Canon Lenses not able to handle 50Mpx Resolution?

[Don Haines]

Still with this? Yikes. Glass is optical. There's no hard ceiling in relation to resolving up to some mythical pixel count. A higher resolution sensor will simply allow more detail passed through the glass to be captured, and the comparable results will be relative to what lenses you use. An 18-55 kit lens (pretending it would fill a FF image circle for a moment here) will yield a better result with 50MP capturing it vs 18MP. The same can be said, relatively speaking, of the 24-70 f2.8 L II. Obviously the L will blow the $100 kit lens out the water by light years, BUT both lenses would see improved resolve thanks to a denser sensor with better pixels. Will every lens yield the full 50MP? No. No lens can perfectly resolve all light because as a matter of physics, there is some relative degree of loss as light passes through the elements of glass. However, we've come to a point where great lenses now have been engineered so darn well that the loss is almost unnoticeable to the naked eye, particularly when compared to lens technology of just a few decades ago. As an example, just pull up some DxO scores on Nikon and Sony lenses and look at how close/far the resolve to 36MP on a D810 or A7R. None of them get to 36. There are few who get you into the 30s, but they are $5000 Zeiss primes. Some of Canon's sharpest glass is currently resolving near 20MP on a 5DIII, whose sensor is only 22MP. That's a 90% resolve. That's about as damn good as it gets. So take that same 90% figure and apply it to a 50MP sensor, and I bet we see 40-45MP scores from some L glass on a 5DSR (think 135L 24-70 L II, 70-200 L II, 300 L, etc....)

How dare you bring logic and common sense into an emotional argument!

Seriously though, well said!

 

[zim]

???

Yes, image size set apart, they don't contain more information than what the Sony 36Mp CMOS produces. … That's for the pre-production 5DsR samples that we have seen, and a miracle is always a possibility, if perhaps Canon deliberately clipped the resolution of those samples to hide the potential of the new camera for some obscure marketing reason. But that would be a second miracle. ;D

??? ??? ???

Then again, the 36 MP D810 with Nikon's 24-70 delivers the resolution of the 22 MP 5D3 with Canon's 24-70, so perhaps Nikon deliberately clipped the resolution of their system, too. :

For a simpleton :-[ what does 'clipped the resolution' mean?

resolution: what is resolved on a chart with a specific camera/lens combination?
clipping: use of cheap glass?

 

[tiredofstitching]

??? ??? ???

Then again, the 36 MP D810 with Nikon's 24-70 delivers the resolution of the 22 MP 5D3 with Canon's 24-70, so perhaps Nikon deliberately clipped the resolution of their system, too. :

;D ;D ;D

Lenses put aside, I believe—taken with a ladle—, that a sensor that has no AA filter resolves what an AA or cancelled-effect AA sensor of the larger capacity resolves. That concerns the resolving power, not the richness of tones and smoother color gradients that more pixels can produce. Of course, concerning the 5DsR, it is mere speculation and the future will tell.

 

[AvTvM]

No Canon lens whatsoever will be able "to handle 50MP resolution". So if you plan to buy a 5DS/R, you absolutely have to buy new, high-resolution Nikon lenses and extra high-DR Sony lenses. But don't worry, just send all your current low-res/low-DR Canon lenses to me, and I'll take care of proper waste disposal. Especially any and all Canon L lenses should be dumped "with extreme prejudice". ;D

 

[Don Haines]

At the risk of using DXO data....

Take a kit lens, a good lens, and a fantastic lens on the 7D... let's use the 18-135, the 70-200F4L, and the 600L as our examples...

DXO rates the sharpness of those three as 6Mpx, 9Mpx and 12Mpx....

now go to a bit denser sensor, like the 70D... DXO now rates the sharpness of those three as 7Mpx, 11Mpx and 14Mpx....

now go to a FF sensor, like the 5D3... DXO now rates the sharpness of those three as 11Mpx, 18Mpx and 22Mpx....

As we can be fairly certain that the optical properties of the lens do not change as we go from one camera body to the other, and as we look at more and more examples, it should become fairly obvious that there are three main parameters that affect the sharpness rating... lens quality, sensor size, and sensor density.


We can make a very good estimation of what these sharpness ratings will be on a new 50Mpixel body... The 70D has about the same pixel density as a 50Mpixel FF camera, so the sharpness rating should be 2.56 times the rating for the 70D, or 15Mpx, 28Mpx and 36Mpx.

and comparing the percent improvement from a 5D3 to a 50Mpixel FF body, for the three lenses we get an increase of resolution of 36%, 55%, and 64%. This tells us that the better the lens, the more of an increase in sharpness you will see.

So yes, it does not matter if you have a kit lens or a $15,000 big white, you will see an increase in sharpness by going to a higher megapixel FF body.

 

[AvTvM]

Don, fully agree with your conclusion. However, 18-135 is an EF-S lens - so no use on FF sensor cameras.
Except for "purely theoretical DXO megapickles."

 

[tiredofstitching]

For a simpleton :-[ what does 'clipped the resolution' mean?

resolution: what is resolved on a chart with a specific camera/lens combination?
clipping: use of cheap glass?

The resolution can be clipped by a gross conversion to JPG. Actually, all the image samples that we have seen were in camera JPG's. Imaging Resources had also some RAW samples for a time, but they have been withdrawn. Maybe Canon realized that their in camera algorithm wasn't rape yet and asked for it. I got one RAW when they were available but have no means of opening it, although IR have a report that LibRaw can actually demosaic the 5DsR RAW's and produce more detailed images than the JPG previews that we see on the 5DsR product page, which in itself is normal. To what extent are the RAW's sharper than the JPG's, that's hard to guess. But I doubt that with the presence of an AA filter with effect cancelled, the images will be much more detailed than for instance the 36Mp A7R RAW's. Again, only taking guesses and the proof by the image should come in a short while.

 

[Quasimodo]

No Canon lens whatsoever will be able "to handle 50MP resolution". So if you plan to buy a 5DS/R, you absolutely have to buy new, high-resolution Nikon lenses and extra high-DR Sony lenses. But don't worry, just send all your current low-res/low-DR Canon lenses to me, and I'll take care of proper waste disposal. Especially any and all Canon L lenses should be dumped "with extreme prejudice". ;D

LOL

 

[charlesa]

All I say is wait and say. All this pre-emptive talk of noise, and dynamic range and lenses not up to the job. No one has had a half decent up close meeting with the camera as yet, except Canon insiders.

 

[tiredofstitching]

All I say is wait and say. All this pre-emptive talk of noise, and dynamic range and lenses not up to the job. No one has had a half decent up close meeting with the camera as yet, except Canon insiders.

Absolutely. The day we will see an image sample that approaches the 645Z quality, I will start to seriously drool on the new camera. Other's expectations may vary.

8)

 

[zim]

For a simpleton :-[ what does 'clipped the resolution' mean?

resolution: what is resolved on a chart with a specific camera/lens combination?
clipping: use of cheap glass?

The resolution can be clipped by a gross conversion to JPG. Actually, all the image samples that we have seen were in camera JPG's. Imaging Resources had also some RAW samples for a time, but they have been withdrawn. Maybe Canon realized that their in camera algorithm wasn't rape yet and asked for it. I got one RAW when they were available but have no means of opening it, although IR have a report that LibRaw can actually demosaic the 5DsR RAW's and produce more detailed images than the JPG previews that we see on the 5DsR product page, which in itself is normal. To what extent are the RAW's sharper than the JPG's, that's hard to guess. But I doubt that with the presence of an AA filter with effect cancelled, the images will be much more detailed than for instance the 36Mp A7R RAW's. Again, only taking guesses and the proof by the image should come in a short while.

Ah right, agreed

I think charlesa has it right, "All I say is wait and see", Canon have a bit of a track record of not supplying great initial images, generally the RAWs are also softer until the good RAW converters come on line, causing much whohaa, gnashing of teeth and forum foder

 

[Loswr]

All I say is wait and say. All this pre-emptive talk of noise, and dynamic range and lenses not up to the job. No one has had a half decent up close meeting with the camera as yet, except Canon insiders.

Wait for what? The first lens-cap shot? We know it doesn't have 14-stops of DR, therefore it will have poor IQ and the camera will be a complete fail, further demonstrating Canon's lack of innovation. Boy, if they don't deliver more low ISO DR soon, they're doomed.

 

[AvTvM]

... further demonstrating Canon's lack of innovation. Boy, if they don't deliver more low ISO DR soon, they're doomed.

+1

 

[Hector1970]

How many Dynamic range stops would it need to have to see contract on the lens cap? :

 

[LDS]

Still with this? Yikes. Glass is optical. There's no hard ceiling in relation to resolving up to some mythical pixel count.

Unluckily, there is, because of the very nature of light itself. Those who had worked with telescopes for a long time, where you often greatly enlarge the image coming from the lens/mirror, know it very well. For a given lens diameter (aperture), there's a maximum theoretical resolution which depends on the light wavelength also.
Theoretical angular resolution is proportional to the wavelengh, and inversely proportional to the lens diameter (so a larger lens will *increase* resolution, all other parameters the same).
That's because of diffraction (changes in refractive index like in a lens cause diffraction as well), and any "point" will become an "Airy disc", a central bright small spot sorrounded by rings of light and dark od decreasing brightness. How large these discs are, and how bright are the rings, will led to the "actual" resolution of the lens - when two Airy discs can be perceived as separated. Of course, in any actual lens the actual resolution is less than the theoretical one.
That's one reason astroboffins want larger and larger telescopes - not only you can gather more light, you're also able to obtain a far better resolution. That's why, also, smartphones cameras with their small diameter lenses won't ever be able to compete with the larger DSLR ones.
This about pure angular resolution, then there are several other kind of aberrations that can also reduce resolution. Having pixels smaller and closer than the Airy disc a lens can generate is useless - they will see a single light "point" anyway.
Thereby there will be a lower limit to the pixel size and dimension to be used with a "portable" lens, and lens needs to be designed to exploit the sensor resolving power - but where are actual lenses and sensor I don't know.

 

[BLFPhoto]

How many Dynamic range stops would it need to have to see contract on the lens cap? :

Well, since the 5Ds/5DsR have a high ISO of ONLY 6400, you're going to be forked with this camera anyway in trying to make those important lens cap test images.

 

[Mt Spokane Photography]

???

Yes, image size set apart, they don't contain more information than what the Sony 36Mp CMOS produces. … That's for the pre-production 5DsR samples that we have seen, and a miracle is always a possibility, if perhaps Canon deliberately clipped the resolution of those samples to hide the potential of the new camera for some obscure marketing reason. But that would be a second miracle. ;D

??? ??? ???

Then again, the 36 MP D810 with Nikon's 24-70 delivers the resolution of the 22 MP 5D3 with Canon's 24-70, so perhaps Nikon deliberately clipped the resolution of their system, too. :

For a simpleton :-[ what does 'clipped the resolution' mean?

resolution: what is resolved on a chart with a specific camera/lens combination?
clipping: use of cheap glass?

Resolution is generally a contrast measurement. It can be measured in various ways, it is based on the ability of the eye to detect the difference in dark and light adjacent areas of a image. In its simplest form, a narrow slit behind a bright light is projected thru a lens that is focused onto a sensing medium. The result can be measured to determine contrast (MTF). MTF is the critical concept since it is used to determine the response of a optical system. Read up on it.

Resolution of light waves is analogous to frequency. The higher resolution produces higher frequencies. Low pass filters are used to block the high frequencies in order to prevent Moiré. It is also done in the electronics. So by clipping or filtering out the high frequencies, resolution can be limited. This is good, because it makes a image viewable rather than just a huge rainbow of Moiré.


Its all about the sensor, low pass filter, and the electronics processing the signal. The glass has no problem with 50mp resolutions. It does distort the image which reduces contrast and acts like a low pass filter, but its not a problem with current sensor technology. Maybe 1000 MP might see some effects that outweigh the sensor.

The increased light fall off at the edges of the image is also sensor driven, the angle of incident light increases towards the edge of the sensor so less light is converted to electrons. There are various strategies for minimizing this, but only partially.

 

[Jeffrey]

It still goes back to that if your images look crappy at 15 or 18 megapixels, just imagine how wonderful your images will look at 50 megapixels! I'm not completely sold on the quality of the lens argument.

I really wonder how many amateurs need more than say 15 megapixels for anything they shoot. Instead what they need are additional study/lessons/workshops in order to become very good shooters.

Isn't equipment marketing wonderful!

 

[PureClassA]

Don, did you burst into flames after typing that? ;D

At the risk of using DXO data....

Take a kit lens, a good lens, and a fantastic lens on the 7D... let's use the 18-135, the 70-200F4L, and the 600L as our examples...

DXO rates the sharpness of those three as 6Mpx, 9Mpx and 12Mpx....

now go to a bit denser sensor, like the 70D... DXO now rates the sharpness of those three as 7Mpx, 11Mpx and 14Mpx....

now go to a FF sensor, like the 5D3... DXO now rates the sharpness of those three as 11Mpx, 18Mpx and 22Mpx....

As we can be fairly certain that the optical properties of the lens do not change as we go from one camera body to the other, and as we look at more and more examples, it should become fairly obvious that there are three main parameters that affect the sharpness rating... lens quality, sensor size, and sensor density.


We can make a very good estimation of what these sharpness ratings will be on a new 50Mpixel body... The 70D has about the same pixel density as a 50Mpixel FF camera, so the sharpness rating should be 2.56 times the rating for the 70D, or 15Mpx, 28Mpx and 36Mpx.

and comparing the percent improvement from a 5D3 to a 50Mpixel FF body, for the three lenses we get an increase of resolution of 36%, 55%, and 64%. This tells us that the better the lens, the more of an increase in sharpness you will see.

So yes, it does not matter if you have a kit lens or a $15,000 big white, you will see an increase in sharpness by going to a higher megapixel FF body.

 

[flux capacitor]

Thank you Guys for the tons of informations you provided.