A7rIII - 70-80 megapixels

[shadowblade]

From Sonyalphrumors

Incredible if it turns out to be true. It's certainly possible, given that Sony's latest lenses are rated to 100MP.

Also, I hope these huge resolution jumps spur on the development of more tilt-shift lenses, better anti-diffraction deconvolution software (including as part of RAW conversion) or the eventual adoption of a Lytro Light Field camera-type design, since depth of field will become a major constraint.

[Christopher]

Why is depth of field the major constraint? First of all nothing changes if the sensor has the same size. Secondly, do we really need the unrealistic dof from a few inches to infinity?

To the sensor. I'm pretty sure we will see it soon in a Sony and Nikon camera.


Christopher Hauser
ch@chauser.eu

[Bo Dez]

It's certainly interesting times. I can't see how medium format can keep up with this sort of pace, if true. But I'm not interested in the current Sony bodies at all. If there is an 80MP Nikon D820 then I would jump in.

as for Lytro, they just announced they are dropping cameras and getting into VR. So, I think that goes to show how much interest there is in that system.

[dwswager]

Why is depth of field the major constraint? First of all nothing changes if the sensor has the same size. Secondly, do we really need the unrealistic dof from a few inches to infinity?

To the sensor. I'm pretty sure we will see it soon in a Sony and Nikon camera.


Christopher Hauser
ch@chauser.eu

While the actual DOF does not change, he may be looking at elargement sizes.  Since more MPs means it can be enlarged to more then the apparent DOF decreases with Enlargement.  The Circle of Confusion chosen for DOF is based on the amount of enlargement (Sensor size and output size).

[Christopher]

True, but when printed it will look the same or better. Or st least from my experience.


Christopher Hauser
ch@chauser.eu

[dwswager]

It's certainly interesting times. I can't see how medium format can keep up with this sort of pace, if true. But I'm not interested in the current Sony bodies at all. If there is an 80MP Nikon D820 then I would jump in.

as for Lytro, they just announced they are dropping cameras and getting into VR. So, I think that goes to show how much interest there is in that system.

There are both benefits and drawbacks to larger and small pixel size on a sensor. But with ever shrinking pixel size come increasing demands on lenses and technique.

[dwswager]

True, but when printed it will look the same or better. Or st least from my experience.


Christopher Hauser
ch@chauser.eu

No it won't.  Every image I take looks wicked sharp on the 3.2" screen until you enlarge it.  Print the same image at 3x (4.5" x 3") and at 20x (30" x 20").  At 3x the whole image might look fairly sharp, but at 20x, not so much.  Of course, we can help it along in post, but it won't make it sharp.

[Christopher]

And? The same image with a lower res camera with the same settings will not look any better. Especially if you go to x20 or more.


Christopher Hauser
ch@chauser.eu

[shadowblade]

And? The same image with a lower res camera with the same settings will not look any better. Especially if you go to x20 or more.


Christopher Hauser
ch@chauser.eu

No, it won't look better.

But I want to make full use of all the megapixels, especially when shooting landscapes at longer focal lengths. And, if I need to shoot at f/32 to make use of all the available resolution, I'd like to be able to remove diffraction-related loss of resolution using software. Diffraction, after all, follows well-known laws of physics, so it can be done.

[dwswager]

And? The same image with a lower res camera with the same settings will not look any better. Especially if you go to x20 or more.

Christopher Hauser
ch@chauser.eu

A 12MP image one might print at 10x while a 75MP one might print at 30x.  That was the point.  With more MP, it is potentially possible to print larger and so DOF becomes a potential constraint depending on the image.

[shadowblade]

There are both benefits and drawbacks to larger and small pixel size on a sensor. But with ever shrinking pixel size come increasing demands on lenses and technique.

Pixel size has no impact on technique. Total pixel count does.

An 80MP full-frame sensor is no more demanding on vibration reduction, focus accuracy, etc. than an 80MP medium-format back at the same angle of view.

[dwswager]

Pixel size has no impact on technique. Total pixel count does.

An 80MP full-frame sensor is no more demanding on vibration reduction, focus accuracy, etc. than an 80MP medium-format back at the same angle of view.

You are correct, but what we are talking about are ever increasing pixel density on the 135 size sensors.  Moving from 12MP, to 24, to 36, to 50 and now potentially 70-80MP.  The sensor remains the same, the pixel size decreases and the pixel count increases.

[Bo Dez]

It only has an effect on technique when taking into account current technology and designs, not suited for high res. Don't expect this to to stay the way it is though.

[Bart_van_der_Wolf]

No, it won't look better.

But I want to make full use of all the megapixels, especially when shooting landscapes at longer focal lengths. And, if I need to shoot at f/32 to make use of all the available resolution, I'd like to be able to remove diffraction-related loss of resolution using software. Diffraction, after all, follows well-known laws of physics, so it can be done.

Hi,

I understand what you are saying, but allow me to underline a few issues and opportunities. A sensor with denser sampling, i.e. more photo-sites per unit area, will extract more resolution from a given lens. The diffraction for a given sensor surface/area will remain the same for a given aperture number, but the per pixel resolution will suffer from more diffraction blur (lower contrast and loss of resolution). Also issues like camera shake become more significant.

However, with more samples of the blur, there are better opportunities for deconvolution software to restore the original signal from the scene before the lens/aperture blurred it, and additionally it reduces aliasing artifacts. Nevertheless, there is a limit to how much diffraction can be restored, and that limit is due to physics that cannot be beaten or improved (unlike the blur).

Assuming Green wavelengths of, say, 555nm and a circular aperture, that means that for e.g. f/32 the physical resolution (where the MTF drops to 0 response) the resolution in limited at 1 / (0.000555 * 32) = 56.3 cycles/mm. That is equal to what an 8.88 micron sensel pitch sensor array achieves, so we might as well use a lower resolution camera, as far as resolution goes.

Cheers,
Bart

[shadowblade]

Hi,

I understand what you are saying, but allow me to underline a few issues and opportunities. A sensor with denser sampling, i.e. more photo-sites per unit area, will extract more resolution from a given lens. The diffraction for a given sensor surface/area will remain the same for a given aperture number, but the per pixel resolution will suffer from more diffraction blur (lower contrast and loss of resolution). Also issues like camera shake become more significant.

However, with more samples of the blur, there are better opportunities for deconvolution software to restore the original signal from the scene before the lens/aperture blurred it, and additionally it reduces aliasing artifacts. Nevertheless, there is a limit to how much diffraction can be restored, and that limit is due to physics that cannot be beaten or improved (unlike the blur).

Assuming Green wavelengths of, say, 555nm and a circular aperture, that means that for e.g. f/32 the physical resolution (where the MTF drops to 0 response) the resolution in limited at 1 / (0.000555 * 32) = 56.3 cycles/mm. That is equal to what an 8.88 micron sensel pitch sensor array achieves, so we might as well use a lower resolution camera, as far as resolution goes.

Cheers,
Bart

Enter new, diffraction-based 'super-lens' technologies based on nano-scale surfaces that can resolve detail beyond the usual diffraction and wavelength-imposed limits.

[Bart_van_der_Wolf]

Just to give an idea, 80 MP on a 36x24mm sensor would translate to 10960 x 7296 pixels, and a photosite pitch of approx. 3.29 micron.

That would produce a maximum resolution of 152 cycles/mm, and an unrecoverable loss of physical resolution at apertures of f/11 or narrower. The first signs of diffraction at the pixel level will become visible at f/4.0 and will gradually increase at narrower apertures.

Cheers,
Bart

[Bart_van_der_Wolf]

Enter new, diffraction-based 'super-lens' technologies based on nano-scale surfaces that can resolve detail beyond the usual diffraction and wavelength-imposed limits.

Diffraction is caused by the diameter of the aperture, not by the lens elements.

Cheers,
Bart

P.S. As a means to reduce the amount of light with a wider aperture (to improve resolution), one could employ Neutral density filters, but that would not have an effect on DOF. DOF requires either narrow apertures, or a different technology like focus bracketing or incident light-angle sensitive sensors.

[shadowblade]

Diffraction is caused by the diameter of the aperture, not by the lens elements.

Cheers,
Bart

In classical optics, yes.

Not so when you bring metamaterials with special optical properties (e.g. negative refractive index) into it. You can bend light around an object, such as the aperture, without diffraction, or correct it optically before it reaches the sensor. Lots of interesting recent (last 5 years) developments in optical materials.

[NancyP]

I still can't wrap my mind around "negative refractive index" - I want to see this in action.

I just keep thinking, FIRST upgrade the computer!     Otherwise processing these big files will feel like using dial-up.

[Paul Roark]

For stitching, etc., whenever there is geometric manipulation of the data, we lose information.  The higher MP count would help preserve what our lens captured. 

My limiting factor with the Sony a7r2 is more from the noise, however.  So, this trade off is what I'll be interested in.

As an example, the image currently on my web page -- http://www.paulroark.com/ -- was hand held (auto bracketed).  Sadly, the hand held shots (at 1000 iso) were not good enough to show all that the lens (Leica apo 135mm) could capture.  So, I took my tripod out and re-did the shot of Jupiter to capture the three of its moons that my spotting scope could (barely) see.  A 100% section of the image is at http://www.paulroark.com/Jupiter-30th-400iso-135mm-Apo-Telyt-at-100pc.jpg .  Yes, three moons of Jupiter show in large prints (20x26 inches minimum).  You just have to love what technology is bringing to us photographers.

Paul
www.PaulRoark.com