5D3 same max dynamic range as the 5D2???

[LetTheRightLensIn]

I've never seen banding on my files unless I've screwed up the exposure and have had to push the image.

It doesn't surprise me that dynamic range hasn't improved either. I mean, there are some drawbacks to using a CMOS sensor. Until they change the sensor type there will always be limits.
They could use a CCD sensor but then they'd be limited to a max ISO of 1600, but you'd get 12 stops of range.

it can be improved even for CMOS, D7000/D3x/D800 demonstrate this, they use a digital CDS and column parallel on sensor ADCs; apparently there are other ways too, one person claims Canon was offered one but didn't want to bother; the exmor method may be patented

the 5D3 should have similar DR to the D800 at ISO800 and above but lower at ISO 400 and much lower at ISO100-200, this is slightly speculative but pretty likely to bear out

high iso performance may be fairly similar between the 5D3/D4/D3s/D800 with the D4 perhaps taking the crown by 1/3 to 1/2 stop (meaningless to a modest difference), this is getting much more speculative now though but not unreasonable, in some ways the higher MP count of the 5D3 and D800 may make their files in some ways have a nicer looking high iso noise than the D4 and especially D3s perhaps, again that is a bit speculative

 

[LetTheRightLensIn]

So... what are you doing? Blowing the highlights and measuring, blowing the shadows and measuring, and computing the magnitude between them? If you repeat the same calcs for every 5D3 sample do you get the same thing? If you repeat for every 5D2 image do you get the same 11.2? Or are you just measuring the DR of a single image?

I'm not doing anything. I'm using IR's files. Thankfully they blew the highlights on some specular highlights so that is where I got the raw saturation levels from. The dark current noise I measured from the masked area of the file that was cut off from light. it seems to be around +/- .1 stops for across three quick tries on files, doing the same thing my 5D2 values happen, by chance to match DxO exactly to the tenth. different copies might vary +/-.15 or so perhaps unless you got a real weird copy

 

[etto72]

I see many folks in this forum deducing and assuming from pictures that are floating on the web that the DR of the new 5D mk III will be about the same of the mk II
Before giving such a underwhelming pictures think a little..
From DP ratings:
5D - 9.2
1Ds mk III - 11.3 (5D mk II - 11.1)
1D mk IV - 12
5D mk III... ?
As you see there was a constant improvement with each sensor generation(Full frame and 1.3)
the last camera generation (1D mk IV) has 12EV and considering that the new 5D (and 1DX)have the latest technology and that the pixel size of the new cameras is bigger
I expect to see around 12.5-13EV
That would be awesome!
Cheers
Ettore

 

[Curmudgeon]

I see many folks in this forum deducing and assuming from pictures that are floating on the web that the DR of the new 5D mk III will be about the same of the mk II
Before giving such a underwhelming pictures think a little..
From DP ratings:
5D - 9.2
1Ds mk III - 11.3 (5D mk II - 11.1)
1D mk IV - 12
5D mk III... ?
As you see there was a constant improvement with each sensor generation(Full frame and 1.3)
the last camera generation (1D mk IV) has 12EV and considering that the new 5D (and 1DX)have the latest technology and that the pixel size of the new cameras is bigger
I expect to see around 12.5-13EV
That would be awesome!
Cheers
Ettore

That would be awesome. Hope you're right. Can't help wondering why Canon would be keeping such performance a secret.

 

[LetTheRightLensIn]

I see many folks in this forum deducing and assuming from pictures that are floating on the web that the DR of the new 5D mk III will be about the same of the mk II
Before giving such a underwhelming pictures think a little..
From DP ratings:
5D - 9.2
1Ds mk III - 11.3 (5D mk II - 11.1)
1D mk IV - 12
5D mk III... ?
As you see there was a constant improvement with each sensor generation(Full frame and 1.3)
the last camera generation (1D mk IV) has 12EV and considering that the new 5D (and 1DX)have the latest technology and that the pixel size of the new cameras is bigger
I expect to see around 12.5-13EV
That would be awesome!
Cheers
Ettore

I was looking forward and expecting 12.5-13 too, but so far nobody has measured that. Five people have tried and all have gotten it to be about the same as for the 5D2. I just hope that by some miracle the IR got a really bum copy or something, not likely though. Or that the masked area is doing something different this time, not likely though either.

Also you are listed per pixel 100% view DR but if you look at normalized per same print/view DR then the 1Ds3 and 5D2 and 1D4 all have basically the same DR and there was no increase of near 11 to 12 for the 1D4, that is only because you were comparing the per pixel DR. Canon seems to have topped out in what they can do with their processes at ISO DR back with the 1Ds3 release.

I really thought that after so many years and keeping it at 22MP they would have a big break through this time. And when I saw the $500 more than the D800 for less MP then I really, really thought they would have the break through. Seems like not. I was actually horrified when I saw that first guy report the same read noise. I couldn't believe it.

Only chance left is for the 1DX. (IMO it would really suck if they held some special new fab and process and tech for the 1DX only when the 5D3 is more of a landscape cam and could arguably use the DR more than the 1DX if anything)


at high ISO it will probably go up a trace, perhaps even a modestly noticeable 1/2+ stops, compared to the 5D2 and even more compared to the 1Ds3 and the SNR does appear to be nearly 2/3 stops better and the banding less so there are some improvements, but for 3.5 years and same MP and $500 more I thought they'd have a bit more, still it should be close to the best at high iso

 

[parsek]

If the findings of these people [who have done their own DR tests] become verified truth, it is very disappointing performance by Canon. DR is the holy grail at this point in high performance digital imaging IMHO. Low light ISO performance being it´s younger brother.

 

[jrista]

It is a sad state of affairs when presented with direct evidence people just stick their head in the hand and shoot the messenger if the results don't mesh with their hopes, internal belief system, what have you. And not just in this forum.

What direct evidence, man?? You've been regurgitating hearsay, speculation, unverifiable test images with no EXIF data, arguments from DPR between individuals who seem to have long reputations as promoting hearsay and speculation, oh, and a few secretive "tests" that ->you<- have done yourself using some kind of mysterious and questionable method to "measure" DR from a few random sample images (WOEFULLY INADEQUATE to the task of properly measuring DR) on the net...the very vast majority of which have been taken with PRE-PRODUCTION 5D III's.

I don't really know what your goal is here, but your just rumormongering and spreading hearsay. Nothing you have provided thus far is verifiably factual, and there don't even exist any sample images on the net yet that could even be remotely considered adequate to properly "measure" dynamic range, let along start a little jihad against the 5D III and how its a step back from the 5D II in terms of noise and DR. (Something which I think most of us who have looked at 5D III sample images can disagree with without even the need to do any sketchy testing...because so far the sample images have been nothing short of fantastic and very competitive (from an IQ perspective...subject wise, I often wonder who screwed Canon's head on, and why they screwed it on backwards.))

Its time to just wait for the darn thing to actually hit the streets, and let the institutions who have solid reputations for accurately measuring these things do their job. Until then, its ALL JUST SPECULATION!

 

[LetTheRightLensIn]

It is a sad state of affairs when presented with direct evidence people just stick their head in the hand and shoot the messenger if the results don't mesh with their hopes, internal belief system, what have you. And not just in this forum.

What direct evidence, man?? You've been regurgitating hearsay, speculation, unverifiable test images with no EXIF data, arguments from DPR between individuals who seem to have long reputations as promoting hearsay and speculation, oh, and a few secretive "tests" that ->you<- have done yourself using some kind of mysterious and questionable method to "measure" DR from a few random sample images (WOEFULLY INADEQUATE to the task of properly measuring DR) on the net...the very vast majority of which have been taken with PRE-PRODUCTION 5D III's.

I don't really know what your goal is here, but your just rumormongering and spreading hearsay. Nothing you have provided thus far is verifiably factual, and there don't even exist any sample images on the net yet that could even be remotely considered adequate to properly "measure" dynamic range, let along start a little jihad against the 5D III and how its a step back from the 5D II in terms of noise and DR.

Its time to just wait for the darn thing to actually hit the streets, and let the institutions who have solid reputations for accurately measuring these things do their job. Until then, its ALL JUST SPECULATION!

I listed my entire procedure in detail! I wasted like ten minutes typing it out. It's the same thing DxO does. A lead software developer for a RAW converter has done the same test and found the same thing. I gave you the link. Another guy has done the same thing and gotten the same results. I'm pretty sure I put that link in this thread too. PhDs, engineers, lead software developers, detailed procedure. What more do you want?

All you need to measure DR is an area of an image that is totally blown out and an area that is 100% black dark current and the IR files gives us that. Unless IR got a really, really bum copy or they suddenly started making the left mask area perform differently and I hope one or the other is the case but I doubt it, this is pretty much solid results.

My 'woefully inadequate' procedure only happened to match the DxO ISO 100 results for ISO 100 to the tenth of a stop (easily within any sample to sample variation, different parts of even a given sensor can easily be +/- 0.1 difference alone).

Smite away, have fun, whatever.

I hope we are wrong and that the pre-pro unit they got stinks or that they messed with the masking area. Every time we hoped that in the past it didn't turn out to be the case, but I hope this time it will. It's not at all likely. But if I (we) are wrong I'll be jumping for joy.

At least the high ISO should be some degree of better even if not the low.

 

[jrista]

It is a sad state of affairs when presented with direct evidence people just stick their head in the hand and shoot the messenger if the results don't mesh with their hopes, internal belief system, what have you. And not just in this forum.

What direct evidence, man?? You've been regurgitating hearsay, speculation, unverifiable test images with no EXIF data, arguments from DPR between individuals who seem to have long reputations as promoting hearsay and speculation, oh, and a few secretive "tests" that ->you<- have done yourself using some kind of mysterious and questionable method to "measure" DR from a few random sample images (WOEFULLY INADEQUATE to the task of properly measuring DR) on the net...the very vast majority of which have been taken with PRE-PRODUCTION 5D III's.

I don't really know what your goal is here, but your just rumormongering and spreading hearsay. Nothing you have provided thus far is verifiably factual, and there don't even exist any sample images on the net yet that could even be remotely considered adequate to properly "measure" dynamic range, let along start a little jihad against the 5D III and how its a step back from the 5D II in terms of noise and DR.

Its time to just wait for the darn thing to actually hit the streets, and let the institutions who have solid reputations for accurately measuring these things do their job. Until then, its ALL JUST SPECULATION!

Dude I freaking listed my entire procedure in detail! I wasted like ten minutes typing it out. It's the same thing DxO does. A lead software developer for a RAW converter has done the same test and found the same thing. I gave you the link. An engineer has done the same thign and gotten the same results. I'm pretty sure I put that link in this thread too. What more do what me to provide?

All you need to measure DR is an area of an image that is totally blown out and an area that is 100% black dark current and the IR files gives us that.

Done with you, smite away, have fun, whatever.

Does your method involved the use of a specialized test device designed just for the purpose of testing DR, that is finely calibrated to produce consistent test images containing tonal swatches of exact known intensities, lit by a very specific kind of light bulb that is capable of emitting a very specific amount of light, with a very specific color balance, at a very specific intensity? Do you take dozens of sample shots of said test device, evaluate each individual shot to produce very accurate measurements into a data sample set? Do you aggregate the measurements of that data sample set to produce a scientifically accurate mean, standard deviation, ratios, etc.? Do you have your results verified by multiple parties for accuracy? Is your final conclusion based on those verified results? Are all of your test cases for every brand of equipment you test subjected to the EXACT SAME TEST, done in the EXACT SAME WAY, to minimize deviation between the results of various brands, models, and even samples of specific equipment? Do you have a reputation as someone who does these things...a track record of verifiable accuracy?

DXO doesn't just take a sample image and perform some little procedure on it to come up with the dynamic range numbers they publish. Claiming that your results are as accurate as theirs comes with a certain responsibility, and I'm sorry...but you repeatedly claiming that your as accurate and reliable as DXO, well its actually rather sad at this point. As it stands, all we really have at this point is your personal word...but someones word only has meaning to those who have reason to trust it.

You've done some "fiddling", and you have your personal conclusions. As far as I can tell, this supposed "engineer" your a fan of has also done some "fiddling", and also has his own personal conclusions. Speculation and hearsay, but personal conclusions nevertheless. You've been repeating your conclusions all over this forum for a few days now. Its certainly your prerogative to continue, but personally, I would stop comparing my own little home-grown, ad-hoc DR evaluation method to the precision testing performed by DXO...save yourself a little face.

We all know what your opinion is, and I think most of us are content to admire the IQ of the images we've seen so far as being a a competitive improvement over prior Canon cameras, comparable to the competition at the very least...and wait for reputable institutions to provide accurate details about DR, ISO, noise, color fidelity, etc. once these cameras actually hit store shelves.

 

[meli]

Somehow i doubt you would be so sensitive about its methodology and findings if LetTheRightLensIn's thread title was "Zomg 5d3 has 14DR u guys!"

In my case i tend to believe him; his approach -with the amount of data we have so far- seems reasonable, his results fits other people tests, and since Canon itself hasnt bothered to announce something about low iso performance, tends to point in the direction LetTheRightLensIn is.

 

[moreorless]

That would be awesome. Hope you're right. Can't help wondering why Canon would be keeping such performance a secret.

Perhaps similar to the pixel binning issue on video? in order to hype it on one product you need to trash your other products.

In areas such as AF and FPS many users simpley don't need them and ISO tend to be hyped by max numbers rather than pointing out the weakness of lesser models at similar levels. DR and video quality on the other hand are issue that relate to most users and talking openly about weaknesses probabley isnt something there keen on.

 

[altenae]

We are all spending hours and hours on numbers.
Very sad. Very sad.

Some time ago I saw an image taken from a MF and I was blown away.
I checked dxo and the DR was worse than the Sony Nex 10.

There are two things :
Numbers and what your eyes see.

I am going out now to take some pictures.
Or should I stay home to do some more calculations.

Please guys use the camera for where it was made for.

Nice article hereeyes vs. numbers)

http://www.luminous-landscape.com/essays/eyes-vs-numbers.shtml

 

[Loswr]

I listed my entire procedure in detail! I wasted like ten minutes typing it out. It's the same thing DxO does.

Really? It doesn't sound like you did this:

[quote author=DxO Labs]
We perform noise measurement using a transmission target placed on top of a uniform light box. The transmission target, designed by DxO Labs, is made of a plate of thick black plastic with precision-drilled holes which are equipped with a range of neutral density filters designed to absorb the light identically for all wavelengths.

The filters are made of pure optical glass with no structures that can be measured as noise. (While other image quality measurement solutions make use of printed targets, we believe such targets are inappropriate for noise measurement testing, as the intrinsic noise pattern of the print paper may be recorded by the tested camera and then confused with the camera’s own noise pattern.)

We place high-density filters on neighboring positions to limit reciprocal illumination of the patches.
The light box (placed behind the target) is composed of two fluorescent daylight spectrum tubes with a diffusing sheet on top, achieving a perfect uniformity on each filter. The luminance is about 1500cd/m2.

We use filters having different light absorption levels ranging from 0% to 99.99% in order to test across a dynamic range of 4 density steps (= 13.3 f-stops — a dynamic range much greater than today’s digital cameras). When shooting such a chart, the sensor of the camera being tested sees a wide range of light levels, with a 1/10,000 ratio from minimum to maximum. For comparison, a printed target dynamic is typically 2 density steps (6.65 f-stops), which is inadequate to simulate high dynamic range or back-lit scenes.

Each uniform zone on the chart (a “patch”) is measured for luminance (cd/m2) with a certified luminance-meter; then all the values are input into DxO Analyzer software.

Once we measure the target and calibrate the DxO Analyzer software, the selected camera shoots an image of the noise target at different ISO settings, and we measure the noise for each color channel of the target image (R, Gr, Gb, B). We compute the mean gray level and noise values for each patch and for all images shot at different ISO settings. We then interpolate these numerical values for all gray levels to calculate and plot signal-to-noise ratio (SNR) curves, from which DxO Analyzer extracts the SNR 18%, the dynamic range, and the tonal range.
[/quote]

Perhaps the 10 minutes you spent typing out your 'procedure' wasn't the only time you've wasted on this 'analysis'...

 

[jrista]

Somehow i doubt you would be so sensitive about its methodology and findings if LetTheRightLensIn's thread title was "Zomg 5d3 has 14DR u guys!"

In my case i tend to believe him; his approach -with the amount of data we have so far- seems reasonable, his results fits other people tests, and since Canon itself hasnt bothered to announce something about low iso performance, tends to point in the direction LetTheRightLensIn is.

If he insisted his means of arriving at such a conclusion was "exactly the same as DXO's", yes, I'd have issue with it. Its not so much about what he is saying as much as it is about the fact that he is claiming his approach and results are just as good and trustworthy as an institution that has a demonstrated track record of using consistent, repeatable, verifiable techniques to produce valid results.

I'd figure most of the people on this forum realistically expect about a 1 stop improvement in DR, but hope for a full two stop improvement that would maximize the potential of a 14-bit sensor. Realistically, I think the 5D III will exhibit around a 1-stop improvement in DR once tested by DXO (12.7-13 stops). However...thats all just expectation and hope. If someone told me they had run some kind of test that used "masked off" dark pixels from a RAW file and was able to prove with as much accuracy as DXO that the 5D III achieved exactly 14.0 stops of DR...I'd be extremely skeptical. (Having an understanding of the CR2 format, sounds a bit sketchy to me as the masked pixels are intended to be used to set black level by RAW converters, and they are purposely fixed at around 1024, rather than around 0 (don't ask me why Canon does that...its just what they do)...I don't believe they are a valid basis to use for measurement of DR.)

At BEST, I think Canon might be able to achieve the same as Sony, 13.86 stops of DR, which is within the margin of error for 3 electron read noise and quantization error in ADC with a 14-bit sensor. Assuming Canon has not actually moved to a column parallel on-die ADC system in their sensors like Sony did (since Sony currently holds all the patents for such technology), realistically I'd be doubtful that DR will improve much past 13 stops, let alone 13.8. Given Canon's track record with improvements to DR, they have consistently improved it with each successive camera in any one of their lines. Taking the 1D APS-H series as an example using DXO data, the II had 11.1 stops, the III had 11.7 stops, and the IV had 12 stops. Read noise for those cameras was 29.7, 22.3, 16.6 e-, respectively. Canon has shown a notable ability to improve their low ISO dynamic range with reductions in read noise, and they have made some significant claims about improved DR that could bite them in the ass if it turns out to be false, or worse, the 5D III exhibited worse DR than the 5D II.

Realistically, I'd expect the 5D III to approach 13 stops of DR, especially if they reduce read noise to 7D levels (8.7 e-). If Canon has been able to reduce read noise to Sony Exmor levels of about 3 e-, I'd expect the 5D III to push 13.5 stops of DR or more. However its all just educated expectation...even if I thought I could get accurate measurements from .CR2 files from sample images on the net, I wouldn't make any bold claims about my measurements being accurate to any degree, and especially not as accurate as DXO. I have no idea what kind of lighting was used to illuminate any sample image I might get my hands on, whether the camera was a pre-production model or not, whether the exposure settings were set exactly correct to measure DR for the type and intensity of the bulb used to illuminate the scene. So far no sample image on the net that I've come across is really ideal for measuring DR with any degree of accuracy as there are no guarantees about the viability of the information contained within them.

 

[Loswr]

his approach -with the amount of data we have so far- seems reasonable...

It's basing a conclusion on a questionable analysis of a single image. It's like asking one person who they plan to vote for in the next presidential election, then predicting that candidate will be the next President, with the data gathered thus far (n=1), that seems reasonable, too.

Sweeping conclusions based on very limited data, combined with outright false statements ("It's the same thing DxO does"), adds up to a lack of credibility.

 

[traveller]

Oh boy, it's started again!

Guys, I appreciate that people have spent quite a bit of time trying to analyse this, but can we not just hold off on this whole debate for a month until DXO Mark publishes their report on the 5D MkIII's and D800's sensors? That way we won't have to argue over the validity of test images, method etc.

This isn't meant as criticism of anyone, it's just that I don't think it's worth getting riled up at this early stage...

 

[Tijn]

Does your method involved the use of a specialized test device designed just for the purpose of testing DR, that is finely calibrated to produce consistent test images containing tonal swatches of exact known intensities, lit by a very specific kind of light bulb that is capable of emitting a very specific amount of light, with a very specific color balance, at a very specific intensity?

Look, you clearly want to introduce a research hardware setup into this story. You're indicating that his method, lacking all this, is therefore not good enough. What exactly about his method is wrong, if you will, and how is it he's getting consistent results with the RAW files he's been "fiddling" with? Consistent results that are in line with "specialized test device" measurements, even?

Claiming that your results are as accurate as theirs comes with a certain responsibility, and I'm sorry...but you repeatedly claiming that your as accurate and reliable as DXO, well its actually rather sad at this point.

He actually doesn't claim to be reliable, and he hopes he's wrong. He's listed out his entire procedure. We can take it with as much of a grain of salt as we like, as we have all the information. We can see what he did, we can see what he did wrong and we can think of the limitations of his findings ourselves. All of that does not take away the fact that his method of measuring these RAW files comes up with consistent results of the DxO measurements. That means that the chance that his method is "valid" has increased. Consistency is a good thing in different approaches to a similar kind of measurement.

As it stands, all we really have at this point is your personal word...but someones word only has meaning to those who have reason to trust it.

No. He's told you how to do it, and he told you what he found. If you don't trust the man, go do it yourself. It's totally reproducable, which he also noted himself by the way because 4 other bloggers on these things came up with the same results themselves.

We all know what your opinion is

In this discussion between him and you, you're the only one whose level of discussion does not come above that of "opinion". He has no personal opinion, just a method and an interest in the matter. If he made an error, be so kind as to point out that error to him. You're the only opinion here, and that opinion seems to be close to sticking your fingers in your ears and saying lalala I don't hear you; because you have faith in Specialized Test Devices© but not in people doing manual measurements in RAW files and getting the same results. If you don't understand the research methods, I can imagine you only really TRUST the people in white lab coats and the big bleepy complex-looking machines. But nobody's asking you to TRUST this research. You are however trying to discredit it simply for not being a complex-machine lab coat. Give this guy some credit, he's getting consistent results. You clearly don't understand it, so really, you might as well just not comment unless you have something to say that will help us evaluate this method of DR data gathering.

 

[Loswr]

...something to say that will help us evaluate this method of DR data gathering.

Still not sure I'd call it data gathering. As for consistent results, the method guarantees them. As for results that match proper empirical tests, a broken analog clock is correct twice a day. But whatever.

Here's a question: when 'measuring' the dark noise level in the 'side masked area', what is the probability that the numbers provided for the dark signal are meaningful? He reports average values of 1024 for the 5DII and 2048 for the 5DIII - exact 2n values. Personally, I find that suspicious, and it seems more likely that those values are not actual signal, but rathet result from the camera firmware setting those pixels, which are outside the image area, to an arbitrary value as it writes out the .CR2 file. In that case, both the absolute value of the dark signal and the noise of that signal are not valid for DR determination.

 

[altenae]

Numbers , numbers, numbers.
Go outside take pictures.
If you are happy and your customer is happy, then who cares if the DR is 10 or 11.

The same story everytime a new camera is about to release.

 

[Tijn]

It's basing a conclusion on a questionable analysis of a single image. It's like asking one person who they plan to vote for in the next presidential election, then predicting that candidate will be the next President, with the data gathered thus far (n=1), that seems reasonable, too.

Now that's just mean, considering the DR measurement to be a binary value of 1 or 0 in a comparison... At least give a fair example.

Frog population. It's said frogs are heavier on island B than they are on island A. Island A frogs weigh 100g, with a standard deviation of about 2.
Give a guy a frog from island B, have him measure a frog's weight for 15 minutes, and then take his weight measurement to be a representative weight for that type of frog on island B.
Frog weighs 98g. Hey, this frog is almost just as heavy! Therefore, it looks like frogs on Island B aren't heavier at all; perhaps even lighter.

And even with all that, I'm not really sure if his measurements isn't actually more reliable than that.

Perhaps a ruler length measurement would be more fair. A ruler is found in some unknown metrics, but with descernible equally-distanced lines on it. You measure the distance between 10 lines to be 3.22 inches. From this measurement you deduce that for the next ruler of that type you find, the distance will also be 3.22 inches when you measure 10 lines.

For this guy's measurement, he's measuring the darkest bit of a RAW file (masked sensor area) and the brightest bit of the file (blown-out highlight). My question would be: if his masked sensor area measurement is valid (and may be taken as "black") and if his highlight may be taken as valid as well; would there be any significant difference examining 20 images rather than 1? For a RAW file linked to a sensor, isn't "blown-out white" always the same value in the RAW? It's blown out at some value, right? Does that value change from picture to picture?