p.35 #1 · Official: Canon Announces the EOS 6D Mark II
I will be surprised if these initial findings hold up with more testing. But even if they do, it's worth considering that extreme dynamic range only provides a potential benefit in scenarios that fall outside the norm for most shooters. And the most important "test" is your own practice.
To illustrate, in the years I have been using my 6Ds for event work, I cannot recall a single time when I felt let down by the sensor. In contrast, I have struggled mightily with the workflow for the 5D Mark IV, which has the superior sensor on paper.
Compared to my 6D, I find that the 5D IV easily blocks up highlights in sunny conditions--even when correctly exposed. Using modified profiles in ACR (having custom parametric curves) I still have to manhandle the files, shoving the shadows and highlights sliders around a lot more than I am used to with other Canon bodies.
For this reason, the 5D Mark IV is the first Canon model where I have chosen to enable Highlight Tone Priority for normal RAW shooting in daylight. I haven't tried this in years, but I am happy to say that Adobe now handles the tone mapping for HTP RAWs seamlessly and beautifully. This makes the 5D4 files a lot more malleable with a longer tonal curve in the highlights. (Sure, I could underexpose my shots manually, but that would mess up my previews and histograms; using HTP is elegant and more convenient.) I will say that the 5D4 sensor's improved shadow noise and utter lack of banding makes this setting practical to use even up to higher ISOs, a definite plus over earlier models.
So, here we have a conundrum: a supposedly "superior" camera, having a sensor with 2 stops greater dynamic range at ISO 100, still manages to record (absent special handling) images with messed up shoulders in the histogram, while the "inferior" camera creates files that are a breeze to edit.
So I will submit to all of you that you should wait until you can test the output of the 6D Mark II using the tools and workflow you are familiar with. There may be some unexpected surprises. And more importantly, a measured superiority in one characteristic only tells you part of the story. This superiority has to be relevant to and visible in what you shoot for it to have any meaning.
p.35 #2 · Official: Canon Announces the EOS 6D Mark II
Well, it is just like the other "previews" from around the world... I think most of them praised the new body by only reusing the terminology from the PR kit (see TIR or ?t=3m4s" target="_blank" rel="nofollow">this video from DigiDirect). I don't think we can blame them (reviewers or sellers) for trying to build hype around the product and not providing an objective and quantitative analysis.
However, if we find out that these claims were officially made by Canon representatives but the final product doesn't reach the announced performance level, then they would possibly be breaching the trust of the people placing pre-orders... Also, let's keep in mind that there are several outlets reporting this, across Asia, Europe and North America which makes it look like a very official statement.
I guess Canon has finally developed the first Quantum Dynamic Range(TM)... As long as nobody measures it, it might be either extremely large... or completely sub-par.
p.35 #3 · Official: Canon Announces the EOS 6D Mark II
M42Nebula wrote:
Ciriaco, have you included the difference in pixel size in your calculations?
I haven't in mine and I am wondering if this would not help the 6D2 a bit due to the smaller photodiodes :
The 6D has 6.5 microns pixel while 6D2 has 5.67 microns pixel (active area/micro-lensed?). This means a 2*log2(6.5/5.67) = +0.39Ev gain.
The boost would however also apply to the 5D4 with its 5.36 microns pixel : 2*log2(6.5/5.36)=+0.56Ev.
Yes, all values in the graphs are normalized to 8 megapixels, following DXOMark convention in the "print" tab (they also publish the original measure in the "screen" tab). So all cameras are comparable.
But I do the maths in a different way, which yields half the value you estimate (I hope more correct!). For example, the conversion from 6D to 6D2: (how much DR must be added to 6D2 DR at pixel level for a fair comparison with 6D):
log2(sqrt(26mp/20mp)) = +0.19 EV (not 0.39)
This done based on area. I also initially found strange how DXO does the normalization, but it is more easy to understand that they are right with a numerical example in terms of binning: if we bin 2 pixels (halving the resolution) the signal at each binned pixel doubles, while the noise only increases by sqrt(N^2+N^2) = sqrt(2)*N due to physics laws.
initial DR = signal / noise
final DR = (signal * 2) / (noise * sqrt(2))
final DR / initial DR = sqrt(2) = sqrt(resolutions ratio)
That is, the DR only increases log2(sqrt(2)) = +0.5 EV
The 5DS resolution increase was the main trick Canon used to compensate its poor read noise performance and achieve the best Canon sensor at the time ;-)
I think that DXO doesn't directly apply this formula (the screen/print values are not exactly appart by this estimation) but use another way to yield a similar result (since they are within 95% or better).
p.35 #6 · Official: Canon Announces the EOS 6D Mark II
RobDickinson wrote:
OK my camera was a preproduction model apparently, so not sure how relevant my files will be to the final release
I am surprised that they let you share the images on the web. If the DR in 6Dm2 turns out to be totally fine (e.g. 5Dm4 level) then this might have caused them some unnecessary grief.
p.35 #7 · Official: Canon Announces the EOS 6D Mark II
Thank you for the clarification. I understand the rationale behind the area normalization, however I am not sure about the log2(sqrt(res6D2/res6D)) you are using. Shouldn't this also reflect an area normalization rather than a length normalization?
On the +0.5Ev note : could they have forgotten/disable in firmware the average of the left and right photo-diode images (from the Dual-Pixel sensor) and make us miss another +0.5Ev? (even though that looks really stupid)
p.35 #8 · Official: Canon Announces the EOS 6D Mark II
A bit if a relief that there is another potential reason for the apparent lack of improvement. Surely having a number of underperforming examples for people to try out would be a bad idea though?
RobDickinson wrote:
OK my camera was a preproduction model apparently, so not sure how relevant my files will be to the final release
p.35 #9 · Official: Canon Announces the EOS 6D Mark II
RobDickinson wrote:
OK my camera was a preproduction model apparently, so not sure how relevant my files will be to the final release
Rob, thank you for letting us know as soon as you discovered this. I/we know you were simply trying to help the community out by sharing your images but I'm very relieved they were not in fact from a production model.
So can I slide my hope counter up to 0.998%
Time will tell I suppose.
p.35 #10 · Official: Canon Announces the EOS 6D Mark II
RobDickinson wrote:
OK my camera was a preproduction model apparently, so not sure how relevant my files will be to the final release
Well thats good to hear, and even though I remain doubtful that there could be any huge changes from how the sensor works from a preproduction to a production model. I will keep hope alive, but I just find it a bit doubtful that there will be significant changes. Either way, I still want the camera, just trying to work out if its worth the price to jump in sooner rather than later or at all.
But one thing is clear, this whole thing made me realize that preordering was a silly thing for me to do and waiting until reviews of actual production units are out is the smart thing for me. And so I might not be able to have it with the first batch. But I still have a camera to use until then, and this way I am guaranteed to make a better decision once I have all the facts.
p.35 #11 · Official: Canon Announces the EOS 6D Mark II
0.999% thank you very much.
+1 Canonwatch ... while referencing FM, they are parroting the assessment that has been made off the single data point here. I can hear a bunch of folks responding to it as though it is gospel, on the title alone. I wonder who else is latching on to this.
Makes me wonder if the ripple effect will reach Canon, and we get something out of them before models start shipping. I anticipate that B&H will see a volume of cancelled pre-orders and reach out to Canon for assistance to calm storm or right the ship if this is indeed errant information that is being promulgated.
I know we still don't know what the reality is ... but hopefully we can get something sooner rather than later. Although, if this thing goes "viral", Canon might have to drop the release price to rekindle the interest in it. Okay, well maybe not.
BTW ... I just fired off an email (hopefully it's still a good one) to Chuck Westfall to see if he can offer insight, information or empirical evidence / data to this strange, anomaly situation.
Now we just need some more cameras to confirm but as I've said from the beginning I don't see how DR is going to change in a pre-production, sample or release body. Prototype sure but no way Canon would let some sensor they decided not to use get out to the public/media.
p.35 #13 · Official: Canon Announces the EOS 6D Mark II
cgarcia wrote:
Yes, I see I saw this definition just in your site, but I think that without reasoning about why the Circle of Confusion has any relation with the dynamic range. To start this is a subjective parameter (visual acuity, viewing distance and enlargement). Unless the CoC it is indirectly used to refer to the pixel size (but I still don't see why the pixel size by itself should be used to calculate the acceptable noise floor).
...
I think if you read some of the stuff referred to in the Further Reading under the chart you may understand this better.
Start by considering the human vision system as "pixel binning" based on visual acuity (CoC).
Comparing two sensors by simulating two identical sized images viewed at the same distance is objective, not subjective.
(What is subjective is the level of PDR that you find acceptable in your own photography.)
p.35 #14 · Official: Canon Announces the EOS 6D Mark II
Good to hear this data is from a non-production model, as it does give a modicum of hope that there is still a tweak or two coming in the final models. Still seems unlikely to me though, as they wouldn't be handing out substandard preview cameras to pros without an NDA.
Hopefully you won't get into any trouble for posting these up Rob - i can't imagine Canon will be too happy about the negative press.
Despite all this, i still have the 6D2 as my likely next camera. It might not be *everything* i'd hoped for, but it still looks like a damn good camera in my opinon, and certainly not "crippled" as so many like to judge it.
Now we just need some more cameras to confirm but as I've said from the beginning I don't see how DR is going to change in a pre-production, sample or release body. Prototype sure but no way Canon would let some sensor they decided not to use get out to the public/media.
This is really bad news. It would be a step backwards for Canon.
p.35 #16 · Official: Canon Announces the EOS 6D Mark II
Fred Miranda wrote:
This is really bad news. It would be a step backwards for Canon.
And if it is true - though I am not sure it's not simply a storm in a tea cup - I hope 'the community' gives Canon a kick in the pants.
Jul 10, 2017 at 01:43 AM
Guest
Guest
p.35 #17 · Official: Canon Announces the EOS 6D Mark II
It's very strange that the Canon does not care about DR. If I buy a big, expensive camera with not the best autofocus, not the fastest, not the lighest and pocketable - I want quality, good potential for processing, good textures and low artefacts. Some inexpensive crop 1.5 body give a slightly less detail, but who cares when this detail will drown in the noise and artifacts after postprocessing ? And I'm not talking about some extreme pushing shadows. All who works with 5d3 or 5d2 and who need to save large resolution files with clean textures and pure details knows what i talking about.
p.35 #18 · Official: Canon Announces the EOS 6D Mark II
MayaTlab wrote:
This website is absolutely not serious at all in any way. Just as an illustration : they frequently use different raw converters to evaluate DR (for example, they used Fuji's own raw converter to compare the X-pro 2's latitude vs other cameras using LR). Or : they don't normalise ISO comparison shots in raw (i.e., the exposure is different between different cameras). Or : they pretend that they can measure absolute WB when testing strobes (unlikely for various reasons).
Focus numérique has its merits, but not for anything that requires an ounce of rigour.
And they are also testing a pre-production unit with the same firmware as Rob's.
p.35 #19 · Official: Canon Announces the EOS 6D Mark II
Can anyone refer to one single case in the 15 years history of DSLR's where image quality has been different between pre-production units available after announcement, and production units?
Jul 10, 2017 at 03:14 AM
Guest
Guest
p.35 #20 · Official: Canon Announces the EOS 6D Mark II
alundeb wrote:
Can anyone refer to one single case in the 15 years history of DSLR's where image quality has been different between pre-production units available after announcement, and production units?
Maybe a7II but the difference was made by updating release firmware that allowed 14-bit compression instead of 12-bit at release. It's strictly not the case of pre-release / after-release difference.
