Technically Speaking.....

[fotolopithecus]

<p>Is there any practical limit to how many pixels can be put onto a sensor? If there is, what would that limit be on a DX, and FX sensor? I'm also kind of interested in peoples opinions on at what point you they think there's no longer an advantage at normal to larger print sizes. Aside from myself who else would be willing to trade increasing pixel density for dynamic range. ;)</p>

[SCL]

<p>How many angels can fit on the head of a pin? If technology advances are any indication, I would postulate that the practical limit to pixels on a sensor can't be estimated, as it is a moving target. What we may see today could well be overshadowed in the next year or two as certain experimental aspects of technology come to maturity in production.</p>

[Ed_Ingold]

<p>Noise management has kept pace with increased pixel density (or is it the other way around). Back-illuminated technology represents a giant step in this direction. At present, the only full-frame sensor with back-illumination is in the Sony A7Rii. The noise, hence ISO capabililty was increased from 25,600 to 109,000. images at 25,600 are now somewhat noisy but suitable for publication in print.</p>

<p>The advantage of increased resolution is seen in even small prints or images sized for web publication. Downsampling preserves a lot of detail not found in lower resolution sources - you can enhance the results, but you can't add detail if it's not there. The downside is images take a lot more room (the A7Rii raw files are 84 MB). However, memory is cheap, and you can always down size for printing or transmission.</p>

<p>In terms of theory, the cells cannot be smaller than half the wavelength of light they're intended to capture, nor can they be smaller than the optics used in their manufacturing allow.</p>

[charles_stobbs3]

If you take an acre of land and keep fencing off smaller and smaller plots pretty soon all you have is

fences. I don't know if this analogy applies to sensors and pixels.

[bobatkins]

<p>There is a limit to the number of pixels. If pixels are much smaller than the wavelength of light then they aren't going to respond. There's also a lithography limit, but that's even smaller.</p>

<p>There is no reason why you couldn't put 350MP on a full frame sensor. Performance would be similar to that of one of today's 20MB digicams in terms of dynamic range and noise. If you put down 1 micron square pixels you'd get around 850MP. Technically, that can be done without any advance over current technology.</p>

[fotolopithecus]

<p>Thanks folks, that's interesting stuff. So given what Bob says I expect they're not going to be slowing the pixel wars anytime soon. I was watching a Tony Northrup video the other day in which he suggested that there wasn't any relationship between pixel density, and noise. Basically I think what he was saying is that if two pixels occupy the same area as one larger pixel the noise is the same, because they're collecting the same amount of light in the same area. It makes sense to me, but as Edward's answer implies a lot of people must think otherwise. I'm not sure myself, because there sure seems to be a lot in photography that's counterintuitive.</p>

[yockenwaithe]

<p>This reminds me of when the Leica M8 was considered the pinnacle of sensor technology, at 8mp. Many cameras now have twice of even three times that many. I expect at some point they'll probably incorporate microtechnology with sensors and get some crazy density going on.</p>

[Brian]

<p>At normal print size (and viewing distance) any visible advantage going beyond about 12 MP is very very small. (This of course assumes excellent lenses and technique.) The limiting factor is human visual acuity.</p>

[q.g._de_bakker]

Memories are funny things. The Leica M8 was 10 MP. ;-)<br>It was one of many 10 MP cameras at the time, and i don't remember it being at or even the pinnacle of sensor technology. In fact, there were complaints about sensor related image quality problems.

[AlanKlein]

<p>More pixels give a greater ability to crop. More pixels are limited by the processor in the camera, memory cards, speed of your computer and its memory/storage, etc. As long as people are willing to pay for more pixels, the industry will satisfy the demand. </p>

<p>I just finished making a DVD for slide show on my HDTV at 1920 x 1080 pixels or about 2mp per picture. That's a lot less then my Micro 4/3 12mp camera takes. Most of the pixels were thrown away for the slide show. If you're printing, a decent measure would be 300ppi which means about 7mp for an 8"x10" if I got my calculations right. It's 14mp for 11"x14". 29mp for 16"x20" but you could probably get by with 250ppi which means 20mp for the 16"x20".</p>

<p>I also shoot medium format film. But that's a whole other story. </p>

[fotolopithecus]

<p>I found that Tony Northrup video which is kind of interesting although I don't quite get everthing he's talking about. Maybe if I watch it again. ;)<br>

<a href="http://www.bing.com/videos/search?q=tony+northrup+youtube&view=detail&&&mid=E02CE8FDB976D5192BE1E02CE8FDB976D5192BE1&rvsmid=E02CE8FDB976D5192BE1E02CE8FDB976D5192BE1#view=detail&mid=E02CE8FDB976D5192BE1E02CE8FDB976D5192BE1">http://www.bing.com/videos/search?q=tony+northrup+youtube&view=detail&&&mid=E02CE8FDB976D5192BE1E02CE8FDB976D5192BE1&rvsmid=E02CE8FDB976D5192BE1E02CE8FDB976D5192BE1#view=detail&mid=E02CE8FDB976D5192BE1E02CE8FDB976D5192BE1</a></p>

[Ed_Ingold]

<p>Noise affects individual pixels in a random fashion. With more resolution, the noise is less visible at a given magnification. To reiterate, down sampling (e.g., printing) produces cleaner, sharper results than an original at the lower resolution. That works with video too, for example from 4K to HD or HD to SD. In particular, you have less "staircasing" effect in prominent diagonal lines.</p>

[mark_pierlot]

<p>I'd trade higher dynamic range for pixels any day. I have no issues with the resolutions of my 21 MP Canon 5DII or 18 MP 7D; it's their relatively low DR values (compared with film) that bother me. </p>

[DavidTriplett]

<p>It seems the quality of our lenses must come into play, no? Once the density of pixels on the sensor substantially exceeds the ability of the lens to resolve, any further density is mute. I don't suppose we'll want to start spending NASA-grade money for our lenses. It would be more economical, I suppose, to make larger sensors which keep the pixel density within the range of normal, high-quality glass that we can afford.</p>

[donald_miller5]

I like Bob's thinking at this point as long as we are thinking in terms of pixels and discreet units. At some point , maybe

there will be a new paradigm. Light does have a dual behavior and maybe in the future we will have a technology that will

be based on a constantly variable field and use something that is not predicated on a binary system.