta software is peeking at high ISO performance where it seems to show the 5D3 doing better than the 5D2 .

Even testing with a pre-production sample and beta software, we do know that Canon HAS DELIVERED several improvements with their latest sensors: two stops better Native ISO for both the 1D X and 5D III;

Yes, but two stop more native ISO is different than 2 stops better SNR in RAW. Not that is a bad thing though.
Even Canon themselves don't claim 2 stops better SNR for RAW.

elimination of fixed pattern noise leaving behind more appealing random noise;

Not at low ISO, although the horizontal banding appears to be 100% gone and even the vertical doesn't appear to show up as much at high ISO (although it hasn't been carefully tested yet). But yes it does seem that the high ISO noise will look nicer, less clumped, less giant chomra-blotched, no horizontal banding at all and potentially less objectional vertical banding (at high iso only). Certainly nice improvements, no doubt.

and 6fps on the 5D III (almost double its predecessor WITH an increase in resolution, even though it was modest.)

6.0 isn't almost double 3.9 unless the D800 as almost double the MP of the 5D3 and it does not.

That said getting to 6 is big since that is the absolute minimum where it just begins to appreciably help for sports.
I wish the 6.9fps or 7.5fps rumors had been correct though since then you'd really be talking. But 6fps will sometimes get you more than one good frame, nearly as often as not, but 7-8fps almost always will. Still it is much better than 4fps which hardly ever will and it's much better than even 5fps which will only do that from time to time.

Saying Canon has not delivered even a "single" improvement over the last four years is just flat out wrong, even in the case of DR. Canon has consistently delivered improvements to DR and low ISO noise...they have just been smaller and smaller improvements as they have approached 12 stops.

When did I say they haven't offered a single improvement? I've said that the video might be much better, the fps are up, the AF should hopefully be way better, high ISO should be somewhat better.

As for DR they really haven't improved that for 4.5 years though (at base ISO).

[jrista]

Perhaps we are talking at cross purpose here.

My point is to maximise the DR because that gives the best possible image with some lattitude to play with. Go for high ISO and you throw that DR away. High ISO looks very flat colourwise so why use it unless you HAVE to?

I guess I'm confused, as I don't remember saying to go for high ISO. I agree, you should use the lowest ISO you can at all times to maximize DR. But you can do things like expose to the right at ISO 100 to make more effective use of the DR available there, correct the overexposure in post, and push down the noise floor digitally. (Certainly you may have some limitations there if your shooting lots of motion or need very deep DOF...however in those cases you are probably shooting at a much higher ISO to start with. In the case of landscapes or any kind of still scenes, your ability to ETTR is extreme.) Canon has a lot of highlight headroom (they tend to favor highlights at the cost of shadows, where as Nikon seems to generally be slightly worse on the highlights for much better shadows), and you can push exposure pretty far before you actually blow out highlights.

I'm not a working professional who regularly creates publishable work, but I have spent several years with Canon cameras exposing bright subjects in the dark. Namely the moon, where its possible to push exposure so far to the right it looks like you have nothing but a white disk, and still not blow out the highlights. You have a TREMENDOUS amount of room to recover at that point, and noise in the low frequencies is rarely a problem. Personally, I have found Canon's low ISO DR to be quite usable and very versatile when you make effective use of the highlight headroom that is available...far more usable than the worries over low ISO read noise that you get off of DPR and CR forums would seem to indicate.

[briansquibb]

I guess I'm confused, as I don't remember saying to go for high ISO.

mmm - that is the topic of the thread .... 

[sarangiman]

People keep touting the notion that ISO performance is increased, but DR at low ISO is not. I'm trying to understand how this is possible... is the following valid:

• Increased QE due to better microlenses = increased SNR across the board. Coupled w/ lower gain, this decreases per pixel noise (shot noise is decreased). You'd think this'd also increase DR even at low ISO by decreasing noise at the lower end...
• No improvement in read noise means less usable data on the lower end

So, ISO performance is increased b/c for any equivalent ISO setting on the, say, 5DII, the 5DIII is actually receiving more photons... which translates to higher SNR.

BUT, DR is only slightly improved at low ISO b/c absolute SNR increase for dark pixels is small compared to absolute SNR increase for brighter pixels (e.g. say read noise on 5DII & 5DIII is 5e-; QE is 0.5 on 5DIII vs. 0.25 on 5DII | then: for a signal of 20e- SNR of 5DIII vs 5DII would be 2 vs. 1, but for a signal of 2000e- SNR of 5DIII vs 5DII would be 200 vs. 100)?

Meaning DR is largely determined by full-well capacity, bit-depth of ADC, & read noise? Neither of which, it'd seem, have changed much for the 5DIII compared to 5DII?

(This of course leaves out practical usability of low-end due to banding, which is another issue in an of itself... less FPN = more forgiving raising of shadows since we're so sensitive to patterns).

Just trying to understand these arguments being thrown around... thanks!

[jrista]

People keep touting the notion that ISO performance is increased, but DR at low ISO is not. I'm trying to understand how this is possible... is the following valid:

• Increased QE due to better microlenses = increased SNR across the board. Coupled w/ lower gain, this decreases per pixel noise (shot noise is decreased). You'd think this'd also increase DR even at low ISO by decreasing noise at the lower end...
• No improvement in read noise means less usable data on the lower end

So, ISO performance is increased b/c for any equivalent ISO setting on the, say, 5DII, the 5DIII is actually receiving more photons... which translates to higher SNR.

BUT, DR is only slightly improved at low ISO b/c absolute SNR increase for dark pixels is small compared to absolute SNR increase for brighter pixels (e.g. say read noise on 5DII & 5DIII is 5e-; QE is 0.5 on 5DIII vs. 0.25 on 5DII | then: for a signal of 20e- SNR of 5DIII vs 5DII would be 2 vs. 1, but for a signal of 2000e- SNR of 5DIII vs 5DII would be 200 vs. 100)?

Meaning DR is largely determined by full-well capacity, bit-depth of ADC, & read noise? Neither of which, it'd seem, have changed much for the 5DIII compared to 5DII?

(This of course leaves out practical usability of low-end due to banding, which is another issue in an of itself... less FPN = more forgiving raising of shadows since we're so sensitive to patterns).

Just trying to understand these arguments being thrown around... thanks!

Just to offer some (to our best knowledge) real numbers for the 5D II, 7D, and 1D IV, if that helps anything (from sensorgen.info, based on DXO testing):

5D II @ ISO 100
Q.E.: 33%
Pixel Size: 6.4 microns
Saturation: 64600
Read Noise: 27.8 e-
DR Stops: 11.2

7D @ ISO 100
Q.E.: 41%
Pixel Size: 4.2 microns
Saturation: 20187
Read Noise: 8.6 e-
DR Stops: 11.2

1D IV @ ISO 100
Q.E.: 44%
Pixel Size: 5.7 microns
Saturation: 48702
Read Noise: 16.6 e-
DR Stops: 11.5

[V8Beast]

You can make the argument that better DR may make your life easier.

I don't mean to sound elitist, but this isn't something to be taken lightly. If you're taking photos for fun, I can see how spending 1 minute in post production to extend the DR of an image vs. spending 10 minutes isn't a big deal. However, if you're working on a tight deadline, need to process six dozen images to present to a client, and your livelihood depends on the quality of your images, out-of-camera files that "make your life easier" in post production isn't a luxury, it's a necessity.

Obviously, this doesn't only apply to DR, but also noise, sharpness, color reproduction, contrast, etc. It all adds up, and any time you can save in post production is time you can be spending behind the lens and making more money. I don't know about you, but I'd rather be shooting than staring into a computer screen and fiddling with a mouse 

If you regularly find yourself dragging up the shadows, then you might as well jump ship and head over to Nikon where the grass is greener. Or you could ETTR, utilize the sensor DR better (Canon does seem to have a bit more highlight headroom than Nikon by about 1/2 a stop based on DPR charts), and correct exposure at the click of a button in post

I would certainly hope that anyone attempting to earn a living with Canon gear utilizes a technique as simple as ETTR  Like you said, Canon files are incredibly good at highlight recovery, which makes ETTR a very useful tool in extending DR. My point is that over time, everyone is going to learn tricks like ETTR, or something as basic as using reflectors, fill light, multiple exposures, etc to extend DR. You're going to do that regardless of whether you shoot Canon or Nikon. Ultimately, however, a file with more latitude right "out of the box" will help you create the best image possible.

I'm not quite sure how this thread turned into a talk about DR, but DR is just one of MANY factors that determine IQ. Even if the D800 proves to have better DR than the 5DIII in the real world, I can just as easily decide that I hate it due to color reproduction, contrast, and sharpness that aren't my cup of tea. I remember the first shoot I did with the 5DC. I was blown away by the film-like image quality of the files. It was like I was shooting color slides again, and the color, contrast, and sharpness were simply stunning.  I'd never seen such incredible IQ on any digital camera before. I didn't care how its DR or ISO measured on some on chart posted by some geek on the internet. The images just had that certain look and feel to them that I cherished, and at the end of the day, that's all that mattered. IMHO, that's why you have to try these things out in the real world before determining a winner.