Where will it stop? (megapixels that is)

[johnellisphotography]

Just thinking and wondering where the limit is going to be on DSLR's for MP count. There are physical limits to it and the notion that well have 100MP canon mount cameras in 10 years is probably way off. I got to thinking about this because of two things: 1.) my 1dMKII does everything my film cameras(35mm) did and does it better. so IMHO, 35mm film has been conquered. 2.) some guys I know that upgraded from 1DS to 1DSMKII dont really see that much more resolution, just image size. They think it is becuase of glass limitations.

So, I suspect that this goofy megapixel race and hype will wane soon and some fairly "standard" sizes will settle in for all practical purposes. Canon seems to have settled on the 1.6x crop for the semi pro cameras and probably FF for the 1 series.

Just wanted to get others thoughts on this as I suspect that we'll see this happen in the next couple of years.

[MDJAK]

A year or so back, there was a thought on that by Michael Reichmann, editor of the LuminousLandscape.com, where he basically predicted the 1ds Mark II's resolution and said it won't get much more than that.

In my humble opinion though, just because digital has surpassed film is no reason for it not to continue to get better. After all, film kept getting better.

My question, not to hijack this topic, is exactly how much have sales of film and film-based SLRs dropped in the last year or two. It's almost like turntables: they still sell 'em, but not many.

[Phil V]

this is a technology (miniaturisation) issue, there's no telling where it will end, smarter photosites, micro lenses, faster processors. Unlike sensor size, which is a physical limit, waiting for the fab plants to be good enough for FF sensors to be affordable.

Beating 35mm quality isn't really a benchmark here. Too many photographers are hung up on the size issue, they can't see the resolution issue, even though it's been staring them in the face all these years. Digital cameras will continue to improve until the resolution of a portable camera exceeds 5x4 film cameras.

[Pyromaniac]

The only real limit to how far it will go will be cost, like computer CPU's as the cost to make smaller transitors comes down you see faster computers so I think the same will hold true for digital sensors.

[CyberDyneSystems]

As far as digital resolution goes,. 35mm was the holy grail for years,.. but now that the benchmark has been surpassed,. there are still the larger formats to challenge.

[johnellisphotography]

Good points. I suppose another way to look at is that we currently have the technology to meet and exceed all practical uses for photography, so I guess I'll count us as lucky and even more lucky as it gets better.

[DocFrankenstein]

Yeah... 5 gigapixel large format digital back...

Don't throw that 50kg monster out just yet. :lol:

[steven]

Would there be a limit of megapix due to lens resolution?

[Citizensmith]

Also, I wonder what will come along that will replace CMOS/CCD. Sooner or later there will be another technology along to prove itself. Maybe see a little of the biological in there. Each photosite has its own shutter controlled by artifically grown nano-muscles.

Of course we'll also have cameras automatically deleting photos if they think they suck.

[johnellisphotography]

Of course we'll also have cameras automatically deleting photos if they think they suck.


HAL, youve made a mistake.


Then the camera will sue you for infringing upon its right to tell you what sucks.

[KelliShaver]

Not only that, you may run into problems of your camera sueing for copyright infringment if you go distributing the photos it took without paying it the proper royalties.

And... Oh, hi. I'm new. ;)

[OceanRider]

Lets beat dust as well as megapixels!

[ron chappel]

THERE DEFINITELY IS A MEGAPIXEL LIMIT

It's not talked about much on this forum, but on other more technical minded forums this subject has been explored at length.
The quick answer is that the limit is not a whole lot above what is being currently offered.

The aspects (as i recall) are:

Noise- this is mostly a function of how much light can get to a pixel .Smaller pixels are noisier as we all see from digicams so there is no way they'll give us more (unesessary?) pixels if it makes noise worse!
The things that they use to combat noise are microlenses and onsite noise reduction.
Microlenses are close to being optomised allready and at-the-pixel-site noise reduction is pretty well advanced.There is no doubt more to come but not heaps
Another possible avenue is the use of CYM sensors instead of RGB.I read a kodak paper on this subject-the CYM sensors were more sensitive to light but were harder to manufacture for some reason

a couple of other factors are lens resolution and the focus system's ability to be able to keep a razor sharp lens acurately focused

[Tom W]

Noise is indeed an issue. As the size of individual photosites decreases, noise will increase. This is especially true when the photosite approached the actual wavelength of the light that they are meant to detect. This is a physical barrier that technology may not be able to deal with.

Glass limitations are an issue as well. The 1Ds Mk II is a stickler for good glass - the combination of good pixel density, the ability to utilize the entire lens, and the ability of people to view their images at 100% (the equivalent of a 20 X 30 inch poster or larger) gives us the ability to exploit every weakness in the system.

[FlipsidE]

I think we have a LONG way before it stops. Once the companies begin to close in on the cap for full frame 35mm, I think they'll make the natural jump to Medium Format backs. Medium format cameras aren't as popular and widely used as 35mm, but they aren't to be put by the wayside either. Once the companies start to reach the limits of medium format, then we'll revisit this topic and see what kind of plans there are to make a mass move to large format digital backs. Just my $0.02.

FlipsidE

[Scottes]

Ron Chappel's and Tom W's speak volumes about the inability of technology to overcome several key points. Photosite size simply can't go smaller than the wavelength, so soon we will reach a photosite denity issue. People have been making glass for many, many years and improvements will slow - though there's always the potential for some huge discovery. Glass quality and photosite size are two (the two?) stoppers for increasing megapixels it seems to me.

And what good would it do if the surrounding technology can't compete? Will they soon commonly develop inexpensive printers that can print the number of colors that a 1DsMkII can capture? (Yes, I'm sure that a LightJet can do so...) What about monitors, too? We should be seeing improvements in LCD monitors fairly regularly, but even CRTs are at - or close to - their limits to us digital photographers.


Technology could help the noise issue though, and I'm sure we'll see improvements there. An increase of dynamic range is another thing that I'd like to think is possible - who wouldn't love a 20 megapixel sensor with a true 14-stop range without noise? Color accuracy is pretty darn good but could always use improvement at the extremes. Autofocus accuracy will surely improve over time, especially as computer chips get smaller and faster. Autofocus speed will also improve I'd say, though maybe not as quickly as we'd like. Image Stabilization should also improve and become cheaper as computer chips get faster and cheaper.

I'm not so sure that I expect many more megapixels - well, photosite density may be a better explanation, as sensors could always get larger. And I might not be able to do anything with the extra data anyway. But I could forsee technology advances in other areas that would enhance the photographic experience.

[FlyingPete]

Ron Chappel's and Tom W's speak volumes about the inability of technology to overcome several key points. Photosite size simply can't go smaller than the wavelength, so soon we will reach a photosite denity issue. People have been making glass for many, many years and improvements will slow - though there's always the potential for some huge discovery. Glass quality and photosite size are two (the two?) stoppers for increasing megapixels it seems to me.

Current technology seems to have the pixel limit at around 2.3 microns (a rule of thumb for a while was not to go below 3 microns, but noise reduction methods are getting better), before noise levels become unacceptable, even at this level anything over ISO200 becomes problematic.

Lets assume that a full frame sensor could be made with pixels of that size, that would make a sensor with around 160 Megapixels, but with the severe limitations on practical usability due to noise levels. A 1.6x sensor would weigh in around 64 Megapixels

I wouldn't buy either of those, as Scottes pointed out there are far better areas to develop the technology in:

Technology could help the noise issue though, and I'm sure we'll see improvements there. An increase of dynamic range is another thing that I'd like to think is possible - who wouldn't love a 20 megapixel sensor with a true 14-stop range without noise? Color accuracy is pretty darn good but could always use improvement at the extremes. Autofocus accuracy will surely improve over time, especially as computer chips get smaller and faster. Autofocus speed will also improve I'd say, though maybe not as quickly as we'd like. Image Stabilization should also improve and become cheaper as computer chips get faster and cheaper.

Yep, put my R&D money there! I would love more dynamic range, it is the one thing I miss from using film, and while they are at it throw in a low noise level ISO6400 sensor :D

[Avalonthas]

Technically there is no limit. Right now there is a scientist who has a 3.2 gigapixel (3200 mega pixel) camera he built himself so years down the road im sure there will be consumer grade ones (many years down the road). (http://www.gigapxl.org/)

As for the 100 megapixel cam, it will happen, however not in 10 years, more like 20 years or 25 years. The only advantage of more megapixels is increasing the resolution (the physical size) of ur pictures. Also when u resize a picture the noise is also less visible. So if u have say a 100MP picture and reduce it to the equivalent of 50MP, u will get a very large image with a respectable quality. So the only real advantage i see for such large megapixels is the physical size of the photo. There is only so much detail the human eye can detect and oneday we will surpass it, so the only thing we can do is make pics larger.

[Scottes]

A 1.6x sensor would weigh in around 64 Megapixels
Well that would be pretty serious, but I'm not so sure that is possible. Does the math for that account for distance between pixels? And I would have to think that 64 MP 1.6x would seriously outresolve the glass.

Though I certainly wouldn't mind a 20 MP 2x sensor... Or 3x... So that would be cool.

Still, give me a 12 MP 1.6x sensor with a 300% improvement in noise reduction and a true 10-stop dynamic range and I'd be very happy.

ISO 800 with the current noise level of a 20D at ISO 100? Yeah, I'll take one of those.

[Scottes]

Hmmm... Don't forget about advances in memory and CF (or it's future replacement).

Imagine RAW at 5 frames per second and being able to fill a CF card without filling the buffer... Hell, we'd have a DSLR with video mode after all... :-)

[FlyingPete]

Well that would be pretty serious, but I'm not so sure that is possible. Does the math for that account for distance between pixels? And I would have to think that 64 MP 1.6x would seriously outresolve the glass.

My numbers are purely theoretical, seeing how many pixels of the same size you would fine on a modern compact (in this case a G6) and see how many you can get on a full frame sensor.

e.g. A 35mm frame is 35.8x23.8mm, so that is 852 square mm, the G6 has a sensor 7.18 x 5.32mm, or 32 square mm, it has 7 million pixels, so the full frame is 26 times the are, so it can have 26x7MP, or 182MP. The original maths I used were a little more refined, hence the difference in the two quoted figures.

Don't think it will happen, what is the point, especially since the 1DsMk2 has been rumoured to already out resolve a lot of glass.

Technically there is no limit.
Wrong.

There is a limit, it is called the wavelength of light, a pixel can only be so small, once there the only way up is to build bigger sensors. "You canna break the laws of physics!"

That is why we have electron microscopes that pick up where optical ones leave off as there is a point where you just can't magnify anymore, so you need something smaller to detect, so they replace light with electrons. I suppose you could create an electron digicam, however everything you shoot will have to be gold plated, and the images will be 'black & white' :lol:

[Scottes]

Right now there is a scientist who has a 3.2 gigapixel (3200 mega pixel) camera he built himself so years down the road im sure there will be consumer grade ones (many years down the road). (http://www.gigapxl.org/)

Um, he's using a rather large piece of film in that camera and scanning the pic in. This camera has nothing at all to do with "where the limit is going to be on DSLR's for MP count."

[Scottes]

I suppose you could create an electron digicam, however everything you shoot will have to be gold plated, and the images will be 'black & white'
Maybe Canon's working on a sensor using quarks for photosites...

[Tom W]

Maybe Canon's working on a sensor using quarks for photosites...

Well, Canon has a few quirks already - perhaps they can exchange them for quarks at the prescribed exchange rate.

[CyberDyneSystems]

Hmmmm,.. the Canon "1Q" would be the pro line,. and the "prosumers" would be scrambling after the "TOP-20"

Of course the P&S crowd would be eyeing the new "Charmed elph"

[Gus00]

I think I remember reading somewhere that the cameras on the Mars Landers are something like 300mp? If that's correct then it seems the technology is already here. The deciding factor is the cost.

[FlyingPete]

I think I remember reading somewhere that the cameras on the Mars Landers are something like 300mp? If that's correct then it seems the technology is already here. The deciding factor is the cost.


Hmmm, the raw images I have seen off the NASA web site look more like 300KP. Imagine the data link required to upload that much data back to earth!

[Jesper]

Here's an article about a new CMOS sensor with 2 x 2 micrometer pixels designed by Matsu****a (Panasonic): http://www.panasonic.co.jp/corp/news/official.data/data.dir/en050208-6/en050208-6.html

Those pixels are very small, but they've managed to make the wiring around the pixels smaller and did clever things to reduce the number of transistors necessary to read out the pixels.

Things like that will likely also help other manufacturers to make sensors for digital cameras with more megapixels.

Edit: Hey, that's funny, the letters s, h, i and t are not allowed (they're automatically replaced by asterisks)? Matsu****a? :rolleyes: Matsushi-ta??

[Citizensmith]

Hmmm, the raw images I have seen off the NASA web site look more like 300KP. Imagine the data link required to upload that much data back to earth!

Bandwidth is around 9600baud I think. Don't want to be shifting no 300mpxl images with that. :) That and the fact that everything has to be designed to still work after being torn through by high energy particles for which there is no atmospheric protection. There is a reason a lot of the stuff in space still runs on old technology. It can take more crap than the new stuff.

[Hellashot]

Just thinking and wondering where the limit is going to be on DSLR's for MP count. There are physical limits to it and the notion that well have 100MP canon mount cameras in 10 years is probably way off. I got to thinking about this because of two things: 1.) my 1dMKII does everything my film cameras(35mm) did and does it better. so IMHO, 35mm film has been conquered. 2.) some guys I know that upgraded from 1DS to 1DSMKII dont really see that much more resolution, just image size. They think it is becuase of glass limitations.

So, I suspect that this goofy megapixel race and hype will wane soon and some fairly "standard" sizes will settle in for all practical purposes. Canon seems to have settled on the 1.6x crop for the semi pro cameras and probably FF for the 1 series.


Science and nanotechnology dictate that pixels can get to be extremely small - beyond comprehension. Your only limitations are processor speed, instrumentation size, and memory space. So there is no limitation to maximum megapixel capacity theoretically.

Someday scientists may finally produce an image sensor that is actually finer than the human eye - which at this point is a long way off, on the consumer end at least.

[Longwatcher]

Science and nanotechnology dictate that pixels can get to be extremely small - beyond comprehension. Your only limitations are processor speed, instrumentation size, and memory space. So there is no limitation to maximum megapixel capacity theoretically.

Someday scientists may finally produce an image sensor that is actually finer than the human eye - which at this point is a long way off, on the consumer end at least.

While you will be able to build a bigger sensor (why I have a full frame camera),
Physics will get in the way of smaller pixels.

Visible light is 400 to 750nm so if the detector(pixel) size drops below 750nm you are not picking up red light any more. below 400 and you are not seeing in the visible. The reason storage goes up is because they have moved from red to blue (or 650nm to around 420nm (if I remember right) so they can fit more on a disk because they are using a smaller wavelength. You could go even smaller. The catch is you and I don't have to see the little tiny changes it makes, so they can get away with it. If we want to see thinks as we normally percieve them, then we must keep the sensors detecting in the 400-750nm range. Thus a physics limit. Currently the cameras are at 2,800nm in size so they only have a short way to go to hit the physics limit of red light, which is why the 8MP P&S are so noisy.

It is theoretically possible to detect the passage of a photon rather then the photon itself, but it becomes very, very noisy at that point and being able to determine what the color is by timing the wavelegth will be problematic unless the whole system is designed including lenses to focus the light directly onto the sensor in a choherent manner. Translation: They won't be camera's any more. Also there would be no financial reason for the camera companies to build them.

I suspect that at some point in the future there will be different camera formats
1. add-on auxillary camera (AKA camera phone) probably top out at 6MP.
2. 35mm size format cameras, both P&S and DSLR topping out at 50MP
3. Medium format cameras, topping out at around 400MP
4. large format cameras (think 4x5 size sensor)
5. very large format cameras (think 8x10 or larger size sensor)

The limit is how big is the camera/sensor body/platform. The standard DSLR body can probably handle a larger sensor then 35mm format, but you would have to have different lenses. (I think the Mamiya adopted this route) The sensor from my 1DsMkII would not fit in a standard camera phone. However, there are a few PDA/phones which it might fit in if willing to give up half the space for the camera. Even the 1.6x sensor size is probably a bit too big for a camera phone.

So given that there is a physics limit to the ability to make smaller sensors that still pick up light and there is a size limit to practical use we are approaching the limits on digital sensors for the bottom formats (camera phones and DSLR). We are not there yet, but only a few years out.

just my opinion and research,

[Jesper]

If it were possible to make pixels of 750 x 750 nm...

1 mm = 1,000,000 nm, so along 36 mm you could fit 36,000,000 / 750 = 48,000 pixels and 24 mm = 32,000 pixels. That would make a sensor with 48,000 x 36,000 = 1.536 gigapixels!

Even when the pixels would be much bigger (2 micrometers, such as in that new Panasonic sensor) there would fit 216 million of them on a 36 x 24 mm sensor.

If you'd make a sensor with 2,800 nm pixels (as Longwatcher says) at 36 x 24 mm, it will already have more than 100 MP. So it should already be possible to make such a sensor (it would probably have horrible noise...).

[tim]

I think I remember reading somewhere that the cameras on the Mars Landers are something like 300mp? If that's correct then it seems the technology is already here. The deciding factor is the cost.

IIRC it's more like 1MP, but with a large, good quality sensor.

[GaryTorello]

Gigapixel Project (actually they're talking about 6.3 Gigapixels..)

http://www.gigapxl.com/

[Scottes]

Gigapixel Project (actually they're talking about 6.3 Gigapixels..)

http://www.gigapxl.com/

Again, that's film based.

[Avalonthas]

why dont they just make the censor smaller to fit into a cellphone cam

[Scottes]

why dont they just make the censor smaller to fit into a cellphone cam

Noise, which increases as the size of the photosite decreases and also (I think) increases as the proximity to other photosites decreases. And there's currently a technological limit as to the size of each photosite.

[Tom W]

While you will be able to build a bigger sensor (why I have a full frame camera),
Physics will get in the way of smaller pixels.

Visible light is 400 to 750nm so if the detector(pixel) size drops below 750nm you are not picking up red light any more. below 400 and you are not seeing in the visible. The reason storage goes up is because they have moved from red to blue (or 650nm to around 420nm (if I remember right) so they can fit more on a disk because they are using a smaller wavelength. You could go even smaller. The catch is you and I don't have to see the little tiny changes it makes, so they can get away with it. If we want to see thinks as we normally percieve them, then we must keep the sensors detecting in the 400-750nm range. Thus a physics limit. Currently the cameras are at 2,800nm in size so they only have a short way to go to hit the physics limit of red light, which is why the 8MP P&S are so noisy.

It is theoretically possible to detect the passage of a photon rather then the photon itself, but it becomes very, very noisy at that point and being able to determine what the color is by timing the wavelegth will be problematic unless the whole system is designed including lenses to focus the light directly onto the sensor in a choherent manner. Translation: They won't be camera's any more. Also there would be no financial reason for the camera companies to build them.

I suspect that at some point in the future there will be different camera formats
1. add-on auxillary camera (AKA camera phone) probably top out at 6MP.
2. 35mm size format cameras, both P&S and DSLR topping out at 50MP
3. Medium format cameras, topping out at around 400MP
4. large format cameras (think 4x5 size sensor)
5. very large format cameras (think 8x10 or larger size sensor)

The limit is how big is the camera/sensor body/platform. The standard DSLR body can probably handle a larger sensor then 35mm format, but you would have to have different lenses. (I think the Mamiya adopted this route) The sensor from my 1DsMkII would not fit in a standard camera phone. However, there are a few PDA/phones which it might fit in if willing to give up half the space for the camera. Even the 1.6x sensor size is probably a bit too big for a camera phone.

So given that there is a physics limit to the ability to make smaller sensors that still pick up light and there is a size limit to practical use we are approaching the limits on digital sensors for the bottom formats (camera phones and DSLR). We are not there yet, but only a few years out.

just my opinion and research,

Good post, Longwatcher. I'd venture to add that the sensor's individual pixel needs to be a few times larger than the wavelength of the light it is detecting to allow it to accurately differentiate from other wavelengths (colors).

[tim]

why dont they just make the censor smaller to fit into a cellphone cam

I will never allow a censor (http://photography-on-the.net/forum/newreply.php?do=newreply&p=413188) into my camera! There are cameras now that automagically blur out the faces of celebrities, an up-and-coming technology that almost everyone will disable as soon as they can.

[FlyingPete]

Good post, Longwatcher. I'd venture to add that the sensor's individual pixel needs to be a few times larger than the wavelength of the light it is detecting to allow it to accurately differentiate from other wavelengths (colors).

That nails the fact we are almost a technological limit, the smallest sensors availalbe are only three times the upper wavelength of light (2100nm vs 750nm).

I imagine now that in the compact digicams we might start to see the number of pixels tapper off, unless they enlarge the sensors.

[ShutteringFocus]

Personally I'm holding out for the current technology to decrease in price. Seriously...$8000 for a single camera body...I honestly dont know if its worth all that. That's a lot of money.

But then again...a gallon of gass is costing upwards of $2.50 these days too so who knows...

[Jesper]

Personally I'm holding out for the current technology to decrease in price. Seriously...$8000 for a single camera body...I honestly dont know if its worth all that. That's a lot of money.

But then again...a gallon of gass is costing upwards of $2.50 these days too so who knows...Do you know what gas costs in other countries than the US.... a liter of gas (1 liter = 0.263 gallons) costs about € 1.30 here ($ 1.70) - that's $ 6.46 per gallon !!!!! 8) $ 2.50 is VERY VERY CHEAP !!!! :)

[FlyingPete]

Do you know what gas costs in other countries than the US.... a liter of gas (1 liter = 0.263 gallons) costs about € 1.30 here ($ 1.70) - that's $ 6.46 per gallon !!!!! 8) $ 2.50 is VERY VERY CHEAP !!!! :)


Almost fell over backwards when I looked at the petrol price in Europe when I was there a couple of years ago!

That is a good price, we are over $3US a USGallon here.

[Avalonthas]

gas ranges from 78 to 90 CDN cents a litre here, which is 63 to 74 cents per litre USD, which brings it to $2.39 to $2.81 per gallon USD. It used to be even lower, but we get much more $$$ per barrel selling it to americans, so more profit = better economy.

[griff2]

I think the wavelength of light will be a major limiting factor. Visible light varies in wavelength from 0.38 to 0.73 microns, wheres the photosite on a modern 11Mpixel camera is ~ 9 microns in size. The limiting factor would therefore seem be about 100 Mpixels.


Do you know what gas costs in other countries than the US.... a liter of gas (1 liter = 0.263 gallons) costs about € 1.30 here ($ 1.70) - that's $ 6.46 per gallon !!!!! :cool: $ 2.50 is VERY VERY CHEAP !!!! :)


Here in the UK a litre of petrol costs £0.79, that's £3.00 or $5.68 per gallon!

[tim]

Petrol costs maybe NZ$1.20 per litre here. That's US$0.86 per litre, or US$2.86 per gallon. Is that good or bad?

[griff2]

Compared to UK prices it's good.

[ShutteringFocus]

Lets see...I fill the tank about once a week...so thats 12 gallons a week...at $6 a piece...so thats like what?...$3,744 a year on gas...

Wow. Just walk everywhere and run a lemonade stand for the extra $300 and you could own yourself a new Mrk II in one year!

[FlyingPete]

Petrol costs maybe NZ$1.20 per litre here. That's US$0.86 per litre, or US$2.86 per gallon. Is that good or bad?

Yep thats what my numbers were based on.

[Moppie]

So how many pixels are there in a gallon?

And does it cost you more to fill your camera with pixels in UK or the US?