Maximum Sensor size

[ncarnick]

<p>I currently own a Canon 5D mk ii. With the upcoming release of the 1Dx I've wondered why Canon moved to larger but fewer pixels in the sensor. It seems they will be getting some great low light performance through this decision and beautiful image quality but the question I don't know is are the denser MP sensors coming into severe limitations of older lens designs.<br>

I tend to believe there is plenty of room left in sensor development but other friends seem to believe we are much closer to diminishing returns than I believe.<br>

For my argument: the sensor size on this full frame digital is about 864mm². The new G1x has a slightly under 262mm² sensor shooting 14.3MP (<a href="http://www.petapixel.com/2012/01/09/canon-unveils-the-g1x-a-large-sensor-compact-answer-to-the-mirrorless-craze/">http://www.petapixel.com/2012/01/09/canon-unveils-the-g1x-a-large-sensor-compact-answer-to-the-mirrorless-craze/</a>). Possibly this math is stupid (feel free to tell me how) but it seems the same sensor technology in a ff 35 version would be 3.3X larger or 47MP. Now obviously the math would lead to an even bigger jump if using a lousy camera phone sensor for the basis of multiplication but it seems the G1x has fairly high optical standards. It also seems like a slightly safe guess that the IQ would be aided by using the technology in combination with a fixed 35mm lens (but I really have no idea how good the attached lens is on the G1x).<br>

In any event, I guess the high level question is do we know at what point sensors will be resolving more than the detail provided by the more common 35mm lenses? And, once past that, is there still room for the sensor/digital filters to add substantial value or will the returns diminish?<br>

The reason I ask this group is I'm clearly not that technically gifted and did not see much on the subject matter. Currently there are rumors out for every manufacturer releasing a 30+ MP beast camera but these rumors have been around for awhile and with Canon and Nikon coming out with 16 to 20MP cameras (though possibly targeted for sports and video work in each case) I'm starting to wonder if there is a bottleneck or where the optical bottleneck will likely end up being.</p>

[danield]

<blockquote>

<p>I've wondered why Canon moved to larger but fewer pixels in the sensor</p>

</blockquote>

<p>That's not quite the way things moved. Canon moved to a larger sensor in a camera that does 10 frames per second. It didn't move from 5D to 1Dx - it moved from 1D IV to 1Dx.<br /> The "smaller" resolution in the 1Dx vs 5D is all to do with processing speed. It requires a bit more hardware to process ten 16mp per second than it does to process three 24mp images per second in the 5D. The fact that the 1Dx can (will?) do 10 full frame 16Mp images per second with AF is quite a feat.</p>

[sarah_fox]

<p>There are a few important factors here:</p>

<p>First, a lens for a full frame camera has a much larger image circle than a lens for a compact. If you compare images from both lenses, you're going to see more fine detail in the image of the smaller format lens. However, there's much more image projected by the larger format lens. The end result is that the total detail from the larger format lens, with respect to the entire image circle, equals or perhaps exceeds the total detail from the smaller format lens. The up-shot of this is that smaller format cameras require higher pixel densities to utilize the sharper, but smaller, images of their lenses.</p>

<p>Second, only just so much detail is available in the image. The highest resolution full frame sensors are getting fairly close to tapping out all of the resolution offered by the best lenses being used. That's not to say some lenses don't outresolve current day sensors. They do, and higher pixel densities would make use of their razor sharp images. It's just that 99% of all shooting situations don't render image sharpness that would benefit substantially from higher pixel density.</p>

<p>Third, higher pixel density comes at a cost of less dynamic range. A smaller photosite can't accumulate as much electrical charge. There are also noise problems with smaller pixels. Many will correctly point out that luminance noise, overall, is unaffected when pixel density increases. However, I believe there is a chromic noise (color noise) issue that expresses itself in subtle ways.</p>

<p>Fourth, most photographers are finding that they don't really need gazillion pixel photos to make a great print at almost any size they want. There are many of us who are still pretty happy at 12 MP, given that we can enjoy the other benefits of large photosites.</p>

<p>Finally, most people don't have computers capable of dealing well with gazillion pixel images, particularly if they're going to edit them.</p>

<p>EDIT -- oh, and processing speed, as noted above!</p>

<p>In the end, I think the offerings on the market are more a reflection of market forces and meeting photographers' needs, rather than a reflection of what is and is not possible to manufacture.</p>

[JDMvW]

<p>Daniel and Sarah have it exactly. The big camera is about speed of capture in things like sports environments. Not for the same market, really. In any case, it's not a step back at all.</p>

[ncarnick]

<p>At Dan, sorry I didn't clarify this well. I get that the 1Dx is the next generation of the 1D family (had a 1D mk ii in the past). If Canon does not replace the 1DS this new body is step down in MP even if it is 18.1MP. The 1Dx is specwise in a weird spot as it is slightly more in MP than the 1D mk iv (16.1) but is full frame and therefore slightly lower resolution than the 1.3X crop current 1D. The 1DS mk iii is something like 21.1 MP in pure MP the 1Dx is a drop.<br>

I'm not claiming this is a bad thing. This might be the greatest camera ever. However it isn't certain if this is a replacement for both 1 models (studio and sports) or not. I don't care so much that all of the .rumors web sites have been wrong so far on releases of giant MP cameras. I'm just wondering how close we are to what the lens can give us in terms of resolution.</p>

<p>At Sarah, thanks a ton as well. I know the original question makes me look like a MP snob but the intent is actually more the opposite. It seems at some point the density of the sensor will be recording more than what the lens can deliver. Take my favorite lens from Canon as an example (85 1.2). It's pretty sharp in the center if not shot wide open (but I mostly got it for the bokeh so this was a dumb choice of an example :) ). In any event, it would be interesting to see how close the sensors actually are from a technical standpoint of pulling in more data than what is actually delivered by my favorite lens or possibly vs the standard 35 or 50. </p>

[danield]

<p>Norman, NP, I understand where you are coming from.</p>

<blockquote>

<p>The 1Dx is specwise in a weird spot</p>

</blockquote>

<p>It is, but thinking about it it's not that surprising. Thorough the EOS history Canon has had the habit of creating not like-for-like replacements and the camera lines have been blurred all the time - for instance the 60D is not quite a continuation of the X0D series, the 5D is not quite a replacement of the 3 film series, there are now two "rebel" lines (and there have been in the past too), etc. Nikon does the same thing placing models between the Canon lines, etc. Competion is fierce so thinking laterally is the key.</p>

<blockquote>

<p>It seems at some point the density of the sensor will be recording more than what the lens can deliver.</p>

</blockquote>

<p>That has been the case from the day 1 - but it depends from lens to lens. I think the consensus is that good primes like the 50 and 35mm have a lot more "megapixels" to give than consumer sensors will ever get to - however this applies in the center, at the edges and with the wrong lens it's a different story. If you are curious about the lens resolution limits, there are some interesting discussions in the forums at photozone.de</p>

<p>I think sensor noise will continue to be for some time the most difficult issue that designers will have to battle with, rather than lens resolution. However, it's also fair to say that some of the latest 24mp cameras are also pushing the limits of what some of the more affordable zoom lenses around can do. Higher density sensors will for sure mean more expensive lenses will be required, but also that diffraction effects will negate their benefit even at lower apertures than usual.</p>

[laur1]

<p>From an optical point of view, the first problem that will be hit with larger resolutions is diffraction. See for example the diffraction section and calculator in this article:</p>

<p><a href="http://www.cambridgeincolour.com/tutorials/digital-camera-sensor-size.htm">http://www.cambridgeincolour.com/tutorials/digital-camera-sensor-size.htm</a></p>

<p>High resolutions limit the smallest aperture that you can use on a camera. Set that diffraction calculator for a sensor size and then increase the resolution and watch how the smallest aperture that you can use before diffraction effects show up will decrease.</p>

<p>Noise or DR will not be affected by higher sensor density - they are affected primarily by sensor size and technology. Sure, at pixel level, smaller sensels are noisier, but when comparing images or prints of the same size, there won't be that much of a difference.</p>

<p>So I guess they prefer to put higher resolutions in larger sensors, so as to permit the usability of a larger range of aperture values. Really, the highest resolution is most useful for landscape photography when you would also want to stop down the most.</p>

<p>Also, if you want to produce images with large resolutions, the trend now is to use a regular camera and just combine hundreds of shots - think panorama, but in two dimensions.</p>

[bruce_rubenstein]

<p>The discussion here is is all theoretical, at least for those who understand some theory. In the world where real people, take real pictures, with real cameras and lenses the practical limit of resolution is how accurately the lens is focused. Unless the camera is mounted on a very stable tripod and focusing is manually done, zoomed in with live view the lens will not be focused precisely enough to approach it's maximum resolution. Bigger, better pixels that have better dynamic ranges and lower noise are more useful to photographers who are taking pictures of things other than resolution charts.</p>

<p>For folks who would like to theorize how many pixels can dance on the head of a pin, there are some very knowledgeable individuals who participate in in this sort of thread in the <em>Luminous Landscape</em> forums.</p>

[bob_meyer6]

<p>Wow, Bruce, that's one of the most reasonable and logical posts I've read on any forum. I made a similar point on another forum in a thread about lens tests. Poor focus, camera and subject movement, inadequate DOF, and inadequate sharpening have a far greater impact on image quality and subjective lack of "sharpness" than any variations in sensor or lens (assuming we're talking the current generation of equipment).</p>

[ncarnick]

<p>This doesn't really speak to much on sensors approaching capabilities of lenses as lens rentals didn't have time with the bodies to compare multiple lenses on the d700 and d800. As much as I would have liked to have seen an intro 50mm on each body (just to see if the Zeiss 100mm improvements moving from a d700 to d800 are likely to be applicable across the Nikon mount lens lineup), I still thought the comparisons of the 4 lenses on the d800 were very interesting.<br>

<a href="http://www.lensrentals.com/blog/2012/03/d-resolution-tests">http://www.lensrentals.com/blog/2012/03/d-resolution-tests</a></p>