richarddd

DxOMark dynamic range for 5D Mark iii is 11.7 Evs and for the E-M5 is 12.3 EVs.

What measure are you using?

cronin85

True

flyboy60

I said the Dynamic Range number is slightly oversimplified, because there is a tradeoff between DR and Response time, and DLSRs - especially the pro bodies, are generally tuned for speed. I was using the overall sensor scores, because it's more representative of overall performance. Canon's full frame cameras get scores from 79-81, while the Olympus E-M5 gets a score of 71. Meanwhile, the Nikon full frame cameras get scores that are mostly in the low-mid 90s.

richarddd

Your original claim, which is what I responded to, was dynamic range, http://www.flyertalk.com/forum/20870482-post262.html

Dynamic range is a highly standard measure, as are related noise measures.

I don't regard DxO's other sensor measures as particularly meaningful, which is a common although not universal view.

According to DxO, "The Sensor Overall Score is an average of the Portrait Score based on color depth, the Landscape Score based on dynamic range, and the Sports Score based on low-light ISO." I don't see response time on the list. Are you talking about how long it takes to read data off the sensor or how quickly the camera operates, such as burst mode?

BTW, you seem to have raised the oversimplification claim in an edit that was after my post, else I would have responded in that post.

flyboy60

It's a holistic system. This is going to be a bit complicated technically, so bear with me: The degree of dynamic range headroom inherent in the sensor (as a result of the size of the pixels) allows the camera designer to trade off absolute dynamic range of the implemented camera for speed if they choose to do so. But it all starts out as a function of how much inherent DR is in the pixels that enables the overall system implementation to be optimized one way or the other. An imaging sensor is constantly collecting light at the pixels when the shutter is open. The camera design basically takes the sensor output and sends it to an A/D converter of some number of bits of resolution and some sampling rate. The sampling algorithm's inherent speed is dictated by how long it takes the A/D converter to make it's complete conversion, one bit at a time - but it can be sampled either after the last bit settles with complete accuracy, or it can be sampled more rapidly, in which case the last bit is as likely to be spurious noise as it is valid data. If the Designer wants Absolute max dynamic range from the system, they will increase the sampling period (thus lowering the sampling rate) to the level where that last bit is reliable data. But if they care more about speed, they may sample at a frequency where the last bit has not settled out yet, and then they discard it. So the system result can either give one bit more of dynamic range - or it can give a higher frame collection speed. That's using the same sensor, with the same fundamentals - it just depends upon which choice the camera designer made in software. And most pro-level DSLRs make the choice in favor of frame rate. I suspect that Olympus went the other way.

So all three of the separate DxO criteria are inter-related. As you get one bit more of color depth along with the greater Dynamic Range, but that comes at the expense of frame rate. Maximizing frame rate costs the system one bit of color depth and DR. The claim that only the Dynamic Range metric of the DxO score matters is not justified on technical merit.

BTW, I made the edit before I read your reply, and you obviously replied before reading my edit. It happens, no harm no foul.

And don't get me wrong - I'm now in the Micro 4/3 camp for my own personal travel use - as I have decided that the weight trade-off is highly desirable vs. my DSLRs.

richarddd

To the extent read-out speed affects DR, it's already reflected in the DR score. Photographers who care about speed can look at a camera's frames per second.

None of the DxO scores cover frames per second. The sports score is something else entirely.

Trying to cram all aspects of sensors performance into a single score is foolish, IMHO. I regard DR as the most meaningful of the three DxO scores. Those who care about other factors can find the best measurements and weightings rather than relying on the single DxO summary.

In any event, we started with the question of whether increasing the number of pixels would hurt performance (all else being equal, including sensor size and efficiency). Are we in agreement that it does not?

flyboy60

No we are certainly not. If you increase the number of pixels on a sensor using the same semiconductor process generation, same number of layers, same lithography feature size, and the same physical sensor size, you will almost certainly diminish the inherent dynamic range of the resultant sensor. For any given process and sensor physical size, there is a knee in the curve for all of the interacting issues, and mostly, the sensor producers are at the knee of the curve for number of pixels for their leading edge products that are in production.

Right now, on the current leading in-production Sony process, we seem to be able to get 16 MP on a 4/3 sensor, and 24 MP on a DX sensor. In full frame, the sensors are less pixel dense, probably because the optics of the lenses or the processor engines may be limiting, and we are at about 36 MP on a full frame sensor using the same process, and those 36 MP are demonstrating higher inherent dynamic range as a result of that lower density, than is being seen in the DX and 4/3 sensors using the same process. And what we see is that the best DX and 4/3 products have similar inherent dynamic range, but the best full frame products using bigger, but less dense sensors from the same technology (and that's what's in the Nikon D800) have higher overall dynamic range.

The next generation of process will allow better sensors in every size - thus again allowing the number of pixels to increase without sacrificing, and maybe even improving dynamic range, but that will only happen due to process improvements, either in lithography feature size, or number of layers, or electrochemical efficiency of the existing process layers.

And you are welcome to continue to misinterpret the DxO metrics as you wish to, but there is a reason that they factor all three elements into the score. It's because the effects of each interact. If that were NOT the case, don't you think DxO would simply report the DR metric as their main score, rather than a combined main score, of which Dynamic Range is but one of three components? I will grant you that the description of what they call their sports score is misleading, because it is predominantly a speed-based metric for the sensors. But the three components are basically speed, color depth, and dynamic range, each of which impacts the others.

richarddd

For pixel density v. noise, I'll again reference http://www.dpreview.com/forums/post/32064270 in which the author explains the issue in great detail

As the author says, "It is often stated that an image sensor with small pixels will create digital images that have worse performance characteristics (more noise, less dynamic range, lower sensitivity, worse color depth, more lens aberrations, worse diffraction, and more motion blur.) than those created by a sensor with large pixels. I disagree."

Anyone interested in DxO can read their explanation of their metrics at http://www.dxomark.com/index.php/About/Sensor-scores

DxO has color depth, dynamic range and low light ISO scores. Low light ISO has nothing to do with how quickly the camera operates.

DxO has three scores because different people care about different things. If you care about DR, then look to the DR score. If you care about speed (in the sense of how quickly the camera operates), then don't look to any of the three DxO scores, as they don't cover the issue.

flyboy60

As I said previously, back in my post #265, the thread that you quoted from DPreview does not fully explain what's going on, and that won't change no matter how many times you quote it.

As for DxO, they also combine all of those three metrics into their single MAIN score, because all three of those metrics impact one another, and EVERYONE who uses a digital camera that DxO is evaluating, is, to some degree or another, affected by EACH of those criteria. And as I stated, their SPORTS score is a function of ISO sensitivity - but ISO sensitivity is DIRECTLY influenced by sampling rate - which is a limiting factor on the frame rate. If the camera design samples the exact same sensor more slowly, the sensitivity, as well as the Dynamic Range will be higher, by a factor of 1 additional bit (a factor of 2). If it is sampled more rapidly, the sensitivity will be lower (it will be halved). Claiming that the Sports/Sensitivity score has nothing to do with how fast the camera can operate is, plain and simply, incorrect. And BTW, that description by the author is by no means in great detail. It is WAY oversimplified. My post #275 attempts to put more detail around it, and there is NOTHING in the link you provided that contradicts that description at all - because they don't go into any sort of detail about what they are actually measuring. But what they are measuring is the span of the output of the sensor's A/D converter at the pixel level - and that is dependent upon the sampling rate.

Only looking at the Dynamic Range is like ONLY evaluating a car's drivetrain by the Engine specs. The Engine cannot operate independent of the transmission and the wheels. But of course, there are still horsepower junkies who will buy the biggest engine even if the power cannot be effectively transmitted to the wheels due to weaknesses in the rest of the drivetrain.

Anyways, let's get back to the point of this whole discussion, and that is, quite simply that, all else being equal with respect to sensor physical size and the process technology with which it was made, the sensor with the higher pixel density WILL, with near certainty have lower dynamic range. And there's nothing in the DxO specs - not the ones YOU quoted, NOR the ones I quoted, that contradict that fact. Because the relevant comparisons show that 16 MP 4/3 sensors and 24 MP DX sensors with the same technology have basically similar DxO scores. And the 36 MP full frame sensors that use the same technology (Nikon's, NOT Canon's) have BETTER DxO scores, because they have lower pixel densities.

ohliuw

Not going anywhere without my amateurish D3100 and my VR18-200

richarddd

If someone can't agree the DxO's dynamic range score is DxO's best measure of dynamic range, there's no point in having a discussion with them.

ohliuw

On a side note, I know people who can take stunning shots with basic point and shoot, whereas I know others who use pro DSLRs and simply suck

flyboy60

I suppose you actually believe that that was my argument. Which is hardly surprising. When someone can't grasp the technical details of an argument and resorts to a misrepresentation of what I said and an ad-hominem attack, then you are quite correct, there is no point in having a discussion with you. That being said, I'm going to try to make this point one last time, simplifying it so that even you, richarddd, can hopefully grasp it.

DxO's dynamic range score only measures dynamic range of the sensor as implemented, and under a particular set of restricted test criteria. What it does NOT measure is the inherent dynamic range that the sensor is capable of overall - Why? because the metric fails to isolate what else is going on that directly affects dynamic range, including the sampling rate - which impacts the ISO sensitivity and the color depth. DxO themselves understand this, even if richarddd does not. That is why DxO reports an aggregate score that also includes these other two factors as the MAIN score for sensor quality.

But back to the main point of the discussion, for the benefit of those that are capable of grasping it: Anyone who actually understands the DxO measurements can plainly see that all else being equal, pixel density DOES affect dynamic range, and that for any specific sensor process technology, the full frame sensors that have lower pixel density also have the greatest dynamic range - more than the 4/3 and DX sensors that have about a 33% higher pixel density.

nkedel

The whole sensor size/pixel sensity thing should come as no surprise to anyone who's ever shot multiple sizes of film. One can do amazing things with 35mm, but within the limits of where you can carry a medium format camera, it's a whole lot easier to get great tonal range and detail out of the much larger film (and it's rather more forgiving of slight variations of focus, too.)

ethan006

Camera picure quality is very important while taking the pictures. If you are going on a trip then you must look for Cyber-shot Digital Camera H200. It is very good.