When will the law of diminishing returns kick in?

[DocFrankenstein]

I am just wondering. The MP counts will be increasing steadily.

Assuming the 35 mm sensor will dominate the market. When will we have a point that increasing MP won't produce an increase in quality.

I did a really simple calculation and approximated it to be somewhere between 50 and 90 mp. Are there any articles on this? I'm just curious.

[Pekka]

What is the "final quality"? It depends on output media, right?

[DocFrankenstein]

Overall quality of the image. Detail I guess.

There will be a point when increasing MP count will actually decrease the quality because the sensors are too small.

[kfong]

That would the diffraction limit.

With Rayleigh's resolution limit for a 35mm sensor, I estimated the following:
f-number pixel count
1 2.2Gp
4 140Mp
8 35Mp
11 11Mp
That's assuming the lens is perfect, i.e. no spherical aberration, no coma, no chromatic aberration. But I wouldn't hold my breath waiting for a f/1 lens with no SA and CA to appear any time soon.

Ken

I am just wondering. The MP counts will be increasing steadily.

Assuming the 35 mm sensor will dominate the market. When will we have a point that increasing MP won't produce an increase in quality.

[DocFrankenstein]

Cool, I was close to your calculation.

[dtrayers]

That would the diffraction limit.

With Rayleigh's resolution limit for a 35mm sensor, I estimated the following:
f-number pixel count
1 2.2Gp
4 140Mp
8 35Mp
11 11Mp
That's assuming the lens is perfect, i.e. no spherical aberration, no coma, no chromatic aberration. But I wouldn't hold my breath waiting for a f/1 lens with no SA and CA to appear any time soon.



HUH? :?: :?

[Rayz]

The limit is the wavelength of light. Just off the top of my head, I think the wavelength of red light is 0.7 microns. I'm not sure if the efficiency of a photosite deteriorates as this limit is approached, but smaller than this limit would seem to be less efficient. In fact, if it's possible to make an analogy with radio reception, efficiency might increase dramatically as the photosite size approaches the wavelength of light. Current small P&S cameras have around 3 micron photosites, so there's a way to go.

The other side of the equation is lens quality. It seems to be a truism that components in a controlled environment (sort of laboratory conditions), develop more rapidly than components that have to interface with the real world. For example, computer chips, camera sensors, hi fi amplifiers, factory robots, to name a few, develop more rapidly than computer monitors, loudspeakers, camera lenses and automobiles.

All lenses are limited ultimately by diffraction, until someone discovers a way of beating this limitation. But not all lenses are diffraction limited at every F stop. It's very difficult to prodice a 35mm lens that's diffraction limited at f8; even more difficult at f5.6 and well nigh impossible at smaller f stops (but perhaps not with an unlimited budget - I stand to be corrected :D ).

With increasing application of nanotechnology and new materials, it might eventually be possible to produce a lens that is truly diffraction limited at f2. Those 1 micron photosites (a guesstimate of the practical limit) might well become a reality.

One should also bear in mind that the Rayleigh's limit is at an MTF of around 9% (ie. the image has lost 91% of its original contrast). Current digital sensors are incapable of picking up such faint signals. I wonder if a photodetector tuned to the frequency of light would be able to.

We need some clever Physicists to answer such questions.

[ron chappel]

That would the diffraction limit.

With Rayleigh's resolution limit for a 35mm sensor, I estimated the following:
f-number pixel count
1 2.2Gp
4 140Mp
8 35Mp
11 11Mp
That's assuming the lens is perfect, i.e. no spherical aberration, no coma, no chromatic aberration. But I wouldn't hold my breath waiting for a f/1 lens with no SA and CA to appear any time soon.



HUH? :?: :?

Read up on diffraction.It's an interesting subject in the way it relates to perfecting lenses.

Diffraction is something that can't be fixed or corrected for so is pretty much the upper limit in resolution

[chris.bailey]

who is driving the need for greater image sizes?

Professional photographers - probably not as current sizes satisfy their needs for magazine/newspapers for the most part.

Advanced amateurs - possibly on a "ive got more pixels that you" basis but lets face it most of us print out no bigger than A3 for which 10mp is ample. For web based pics (which is the growing media for distribution), 10mp is too many and we downsize them!

I would say therefore that the thing that will limit mp growth is diminishing consumer demand. Development costs for new sensors is huge and the likes of Canon will increasingly find it is difficult to justify R&D costs from sales. I would think this point will be reached long before the laws of physics kick in.

[Cadwell]

The fight for more megapixels will go on and on for a while yet. Why? Marketing. "More megapixels = better", yeah? :roll:

It's the same as "digital zoom" on camcorders. I've got one that claims "1000x digital zoom" or something like that. It's as much use as a chocolate teapot, you get a couple of very large pixels on screen and that's about it... but it's a number that the public can compare. I can look down on the chap next to me who only has 300x digital zoom ;)

[ron chappel]

I dearly hope for -instead of the same dreary increasing pixel count-
that they will make genuine improvements in brightness range and also in-body image stabilization.
Sadly canon and nikon seem to be officially not very interested in the latter :cry: :cry: :cry:

[DocFrankenstein]

who is driving the need for greater image sizes?
Competition between the manufacturers.
Progress...

[wibbly]

That would the diffraction limit.

Where's the market on film and the diffraction limit? As much as film might be further improved over time, isn't the resoltion achievable by all the non-specialist emulsions considered adequate, even by photographers using the most expensive 35mm film bodies & lenses?

If 35mm film resolution *is* less than the diffraction limit, I guess *that* resolution will be enough for digital too. We may find the maket migrates to other "mine's bigger than yours" metrics like max FPS, min shutter lag, max shutter speed, etc, etc, which could all approach minimum real usefulness pretty quickly.

With bodies into £1000's in the UK, many would say we're already well into the general law of diminishing returns... You spend a LOT more to get a little extra. It's just a case of what that little extra is worth to you and your work or hobby.

W

[DocFrankenstein]

Phallometry...

I can't find the dictionary definition for some reason. :?

[imagesense]

I'm really surprised that people are still in a megapixel war. As we have seen very clearly with cameras like the D1x and D2H, megapixels aren't the sole element in resolution quality. There are lots of pros producing stunning images with less megapixels. The manufacturers should be competing on the sensors and their ability to capture data in the best way to maximize the efficiencies of the rest of the gear, like the great optics.

Unfortunately, they are probably finding it easier to market to the public based on the megapixel myth - the higher, the better.
Very few understand the sensor technology to make purchasing decisions based on that criteria.

Lou

[Perfect_10]

The fight for more megapixels will go on and on for a while yet. Why? Marketing. "More megapixels = better", yeah? :roll:

It's the same as "digital zoom" on camcorders. I've got one that claims "1000x digital zoom" or something like that. It's as much use as a chocolate teapot, you get a couple of very large pixels on screen and that's about it... but it's a number that the public can compare. I can look down on the chap next to me who only has 300x digital zoom ;)

More megapixels = better sales :wink:

[Rayz]

Some of you guys seem to be going off at a tangent. We haven't yet got to the stage, and maybe never will, where manufacturers are producing cameras with higher megapixels that do not resolve more detail than previous models with fewer megapixels. Within a given format, so far, the more pixels the better. The 8 megapixel P&S such as the Sony F828 and Minolta A2 produce better image quality than previous 5 megapixel models. And the A2 is also a better camera in other respects with it's image stabilisation and general handling.

It's true that many, even enthusiatic amateurs might not have a printer larger than A3+. So how many pixels do you need to produce an A3+ size print at a recommended 300ppi without interpolation. According to my calculation, around 20-22MP, and that's without any cropping.

Why some of you guys seem to be so derisive of high megapixel counts beats me. Is the D60 not better than the D30? Is the 20D not better than the 10D? Is the 1D mark ll not better than the 1D? Will the 1Ds mark ll not be better than the 1Ds?

[booggerg]

You should really apply the law of diminishing return to lenses.

[DocFrankenstein]

I'm really surprised that people are still in a megapixel war. As we have seen very clearly with cameras like the D1x and D2H, megapixels aren't the sole element in resolution quality. There are lots of pros producing stunning images with less megapixels.
Thanks for the newsflash. :roll:

For me, it's still important. I'd rather have a 20 mp image than a 4 mp, even if it's full frame 35 mm sensor.

[Perfect_10]

I'm really surprised that people are still in a megapixel war. As we have seen very clearly with cameras like the D1x and D2H, megapixels aren't the sole element in resolution quality. There are lots of pros producing stunning images with less megapixels.
Thanks for the newsflash. :roll:

For me, it's still important. I'd rather have a 20 mp image than a 4 mp, even if it's full frame 35 mm sensor.

size isn't everything .. it's quality (less noise) that counts .. :lol:

[SnJPhoto]

At some point when you get past the megapixel issue, ther soon comes the need to process these monster files. I think along with the pixel count an advance in the compression and storage methods is going to have to occur. How many folks can zip through a few hundred 22MP frames for post processing purposes?

I did notice the new 1Ds is using the CR2 format raw file....thank god! Those TIF files were killer slow to process.

Scott

Yep....been there done that.... got the name on the list for the Mk2....

[pcasciola]

At some point when you get past the megapixel issue, ther soon comes the need to process these monster files. I think along with the pixel count an advance in the compression and storage methods is going to have to occur. How many folks can zip through a few hundred 22MP frames for post processing purposes?


I think storage technology is advancing way faster than digital camera technology, so storage will never be an issue for still pictures.

I have over 1TB on my home machine now, and that's only about $600 in hard drives nowadays. That's enough for over 100,000 20D Raw files. Even if we get to 100 Megapixels tomorrow, we'd be able to store about 10,000 RAW (or 50,000 JPEGs) on $600 in hard drives. At ~40MB/sec read rate, even a 22 MP image would only take a fraction of a second to read in, too.

[pradeep1]

I think storage technology is advancing way faster than digital camera technology, so storage will never be an issue for still pictures.

I have over 1TB on my home machine now, and that's only about $600 in hard drives nowadays. That's enough for over 100,000 20D Raw files. Even if we get to 100 Megapixels tomorrow, we'd be able to store about 10,000 RAW (or 50,000 JPEGs) on $600 in hard drives. At ~40MB/sec read rate, even a 22 MP image would only take a fraction of a second to read in, too.

I agree with you. You can put together a 1 terabyte hard drive array on your personal computer for $400 at the current prices (11/2/2004). That price will be cut in half every year (or every six months, more like it) going forward pretty much.

I think the MP race will slow down when we start approaching 35mm film (for full frame sensors) in information capacity. I read somewhere that would be about 30-35MP. After that, the best 35mm lenses cannot resolve any more. Then they'll start layering on the megapixels for each color like the Foveon sensors, so we'll have 100MP sensors, but with 33 MP of constituent colors...or something like that. 8)

You have about as much luck stopping the MP race as you have stopping the GHz (gigahertz) race in computers. A 3 GHz machine is not always faster than a 2 Ghz or even a really decked out 1.7 GHz machine. It is based on so many other factors...including processor pipelines, cache, bus design, memory speed, etc. , etc. But do expect to see 4 GHz computers next year and 5GHz by Christmas next year...and so on and so forth.

Where will the madness end? :)

[imagesense]

For me, it's still important. I'd rather have a 20 mp image than a 4 mp, even if it's full frame 35 mm sensor.


At some point the larger the mp, the slower the shooting unless you have cameras with enormous buffers. Then you will need to store the images on really big compact flash cards. Then you will need super computers to not only upload pictures in a less than a few seconds per image but you will need trilobytes of storage for the large files.

The diminishing returns concept begins when you have to buy ALL new equipment every year just to keep up with the competition. (not only cameras but computers, memory, storage, flash cards, etc, etc.

At some point we will need to decide what is an acceptable photographic image and stay at a plateau for a while. Otherwise photography will price itself out of business. You can only charge so much for a photograph and if you keep upping the ante on photographic tools you will spend more than you make real fast.

Lou

[roanjohn]

I wonder about the same thing too.

Technology is moving so fast.......before you know it, you'll have a 100 MB jpeg file!!! :shock: :shock: :shock:

But this is progress...........and we must adapt.

Ro1

[pradeep1]

I found this interesting article on photo.net:

http://www.photo.net/oped/bobatkins/full_frame.html

Thought you guys might like to read it. Also this article by Mike Johnston:

http://www.luminous-landscape.com/columns/sm-oct-24-04.shtml

and Micheal Reichmann's reply:

http://www.luminous-landscape.com/columns/sm-oct-24-04.shtml#re

I agree with Bob Atkins (first article) and Michael Reichmann

[Andy_T]

Pradeep,

thanks for the links!

OT: Is that Pavitr Prabhakar in your Avatar?

Best regards,
Andy