Where are cameras going with MP's ?

More is almost always better, but the advantages of more keep diminishing.

The Megapixel Race was really settled a few years ago when most cameras were at 16MP or more. Everything beyond that is nice to have, but not really necessary for most amateur shooters.

This was a big deal when hard drive and media storage was expensive, but not so much today now that storage is cheaper and image processing is faster. Higher resolution sensors always demand more processing power from both your camera and your computer. So it was a real issue around ten years ago.

To be clear, there are always advantages to having more resolution. Because it will result in getting better detail, the ability to crop more, and even in better performance in dim light. But that advantage isn't linear. Twice as many pixels doesn't get you twice as much resolution.

It really is our nature to want better, even if that is just a slight improvement. Even if it is some improvement we don't need, or cannot see, we still want better. Sometimes it just fills some psychic need. Like owning a Leica does. (And yes, I do own a Leica M3.)

Do amateur photographers who never print larger than 8x10 really need 36 MP sensors? I doubt they do, but it is their money after all, and they can spend it any way they want to. And to be fair, they really DO pick up some other advantages by having higher resolution sensors.
 
Just a few years ago, I remember seeing guys bashing on cameras because they went up from say 6mp's to 10mp's..... or from 14 to 20...... etc. And the argument was always, "Id rather have better quality MP's than more of them".

Personally, even from the start, I always wanted BOTH :) More MP's of better quality too !

Anyway, skip to now, where we have many choices of more than 30MP's..... and compared to just a few years ago, those are some awesome quality MP's also !!!

So I'm not really hearing too many people complain anymore, about "too many MP's... Low quality... etc, etc"

What's around the corner ? At what point will more MP's (regardless of the IQ) make any difference to our eyes ? How many MP's are required to do a 60" x 80" print, that you can pixel peep from 6", to see the fleck of dust, on the nats leg, on the rose ? :) lol

Okay, I'm being a bit facetious with that example..... So let me ask it this way, at what point will more (and higher quality) MP's start to make no difference to the realism of a photo, in real world viewing situations ?

In 10 or 15 years, will we have GigaPixel cameras ? And if so, how will it matter ?
 
The simple truth is the size of a FF sensor is big enough to contain half a billion photosites with each being tiny enough to produce average quality images but with half a billion pixels.
We can already simulate the center of such an image by using cameras like the Pentax Q with an adapter. You can adapt almost any lens ever made for a larger format on it, if there is enough demand for a particular adapter.

Every EOS-mount lens I have adapted to the Q shows pixel-level details in a pixel density that would be 450MP in FF format, stopped down a little, if not wide-open. The contrast of such details falls below what the finest details would be on a DSLR, but they are details that wouldn't exist at all with much larger pixels, and they buffer the capture from aliasing, because any aliasing that they experience in downsizing is at a low contrast. Strong aliasing does occur on the Q with its better native lenses, because they are much sharper than DSLR lenses on the focal plane.

One thing that we need to keep in mind is that DOF is an abstraction; there is not a range of depth within which everything is equally sharp, and it is just the extremes at which focus drops to a certain threshold. As we resolve lenses further, we find that the depth at which the very finest details are clearly visible is much narrower than the standard-definition "DOF".
A quarter of a billion and they're above average but not great. !00 million and that's two 5DS sensors worth and approaching excellent.

But with 100 million pixel images - say RAW, and using layers as an integral part of your editing regime - you need a fast processor, a lot of RAM, state of the art SSHD or two, and preferably a decent graphics card. And a big monitor, and all this costs in energy, so you want to be able to make a bit of coin from your photographs to ensure you get your money's worth.

But then there's photo-stitching. A 24 MPX camera (with computer) can make a higher resolution image much easier than a 35mm film camera ever could regardless of how well equipped the darkroom. So it's where your priorities lay. Get something that isn't a strain on your present computer and you needn't upgrade, ready yourself for billshock, nor feel obliged to use all those pixels all day/night long.
Stitching has its advantages and disadvantages. One advantage is higher potential image MTF compared to higher pixel density in a single image, but stitching needs a lot more photos than most people use to be done very well, with a lower-res camera, because of under-sampling. Even if you stitch, you are better off with a much higher pixel density, so that you can rotate, distortion-correct, and position the images with no loss of precious acuity. All geometric operations such as resizing, rotating, shifting, distortion and perspective correction, mapping to a sphere, CA correction, etc, are best done on images with very high pixel densities.

We are making big sacrifices in precise imaging with our current coarse, undersampling cameras, heavily limiting practical levels of image display magnification, and crop-ability.

Imaging that captures everything a lens can do on a focal plane, and allows corrections without loss, is very data-intensive, unfortunately, so we settle for artificial, pixelated imaging while we wait for high bandwidth. Even if we don't want our hard disks full of 500MP RAW images, we can use much higher levels of compression on images from higher pixel densities, and cameras could work 500MP under the hood, do its CA and lens correction, and then give us a downsampled image at the desired resolution and compression lossiness. There isn't much worse you can do to an image than shoot it at 16MP, sharp and crispy, correct CA and lens distortion, and then resize it to 7MP for a huge 90-inch 4k display available for close inspection. That is a resampling disaster. 500MP -> 7MP is much better, and the only artifacts it forces is the limitation of 7MP. 16MP -> 7MP makes you choose between increased spatial distortion or loss of acuity.
 
Do amateur photographers who never print larger than 8x10 really need 36 MP sensors?
Correct me if I'm wrong, but my impression is that if you crop to the equivalent of 100mm zoom, which really isn't that much, you will not get a decent print much larger than 4x6.

If that is the case, then what if you need to crop significantly more than 100mm equivalent zoom and want that 8x10?
 
Do amateur photographers who never print larger than 8x10 really need 36 MP sensors?
Correct me if I'm wrong, but my impression is that if you crop to the equivalent of 100mm zoom, which really isn't that much, you will not get a decent print much larger than 4x6.
Depends where you start from. If you have an 85mm lens, then you are still at 26MP after cropping to 100mm equivalent from a 36MP sensor.

If you start at 25mm then you are only left with 2.2MP.

Whether that makes a good 4x6 depends on many factor. There are sharp 2.2MP shots that would make decent 11x14's, honestly.
If that is the case, then what if you need to crop significantly more than 100mm equivalent zoom and want that 8x10?
Why not just buy the right lens for the job?
 
Do amateur photographers who never print larger than 8x10 really need 36 MP sensors?
Correct me if I'm wrong, but my impression is that if you crop to the equivalent of 100mm zoom, which really isn't that much, you will not get a decent print much larger than 4x6.
Depends where you start from. If you have an 85mm lens, then you are still at 26MP after cropping to 100mm equivalent from a 36MP sensor.

If you start at 25mm then you are only left with 2.2MP.

Whether that makes a good 4x6 depends on many factor. There are sharp 2.2MP shots that would make decent 11x14's, honestly.
If that is the case, then what if you need to crop significantly more than 100mm equivalent zoom and want that 8x10?
Why not just buy the right lens for the job?
Thank you.
 
More is almost always better, but the advantages of more keep diminishing.

The Megapixel Race was really settled a few years ago when most cameras were at 16MP or more. Everything beyond that is nice to have, but not really necessary for most amateur shooters.

This was a big deal when hard drive and media storage was expensive, but not so much today now that storage is cheaper and image processing is faster. Higher resolution sensors always demand more processing power from both your camera and your computer. So it was a real issue around ten years ago.

To be clear, there are always advantages to having more resolution. Because it will result in getting better detail, the ability to crop more, and even in better performance in dim light. But that advantage isn't linear. Twice as many pixels doesn't get you twice as much resolution.
Resolution is linear, so it increases with the square root of the number of Megapixels. The height and width figures give a clearer idea.
It really is our nature to want better, even if that is just a slight improvement. Even if it is some improvement we don't need, or cannot see, we still want better. Sometimes it just fills some psychic need. Like owning a Leica does. (And yes, I do own a Leica M3.)
I always had a psychic need to own an M4, but I never had enough psychic money. ;-)
Do amateur photographers who never print larger than 8x10 really need 36 MP sensors? I doubt they do, but it is their money after all, and they can spend it any way they want to. And to be fair, they really DO pick up some other advantages by having higher resolution sensors.
But who only prints 8x10 ? That is a very small print.

In the days of film, my routine size was A4, but I regularly printed 20x16 inch from 35mm. This still seems small compared to typical paintings.
 
The simple truth is the size of a FF sensor is big enough to contain half a billion photosites with each being tiny enough to produce average quality images but with half a billion pixels.
We can already simulate the center of such an image by using cameras like the Pentax Q with an adapter. You can adapt almost any lens ever made for a larger format on it, if there is enough demand for a particular adapter.

Every EOS-mount lens I have adapted to the Q shows pixel-level details in a pixel density that would be 450MP in FF format, stopped down a little, if not wide-open. The contrast of such details falls below what the finest details would be on a DSLR, but they are details that wouldn't exist at all with much larger pixels, and they buffer the capture from aliasing, because any aliasing that they experience in downsizing is at a low contrast. Strong aliasing does occur on the Q with its better native lenses, because they are much sharper than DSLR lenses on the focal plane.

One thing that we need to keep in mind is that DOF is an abstraction; there is not a range of depth within which everything is equally sharp, and it is just the extremes at which focus drops to a certain threshold. As we resolve lenses further, we find that the depth at which the very finest details are clearly visible is much narrower than the standard-definition "DOF".
That is already noticeable on the sdQH, and I am sure it will be more so on a 100 Megapixel camera.
 
Do amateur photographers who never print larger than 8x10 really need 36 MP sensors?
Correct me if I'm wrong, but my impression is that if you crop to the equivalent of 100mm zoom, which really isn't that much, you will not get a decent print much larger than 4x6.
Depends where you start from. If you have an 85mm lens, then you are still at 26MP after cropping to 100mm equivalent from a 36MP sensor.

If you start at 25mm then you are only left with 2.2MP.

Whether that makes a good 4x6 depends on many factor. There are sharp 2.2MP shots that would make decent 11x14's, honestly.
If that is the case, then what if you need to crop significantly more than 100mm equivalent zoom and want that 8x10?
Why not just buy the right lens for the job?
That isn't always affordable, carry-able, or even available.

If the subject is jumping around, you may need to have a smaller sensor area selectively, after exposure.

As an educating experience, you might want to go to a pond with swallows flying around, and use TCs or a superzoom or digital zoom or whatever, and try to keep a swallow in the frame.

There's a whole world of photographing active subjects that will open up when we have 1 gigapixel cameras with very sharp medium-FOV or barely telephoto lenses.
 
More is almost always better, but the advantages of more keep diminishing.

The Megapixel Race was really settled a few years ago when most cameras were at 16MP or more. Everything beyond that is nice to have, but not really necessary for most amateur shooters.

This was a big deal when hard drive and media storage was expensive, but not so much today now that storage is cheaper and image processing is faster. Higher resolution sensors always demand more processing power from both your camera and your computer. So it was a real issue around ten years ago.

To be clear, there are always advantages to having more resolution. Because it will result in getting better detail, the ability to crop more, and even in better performance in dim light. But that advantage isn't linear. Twice as many pixels doesn't get you twice as much resolution.
Resolution is linear, so it increases with the square root of the number of Megapixels. The height and width figures give a clearer idea.
It really is our nature to want better, even if that is just a slight improvement. Even if it is some improvement we don't need, or cannot see, we still want better. Sometimes it just fills some psychic need. Like owning a Leica does. (And yes, I do own a Leica M3.)
I always had a psychic need to own an M4, but I never had enough psychic money. ;-)
Do amateur photographers who never print larger than 8x10 really need 36 MP sensors? I doubt they do, but it is their money after all, and they can spend it any way they want to. And to be fair, they really DO pick up some other advantages by having higher resolution sensors.
But who only prints 8x10 ? That is a very small print.
That was the standard enlargement for 35mm when I was a kid (1970s). I never saw a larger photo print at anyone's house other than rock band posters.
In the days of film, my routine size was A4, but I regularly printed 20x16 inch from 35mm. This still seems small compared to typical paintings.
There is no typical painting size, although there are historically preferred aspect ratios. And if you add in amateurs, who are unlikely to start an 8ft long Battle of Trafalgar epic, small canvas sizes probably outsell large ones by a very wide margin.
 
Do amateur photographers who never print larger than 8x10 really need 36 MP sensors?
Correct me if I'm wrong, but my impression is that if you crop to the equivalent of 100mm zoom, which really isn't that much, you will not get a decent print much larger than 4x6.
Depends where you start from. If you have an 85mm lens, then you are still at 26MP after cropping to 100mm equivalent from a 36MP sensor.

If you start at 25mm then you are only left with 2.2MP.

Whether that makes a good 4x6 depends on many factor. There are sharp 2.2MP shots that would make decent 11x14's, honestly.
If that is the case, then what if you need to crop significantly more than 100mm equivalent zoom and want that 8x10?
Why not just buy the right lens for the job?
That isn't always affordable, carry-able, or even available.

If the subject is jumping around, you may need to have a smaller sensor area selectively, after exposure.

As an educating experience, you might want to go to a pond with swallows flying around, and use TCs or a superzoom or digital zoom or whatever, and try to keep a swallow in the frame.

There's a whole world of photographing active subjects that will open up when we have 1 gigapixel cameras with very sharp medium-FOV or barely telephoto lenses.
You still have the competing struggle of sensor size and noise to deal with. Crop a FF gigapixel image down to 36MP and you have an effective 27x crop sensor with all the noise and artifacts to match.
 
Last edited:
There's a whole world of photographing active subjects that will open up when we have 1 gigapixel cameras with very sharp medium-FOV or barely telephoto lenses.
You still have the competing struggle of sensor size and noise to deal with. Crop a FF gigapixel image down to 36MP and you have an effective 27x crop sensor with all the noise and artifacts to match.
No; that would be equivalent to a 5.27x crop.
 
The extreme version would be something like the Nokia Lumia, which had 41 MP in a 2/3" sensor for 1.12 μm pixel pitch. On full frame, that would be something like 650 MP.
I don't get it. According to this thread, the required thickness of silicon places a practical limit on pixel pitch. At 600 nm wavelength, a thickness of about 5 μm is required for 90% photon absorption. There's no point in having pixels that are much smaller than the layer thickness. I would say 1.12 μm is really pushing it. If the thickness requirement is correct, then 1.12 μm must either be marketing hype, or the quantum efficiency must be low.

For what it's worth, a pixel size of 4 μm corresponds to 50 MP for full frame. Thus, 4 μm seems about optimum to me--within a factor of 2 or 3. :) So yes, you could have 200 to 450 MP images, but you're really going to have a drastically limited return above 50.

20 MP is spectacular right now, but I say the more the merrier. When we get cheap 8"x10" sensors with 5 μm pixels (or even 10), then all will be well. :)
 
Last edited:
There's a whole world of photographing active subjects that will open up when we have 1 gigapixel cameras with very sharp medium-FOV or barely telephoto lenses.
You still have the competing struggle of sensor size and noise to deal with. Crop a FF gigapixel image down to 36MP and you have an effective 27x crop sensor with all the noise and artifacts to match.
No; that would be equivalent to a 5.27x crop.
You are correct, I forgot to square.

That's still down to about 1/1.7" sensor size now, though. Seems like it would be easier to start with a smaller sensor in the first place if you just want free digital zoom.
 
No; that would be equivalent to a 5.27x crop.
You are correct, I forgot to square.

That's still down to about 1/1.7" sensor size now, though. Seems like it would be easier to start with a smaller sensor in the first place if you just want free digital zoom.
How's that swallow doing? Is it still in the frame?
 
No; that would be equivalent to a 5.27x crop.
You are correct, I forgot to square.

That's still down to about 1/1.7" sensor size now, though. Seems like it would be easier to start with a smaller sensor in the first place if you just want free digital zoom.
How's that swallow doing? Is it still in the frame?
I like your point.

But ideally, you'd be better off zooming with a lens than cropping, wouldn't you, because the crop is going to have more noise?
 
No; that would be equivalent to a 5.27x crop.
You are correct, I forgot to square.

That's still down to about 1/1.7" sensor size now, though. Seems like it would be easier to start with a smaller sensor in the first place if you just want free digital zoom.
How's that swallow doing? Is it still in the frame?
I don't know. I'm using a gigapixel camera and this tiny black dot is either the swallow or a spec of dust on the sensor. I guess I'll find out when I get back to the computer and pixel peep. :)
 
How's that swallow doing? Is it still in the frame?
I like your point.

But ideally, you'd be better off zooming with a lens than cropping, wouldn't you, because the crop is going to have more noise?
The amount of light hitting the sensor from the subject from any fixed distance, and with the same shutter speed, is proportional to the area of the physical aperture.
 
How's that swallow doing? Is it still in the frame?
I like your point.

But ideally, you'd be better off zooming with a lens than cropping, wouldn't you, because the crop is going to have more noise?
The amount of light hitting the sensor from the subject from any fixed distance, and with the same shutter speed, is proportional to the area of the physical aperture.
So let's say you have a fixed aperture 18-100mm lens.

The 18mm shot, cropped to 100mm equivalent zoom, is going to have more noise than zooming 100mm with the lens, given the same aperture in both cases, is that correct?
 
Last edited:

Keyboard shortcuts

Back
Top