October 26, 2014, 05:40:41 AM

Author Topic: A Big Megapixel Discussion  (Read 39670 times)

CarlTN

  • Canon EF 300mm f/2.8L IS II
  • *******
  • Posts: 2227
    • View Profile
Re: A Big Megapixel Discussion
« Reply #135 on: June 19, 2013, 02:07:39 PM »
It will be interesting to see if Canon can actually achieve parity with the current SOA sensor, regarding those attributes that people like to argue about.  What would be mind blowing, is if Canon actually exceeds those attributes...regardless of the pixel dimensions.  It seems to me that if the number is closer to 60MP rather than 45, the performance might be more compromised.  Certainly it will be difficult to make use of all that resolution outside the center 50% of the image on most, if not all Canon lenses...even the 24-70 ii.

canon rumors FORUM

Re: A Big Megapixel Discussion
« Reply #135 on: June 19, 2013, 02:07:39 PM »

privatebydesign

  • Canon EF 300mm f/2.8L IS II
  • *******
  • Posts: 2580
  • Ermintrude says "moo"
    • View Profile
Re: A Big Megapixel Discussion
« Reply #136 on: June 19, 2013, 02:49:25 PM »
One of the points that everyone is missing when comparing a very high MP FF camera to MF, is that the latter uses a longer focal length to achieve the same field of view. This is even more pronounced on a LF camera.

Ever wondered why a landscape shot on a 10x8 has a certain je ne sais quoi ? It's not just the resolution.  ;)

The reason 8x10 images are so distinctive has nothing to do with the lenses, which are comparatively poor performers when compared to good 135 format lenses, or resolution, or indeed format size, it is because people only ever shoot black and white images of rocks, dead trees, and swamps with them.

Equivalence takes care of all the rest.  The difference in magnification is taken care of by a larger coc that is needed by the poor lenses. At 350mm for a standard lens you need heaps more aperture to achieve the same dof, a fast lens that is totally unusable wide open is an f6.5. 8X10 cameras and lenses were designed for contact prints, not scanning and enlarging, you need to do a lot of work on a scanned 8X10 sheet to make it look good enlarged.

Now find me one person that can tell the difference between an 8x10 contact sheet print from an 8X10 camera and a 24mm TS-E image from a 5D MkIII shot for equivalence, and I will show you somebody who understands the zone system and tonal gradation. Your "je ne sais quoi" has nothing to do with the lenses or format, it has everything to do with sensors and film.
The best time to plant a tree is twenty-five years ago. The second best time is today.

Pi

  • 1D Mark IV
  • ******
  • Posts: 937
    • View Profile
    • Math and Photography
Re: A Big Megapixel Discussion
« Reply #137 on: June 19, 2013, 02:57:32 PM »
What's the relationship between focal length and je ne sais quoi?   ;)

He did tell you that he did not know...  ;)

unfocused

  • Canon EF 300mm f/2.8L IS II
  • *******
  • Posts: 2189
    • View Profile
    • Unfocused: A photo website
Re: A Big Megapixel Discussion
« Reply #138 on: June 19, 2013, 03:13:49 PM »
I have to wonder if the convergence of technologies may soon make the big megapixel discussion obsolete or at least more complicated.

Software solutions for upscaling are becoming extremely sophisticated. I recently used On-One's Perfect Resize on some cropped 7D images that I wanted to print at 20 x 30. I was amazed at the quality. Adobe's latest Photoshop CC has a new resizing engine as well.

In theory, upscaling may not be as good as having the original image shot at the desired resolution, but "in theory" doesn't always match real world practice, especially when it is impossible to tell the difference in the final print.
pictures sharp. life not so much. www.unfocusedmg.com

Pi

  • 1D Mark IV
  • ******
  • Posts: 937
    • View Profile
    • Math and Photography
Re: A Big Megapixel Discussion
« Reply #139 on: June 19, 2013, 06:20:37 PM »
I have to wonder if the convergence of technologies may soon make the big megapixel discussion obsolete or at least more complicated.

Software solutions for upscaling are becoming extremely sophisticated. I recently used On-One's Perfect Resize on some cropped 7D images that I wanted to print at 20 x 30. I was amazed at the quality. Adobe's latest Photoshop CC has a new resizing engine as well.

In theory, upscaling may not be as good as having the original image shot at the desired resolution, but "in theory" doesn't always match real world practice, especially when it is impossible to tell the difference in the final print.

No mathematical trickery can remedy lack of information.  Cornelius Lanczos

jrista

  • Canon EF 400mm f/2.8L IS II
  • *******
  • Posts: 4484
  • EOL
    • View Profile
    • Nature Photography
Re: A Big Megapixel Discussion
« Reply #140 on: June 19, 2013, 09:26:32 PM »
It will be interesting to see if Canon can actually achieve parity with the current SOA sensor, regarding those attributes that people like to argue about.  What would be mind blowing, is if Canon actually exceeds those attributes...regardless of the pixel dimensions.  It seems to me that if the number is closer to 60MP rather than 45, the performance might be more compromised.  Certainly it will be difficult to make use of all that resolution outside the center 50% of the image on most, if not all Canon lenses...even the 24-70 ii.

Remember that total system resolution is effectively (closely approximated by) the root mean square of the resolution of each component that makes up the system. In a DSLR, to keep things simple, the final resolution of the photographs you make is the RMS of the resolutions of the lens and the sensor. There is no such thing as one outresolving the other. Increasing the resolution of either lens or sensor increases the resolution of the system as a whole, and produces higher resolution photographs.

You get the most bang for the buck by increasing the lowest common denominator, but if you have a lens, like the 24-70 II, and you use it on a 60mp FF camera...you WILL realize better results (all other things being equal...i.e. assuming the best tech available is used to produce said 60mp sensor.)
« Last Edit: June 19, 2013, 09:29:07 PM by jrista »

CarlTN

  • Canon EF 300mm f/2.8L IS II
  • *******
  • Posts: 2227
    • View Profile
Re: A Big Megapixel Discussion
« Reply #141 on: June 20, 2013, 02:48:36 PM »
It will be interesting to see if Canon can actually achieve parity with the current SOA sensor, regarding those attributes that people like to argue about.  What would be mind blowing, is if Canon actually exceeds those attributes...regardless of the pixel dimensions.  It seems to me that if the number is closer to 60MP rather than 45, the performance might be more compromised.  Certainly it will be difficult to make use of all that resolution outside the center 50% of the image on most, if not all Canon lenses...even the 24-70 ii.

Remember that total system resolution is effectively (closely approximated by) the root mean square of the resolution of each component that makes up the system. In a DSLR, to keep things simple, the final resolution of the photographs you make is the RMS of the resolutions of the lens and the sensor. There is no such thing as one outresolving the other. Increasing the resolution of either lens or sensor increases the resolution of the system as a whole, and produces higher resolution photographs.

You get the most bang for the buck by increasing the lowest common denominator, but if you have a lens, like the 24-70 II, and you use it on a 60mp FF camera...you WILL realize better results (all other things being equal...i.e. assuming the best tech available is used to produce said 60mp sensor.)

That was not my point.  My point was as stated.  I never said the overall results would not be "better".  To belabor my point, since you are intentionally missing it...I will quote myself: "...it will be difficult to make use of all that resolution outside the center 50% of the image on most, if not all Canon lenses...even the 24-70 ii." 

I stand by this.  Your point does not disprove my point.  You might have your own idea about how you define the phrase "make use of all that resolution".  I have mine.  My point was never that you could not get improved resolution and image quality, from a higher megapixel sensor.  Only a fool would argue that.  Yet you seem to want to believe that's what I meant.  I wonder why?  Up to your same old tricks I see.

As for "bang for buck", that is an entirely separate issue altogether, which I hope you realize...and has nothing to do what my point.

canon rumors FORUM

Re: A Big Megapixel Discussion
« Reply #141 on: June 20, 2013, 02:48:36 PM »

jrista

  • Canon EF 400mm f/2.8L IS II
  • *******
  • Posts: 4484
  • EOL
    • View Profile
    • Nature Photography
Re: A Big Megapixel Discussion
« Reply #142 on: June 20, 2013, 08:51:41 PM »
It will be interesting to see if Canon can actually achieve parity with the current SOA sensor, regarding those attributes that people like to argue about.  What would be mind blowing, is if Canon actually exceeds those attributes...regardless of the pixel dimensions.  It seems to me that if the number is closer to 60MP rather than 45, the performance might be more compromised.  Certainly it will be difficult to make use of all that resolution outside the center 50% of the image on most, if not all Canon lenses...even the 24-70 ii.

Remember that total system resolution is effectively (closely approximated by) the root mean square of the resolution of each component that makes up the system. In a DSLR, to keep things simple, the final resolution of the photographs you make is the RMS of the resolutions of the lens and the sensor. There is no such thing as one outresolving the other. Increasing the resolution of either lens or sensor increases the resolution of the system as a whole, and produces higher resolution photographs.

You get the most bang for the buck by increasing the lowest common denominator, but if you have a lens, like the 24-70 II, and you use it on a 60mp FF camera...you WILL realize better results (all other things being equal...i.e. assuming the best tech available is used to produce said 60mp sensor.)

That was not my point.  My point was as stated.  I never said the overall results would not be "better".  To belabor my point, since you are intentionally missing it...I will quote myself: "...it will be difficult to make use of all that resolution outside the center 50% of the image on most, if not all Canon lenses...even the 24-70 ii." 

I stand by this.  Your point does not disprove my point.  You might have your own idea about how you define the phrase "make use of all that resolution".  I have mine.  My point was never that you could not get improved resolution and image quality, from a higher megapixel sensor.  Only a fool would argue that.  Yet you seem to want to believe that's what I meant.  I wonder why?  Up to your same old tricks I see.

As for "bang for buck", that is an entirely separate issue altogether, which I hope you realize...and has nothing to do what my point.

I am not intentionally trying to misquote you... Simple matter of the facts.

I does not matter whether you are talking about resolution at the center of the frame, edge of the frame, or corner of the frame. The basis of "system resolution", which is a convolution of the effects of each and every component, holds true regardless of which region of the frame you apply it to. Sure, poorer quality lenses and wider angle lenses tend to have more detractors to resolution in the corners. That does not mean that suddenly the rules that govern overall system resolution change. Just as much as a higher resolution sensor will improve the outcome of what the lens resolves at the center, so too will it improve the outcome of what the lens resolves at the edge. A higher resolution sensor can never produce WORSE results than a lower resolution sensor, all else being equal.

A horrible lens is a horrible lens, and while you might see marginal improvements in corner resolution with a higher resolution sensor, you experience diminishing returns. An excellent lens, such as the 24-70 II, which performs quite well in the corners, will realize a greater benefit from a move to a higher resolution lens than, say, the 16-35 II (which performs only moderately well in the corners), both of which would benefit considerably better than say the EF-S 18-55mm, which performs terribly in the corners.

The benefit boils down to a matter of degree for every component involved, not whether or not you get any benefit at all outside of the center of the lens. You don't "make use" of resolution...you convolve a result via a functional process as a real-world image passes through each and every lens element, the aperture, the sensor's filter stack, the CFA, and even the pixel well itself. A higher resolution sensor, assuming equivalence in terms of noise, could never "compromise" IQ in any way. Even with a relative increase in noise, a higher resolution sensor, when its image size is normalized to that of any lower resolution sensor, would still produce results that are as good as or better. (The only time I believe a higher resolution sensor can be detrimental to IQ is when there is a disproportionate increase in the amount of noise due to smaller pixels, which to some degree is the case with the 7D (a fact I blame on Canon's 500nm process, which wastes a lot of photodiode space resulting in disproportionately smaller light sensitive photodiode area relative to sensors with larger pixels...a defect I believe a move to a smaller process, such as 180nm, can resolve.))

A far greater concern, in my opinion, for high density sensors than "making use of" any given lenses resolving power would be avoiding softening from camera shake. As pixel sizes shrink for both APS-C and FF sensors, the effects of camera shake will become increasingly magnified. The slightest vibration caused by even a light wind across a camera on a tripod is likely to introduce detrimental softening on a 24mp APS-C or 61mp FF sensor. I've experienced moderate winds that, even on my very stable GT3532LS tripod, introduce some softening, and the 7D is becoming something only moderately dense as sensor technology continues to evolve and push the envelope towards smaller and smaller pixels.

CarlTN

  • Canon EF 300mm f/2.8L IS II
  • *******
  • Posts: 2227
    • View Profile
Re: A Big Megapixel Discussion
« Reply #143 on: June 20, 2013, 11:36:56 PM »
It will be interesting to see if Canon can actually achieve parity with the current SOA sensor, regarding those attributes that people like to argue about.  What would be mind blowing, is if Canon actually exceeds those attributes...regardless of the pixel dimensions.  It seems to me that if the number is closer to 60MP rather than 45, the performance might be more compromised.  Certainly it will be difficult to make use of all that resolution outside the center 50% of the image on most, if not all Canon lenses...even the 24-70 ii.

Remember that total system resolution is effectively (closely approximated by) the root mean square of the resolution of each component that makes up the system. In a DSLR, to keep things simple, the final resolution of the photographs you make is the RMS of the resolutions of the lens and the sensor. There is no such thing as one outresolving the other. Increasing the resolution of either lens or sensor increases the resolution of the system as a whole, and produces higher resolution photographs.

You get the most bang for the buck by increasing the lowest common denominator, but if you have a lens, like the 24-70 II, and you use it on a 60mp FF camera...you WILL realize better results (all other things being equal...i.e. assuming the best tech available is used to produce said 60mp sensor.)

That was not my point.  My point was as stated.  I never said the overall results would not be "better".  To belabor my point, since you are intentionally missing it...I will quote myself: "...it will be difficult to make use of all that resolution outside the center 50% of the image on most, if not all Canon lenses...even the 24-70 ii." 

I stand by this.  Your point does not disprove my point.  You might have your own idea about how you define the phrase "make use of all that resolution".  I have mine.  My point was never that you could not get improved resolution and image quality, from a higher megapixel sensor.  Only a fool would argue that.  Yet you seem to want to believe that's what I meant.  I wonder why?  Up to your same old tricks I see.

As for "bang for buck", that is an entirely separate issue altogether, which I hope you realize...and has nothing to do what my point.

I am not intentionally trying to misquote you... Simple matter of the facts.

I does not matter whether you are talking about resolution at the center of the frame, edge of the frame, or corner of the frame. The basis of "system resolution", which is a convolution of the effects of each and every component, holds true regardless of which region of the frame you apply it to. Sure, poorer quality lenses and wider angle lenses tend to have more detractors to resolution in the corners. That does not mean that suddenly the rules that govern overall system resolution change. Just as much as a higher resolution sensor will improve the outcome of what the lens resolves at the center, so too will it improve the outcome of what the lens resolves at the edge. A higher resolution sensor can never produce WORSE results than a lower resolution sensor, all else being equal.

A horrible lens is a horrible lens, and while you might see marginal improvements in corner resolution with a higher resolution sensor, you experience diminishing returns. An excellent lens, such as the 24-70 II, which performs quite well in the corners, will realize a greater benefit from a move to a higher resolution lens than, say, the 16-35 II (which performs only moderately well in the corners), both of which would benefit considerably better than say the EF-S 18-55mm, which performs terribly in the corners.

The benefit boils down to a matter of degree for every component involved, not whether or not you get any benefit at all outside of the center of the lens. You don't "make use" of resolution...you convolve a result via a functional process as a real-world image passes through each and every lens element, the aperture, the sensor's filter stack, the CFA, and even the pixel well itself. A higher resolution sensor, assuming equivalence in terms of noise, could never "compromise" IQ in any way. Even with a relative increase in noise, a higher resolution sensor, when its image size is normalized to that of any lower resolution sensor, would still produce results that are as good as or better. (The only time I believe a higher resolution sensor can be detrimental to IQ is when there is a disproportionate increase in the amount of noise due to smaller pixels, which to some degree is the case with the 7D (a fact I blame on Canon's 500nm process, which wastes a lot of photodiode space resulting in disproportionately smaller light sensitive photodiode area relative to sensors with larger pixels...a defect I believe a move to a smaller process, such as 180nm, can resolve.))

A far greater concern, in my opinion, for high density sensors than "making use of" any given lenses resolving power would be avoiding softening from camera shake. As pixel sizes shrink for both APS-C and FF sensors, the effects of camera shake will become increasingly magnified. The slightest vibration caused by even a light wind across a camera on a tripod is likely to introduce detrimental softening on a 24mp APS-C or 61mp FF sensor. I've experienced moderate winds that, even on my very stable GT3532LS tripod, introduce some softening, and the 7D is becoming something only moderately dense as sensor technology continues to evolve and push the envelope towards smaller and smaller pixels.

Where are you quoting "system resolution"?  I made no mention of it.  I agree on the point of camera shake making its presence more known as sensor resolution increases.  Vibration from the mirror, and to a lesser degree the shutter, are also factors.  But I'm sure you will find a way to disagree with me on that, or otherwise tell me I am thinking of it in the wrong way.
« Last Edit: June 20, 2013, 11:39:53 PM by CarlTN »

jrista

  • Canon EF 400mm f/2.8L IS II
  • *******
  • Posts: 4484
  • EOL
    • View Profile
    • Nature Photography
Re: A Big Megapixel Discussion
« Reply #144 on: June 22, 2013, 01:39:01 AM »
It will be interesting to see if Canon can actually achieve parity with the current SOA sensor, regarding those attributes that people like to argue about.  What would be mind blowing, is if Canon actually exceeds those attributes...regardless of the pixel dimensions.  It seems to me that if the number is closer to 60MP rather than 45, the performance might be more compromised.  Certainly it will be difficult to make use of all that resolution outside the center 50% of the image on most, if not all Canon lenses...even the 24-70 ii.

Remember that total system resolution is effectively (closely approximated by) the root mean square of the resolution of each component that makes up the system. In a DSLR, to keep things simple, the final resolution of the photographs you make is the RMS of the resolutions of the lens and the sensor. There is no such thing as one outresolving the other. Increasing the resolution of either lens or sensor increases the resolution of the system as a whole, and produces higher resolution photographs.

You get the most bang for the buck by increasing the lowest common denominator, but if you have a lens, like the 24-70 II, and you use it on a 60mp FF camera...you WILL realize better results (all other things being equal...i.e. assuming the best tech available is used to produce said 60mp sensor.)

That was not my point.  My point was as stated.  I never said the overall results would not be "better".  To belabor my point, since you are intentionally missing it...I will quote myself: "...it will be difficult to make use of all that resolution outside the center 50% of the image on most, if not all Canon lenses...even the 24-70 ii." 

I stand by this.  Your point does not disprove my point.  You might have your own idea about how you define the phrase "make use of all that resolution".  I have mine.  My point was never that you could not get improved resolution and image quality, from a higher megapixel sensor.  Only a fool would argue that.  Yet you seem to want to believe that's what I meant.  I wonder why?  Up to your same old tricks I see.

As for "bang for buck", that is an entirely separate issue altogether, which I hope you realize...and has nothing to do what my point.

I am not intentionally trying to misquote you... Simple matter of the facts.

I does not matter whether you are talking about resolution at the center of the frame, edge of the frame, or corner of the frame. The basis of "system resolution", which is a convolution of the effects of each and every component, holds true regardless of which region of the frame you apply it to. Sure, poorer quality lenses and wider angle lenses tend to have more detractors to resolution in the corners. That does not mean that suddenly the rules that govern overall system resolution change. Just as much as a higher resolution sensor will improve the outcome of what the lens resolves at the center, so too will it improve the outcome of what the lens resolves at the edge. A higher resolution sensor can never produce WORSE results than a lower resolution sensor, all else being equal.

A horrible lens is a horrible lens, and while you might see marginal improvements in corner resolution with a higher resolution sensor, you experience diminishing returns. An excellent lens, such as the 24-70 II, which performs quite well in the corners, will realize a greater benefit from a move to a higher resolution lens than, say, the 16-35 II (which performs only moderately well in the corners), both of which would benefit considerably better than say the EF-S 18-55mm, which performs terribly in the corners.

The benefit boils down to a matter of degree for every component involved, not whether or not you get any benefit at all outside of the center of the lens. You don't "make use" of resolution...you convolve a result via a functional process as a real-world image passes through each and every lens element, the aperture, the sensor's filter stack, the CFA, and even the pixel well itself. A higher resolution sensor, assuming equivalence in terms of noise, could never "compromise" IQ in any way. Even with a relative increase in noise, a higher resolution sensor, when its image size is normalized to that of any lower resolution sensor, would still produce results that are as good as or better. (The only time I believe a higher resolution sensor can be detrimental to IQ is when there is a disproportionate increase in the amount of noise due to smaller pixels, which to some degree is the case with the 7D (a fact I blame on Canon's 500nm process, which wastes a lot of photodiode space resulting in disproportionately smaller light sensitive photodiode area relative to sensors with larger pixels...a defect I believe a move to a smaller process, such as 180nm, can resolve.))

A far greater concern, in my opinion, for high density sensors than "making use of" any given lenses resolving power would be avoiding softening from camera shake. As pixel sizes shrink for both APS-C and FF sensors, the effects of camera shake will become increasingly magnified. The slightest vibration caused by even a light wind across a camera on a tripod is likely to introduce detrimental softening on a 24mp APS-C or 61mp FF sensor. I've experienced moderate winds that, even on my very stable GT3532LS tripod, introduce some softening, and the 7D is becoming something only moderately dense as sensor technology continues to evolve and push the envelope towards smaller and smaller pixels.

Where are you quoting "system resolution"?  I made no mention of it.  I agree on the point of camera shake making its presence more known as sensor resolution increases.  Vibration from the mirror, and to a lesser degree the shutter, are also factors.  But I'm sure you will find a way to disagree with me on that, or otherwise tell me I am thinking of it in the wrong way.

System resolution describes the resolution of the final output of any optical system, which is what a camera with a lens attached is. It's a mathematical concept, thus easily provable. I don't disagree about mirror slap and shutter vibration...definitely two things that can contribute to camera shake, and affect IQ as pixel size diminishes.

Pi

  • 1D Mark IV
  • ******
  • Posts: 937
    • View Profile
    • Math and Photography
Re: A Big Megapixel Discussion
« Reply #145 on: June 22, 2013, 08:31:00 AM »
Where are you quoting "system resolution"?  I made no mention of it.  I agree on the point of camera shake making its presence more known as sensor resolution increases. 
Actually, with smaller pixels. the effect of camera shale decreases slightly. Blur due to camera shake is no different than blur due to a soft lens, for example. Relatively, as a percentage of the total blur, it increases. In absolute terms, camera shake blur remains the same. Go figure.  :)

Don Haines

  • Canon EF 300mm f/2.8L IS II
  • *******
  • Posts: 3372
  • Posting cat pictures on the internet since 1986
    • View Profile
Re: A Big Megapixel Discussion
« Reply #146 on: June 22, 2013, 11:29:30 AM »
(The only time I believe a higher resolution sensor can be detrimental to IQ is when there is a disproportionate increase in the amount of noise due to smaller pixels, which to some degree is the case with the 7D (a fact I blame on Canon's 500nm process, which wastes a lot of photodiode space resulting in disproportionately smaller light sensitive photodiode area relative to sensors with larger pixels...a defect I believe a move to a smaller process, such as 180nm, can resolve.))

Definitely true, but personally, I doubt that Canon will move from 500nm to 180nm technology on their sensors. It is a big thing to change fabrication processes..... everything has to be re-designed and tested, and then you need to set up the production facilities and debug that... It isn't worth it to go down one step.

90nm designs for IC's came out in 2004 or 2005.... That's certainly long ago enough to work the bugs out. I would be willing to bet that Canon will skip over 180nm to 90 or even 60nm. It seems far more reasonable to take one big step than 2 or 3 small steps, particularly since the cost in dollars and time would be about the same.
The best camera is the one in your hands

jrista

  • Canon EF 400mm f/2.8L IS II
  • *******
  • Posts: 4484
  • EOL
    • View Profile
    • Nature Photography
Re: A Big Megapixel Discussion
« Reply #147 on: June 22, 2013, 12:01:57 PM »
(The only time I believe a higher resolution sensor can be detrimental to IQ is when there is a disproportionate increase in the amount of noise due to smaller pixels, which to some degree is the case with the 7D (a fact I blame on Canon's 500nm process, which wastes a lot of photodiode space resulting in disproportionately smaller light sensitive photodiode area relative to sensors with larger pixels...a defect I believe a move to a smaller process, such as 180nm, can resolve.))

Definitely true, but personally, I doubt that Canon will move from 500nm to 180nm technology on their sensors. It is a big thing to change fabrication processes..... everything has to be re-designed and tested, and then you need to set up the production facilities and debug that... It isn't worth it to go down one step.

90nm designs for IC's came out in 2004 or 2005.... That's certainly long ago enough to work the bugs out. I would be willing to bet that Canon will skip over 180nm to 90 or even 60nm. It seems far more reasonable to take one big step than 2 or 3 small steps, particularly since the cost in dollars and time would be about the same.

I am not sure. Process shrinks in the sensor world aren't quite the same as they are in the IC world. If you dig around into more recent CIS patents, you'll find that a LOT of them are based on a 180nm process. I don't know if the body of 90nm CIS patents and fabrication knowledge is as large or as viable. It may be that the time is right to make the next jump...and if Canon does, I certainly am not complaining. I haven't seen any patents from Canon that would indicate such a thing, however. (For that matter, neither have I seen any patents that seem to indicate a jump to 180nm, though.)

Where are you quoting "system resolution"?  I made no mention of it.  I agree on the point of camera shake making its presence more known as sensor resolution increases. 
Actually, with smaller pixels. the effect of camera shale decreases slightly. Blur due to camera shake is no different than blur due to a soft lens, for example. Relatively, as a percentage of the total blur, it increases. In absolute terms, camera shake blur remains the same. Go figure.  :)

True, assuming you intent to publish images from a higher resolution sensor at the same size as a lower resolution sensor. If you use a higher resolution sensor in order to publish larger (i.e. larger prints), then the per-pixel blur, rather than per-frame blur, is what matters. The same would hold true if you get a higher density sensor to support cropping.

canon rumors FORUM

Re: A Big Megapixel Discussion
« Reply #147 on: June 22, 2013, 12:01:57 PM »

CarlTN

  • Canon EF 300mm f/2.8L IS II
  • *******
  • Posts: 2227
    • View Profile
Re: A Big Megapixel Discussion
« Reply #148 on: June 25, 2013, 05:58:31 PM »
(The only time I believe a higher resolution sensor can be detrimental to IQ is when there is a disproportionate increase in the amount of noise due to smaller pixels, which to some degree is the case with the 7D (a fact I blame on Canon's 500nm process, which wastes a lot of photodiode space resulting in disproportionately smaller light sensitive photodiode area relative to sensors with larger pixels...a defect I believe a move to a smaller process, such as 180nm, can resolve.))

Definitely true, but personally, I doubt that Canon will move from 500nm to 180nm technology on their sensors. It is a big thing to change fabrication processes..... everything has to be re-designed and tested, and then you need to set up the production facilities and debug that... It isn't worth it to go down one step.

90nm designs for IC's came out in 2004 or 2005.... That's certainly long ago enough to work the bugs out. I would be willing to bet that Canon will skip over 180nm to 90 or even 60nm. It seems far more reasonable to take one big step than 2 or 3 small steps, particularly since the cost in dollars and time would be about the same.

I am not sure. Process shrinks in the sensor world aren't quite the same as they are in the IC world. If you dig around into more recent CIS patents, you'll find that a LOT of them are based on a 180nm process. I don't know if the body of 90nm CIS patents and fabrication knowledge is as large or as viable. It may be that the time is right to make the next jump...and if Canon does, I certainly am not complaining. I haven't seen any patents from Canon that would indicate such a thing, however. (For that matter, neither have I seen any patents that seem to indicate a jump to 180nm, though.)

Where are you quoting "system resolution"?  I made no mention of it.  I agree on the point of camera shake making its presence more known as sensor resolution increases. 
Actually, with smaller pixels. the effect of camera shale decreases slightly. Blur due to camera shake is no different than blur due to a soft lens, for example. Relatively, as a percentage of the total blur, it increases. In absolute terms, camera shake blur remains the same. Go figure.  :)

True, assuming you intent to publish images from a higher resolution sensor at the same size as a lower resolution sensor. If you use a higher resolution sensor in order to publish larger (i.e. larger prints), then the per-pixel blur, rather than per-frame blur, is what matters. The same would hold true if you get a higher density sensor to support cropping.

I was speaking of "per pixel" blur...actually.

CarlTN

  • Canon EF 300mm f/2.8L IS II
  • *******
  • Posts: 2227
    • View Profile
Re: A Big Megapixel Discussion
« Reply #149 on: June 25, 2013, 06:01:24 PM »
It will be interesting to see if Canon can actually achieve parity with the current SOA sensor, regarding those attributes that people like to argue about.  What would be mind blowing, is if Canon actually exceeds those attributes...regardless of the pixel dimensions.  It seems to me that if the number is closer to 60MP rather than 45, the performance might be more compromised.  Certainly it will be difficult to make use of all that resolution outside the center 50% of the image on most, if not all Canon lenses...even the 24-70 ii.

Remember that total system resolution is effectively (closely approximated by) the root mean square of the resolution of each component that makes up the system. In a DSLR, to keep things simple, the final resolution of the photographs you make is the RMS of the resolutions of the lens and the sensor. There is no such thing as one outresolving the other. Increasing the resolution of either lens or sensor increases the resolution of the system as a whole, and produces higher resolution photographs.

You get the most bang for the buck by increasing the lowest common denominator, but if you have a lens, like the 24-70 II, and you use it on a 60mp FF camera...you WILL realize better results (all other things being equal...i.e. assuming the best tech available is used to produce said 60mp sensor.)

That was not my point.  My point was as stated.  I never said the overall results would not be "better".  To belabor my point, since you are intentionally missing it...I will quote myself: "...it will be difficult to make use of all that resolution outside the center 50% of the image on most, if not all Canon lenses...even the 24-70 ii." 

I stand by this.  Your point does not disprove my point.  You might have your own idea about how you define the phrase "make use of all that resolution".  I have mine.  My point was never that you could not get improved resolution and image quality, from a higher megapixel sensor.  Only a fool would argue that.  Yet you seem to want to believe that's what I meant.  I wonder why?  Up to your same old tricks I see.

As for "bang for buck", that is an entirely separate issue altogether, which I hope you realize...and has nothing to do what my point.

I am not intentionally trying to misquote you... Simple matter of the facts.

I does not matter whether you are talking about resolution at the center of the frame, edge of the frame, or corner of the frame. The basis of "system resolution", which is a convolution of the effects of each and every component, holds true regardless of which region of the frame you apply it to. Sure, poorer quality lenses and wider angle lenses tend to have more detractors to resolution in the corners. That does not mean that suddenly the rules that govern overall system resolution change. Just as much as a higher resolution sensor will improve the outcome of what the lens resolves at the center, so too will it improve the outcome of what the lens resolves at the edge. A higher resolution sensor can never produce WORSE results than a lower resolution sensor, all else being equal.

A horrible lens is a horrible lens, and while you might see marginal improvements in corner resolution with a higher resolution sensor, you experience diminishing returns. An excellent lens, such as the 24-70 II, which performs quite well in the corners, will realize a greater benefit from a move to a higher resolution lens than, say, the 16-35 II (which performs only moderately well in the corners), both of which would benefit considerably better than say the EF-S 18-55mm, which performs terribly in the corners.

The benefit boils down to a matter of degree for every component involved, not whether or not you get any benefit at all outside of the center of the lens. You don't "make use" of resolution...you convolve a result via a functional process as a real-world image passes through each and every lens element, the aperture, the sensor's filter stack, the CFA, and even the pixel well itself. A higher resolution sensor, assuming equivalence in terms of noise, could never "compromise" IQ in any way. Even with a relative increase in noise, a higher resolution sensor, when its image size is normalized to that of any lower resolution sensor, would still produce results that are as good as or better. (The only time I believe a higher resolution sensor can be detrimental to IQ is when there is a disproportionate increase in the amount of noise due to smaller pixels, which to some degree is the case with the 7D (a fact I blame on Canon's 500nm process, which wastes a lot of photodiode space resulting in disproportionately smaller light sensitive photodiode area relative to sensors with larger pixels...a defect I believe a move to a smaller process, such as 180nm, can resolve.))

A far greater concern, in my opinion, for high density sensors than "making use of" any given lenses resolving power would be avoiding softening from camera shake. As pixel sizes shrink for both APS-C and FF sensors, the effects of camera shake will become increasingly magnified. The slightest vibration caused by even a light wind across a camera on a tripod is likely to introduce detrimental softening on a 24mp APS-C or 61mp FF sensor. I've experienced moderate winds that, even on my very stable GT3532LS tripod, introduce some softening, and the 7D is becoming something only moderately dense as sensor technology continues to evolve and push the envelope towards smaller and smaller pixels.

Where are you quoting "system resolution"?  I made no mention of it.  I agree on the point of camera shake making its presence more known as sensor resolution increases.  Vibration from the mirror, and to a lesser degree the shutter, are also factors.  But I'm sure you will find a way to disagree with me on that, or otherwise tell me I am thinking of it in the wrong way.

System resolution describes the resolution of the final output of any optical system, which is what a camera with a lens attached is. It's a mathematical concept, thus easily provable. I don't disagree about mirror slap and shutter vibration...definitely two things that can contribute to camera shake, and affect IQ as pixel size diminishes.

I was speaking of the ability to make use of all the resolution allowed by the pixels of the sensor, and where they would make less of an improvement towards the periphery.  I stand by this.  Again, I never said the overall resolution would not increase.

canon rumors FORUM

Re: A Big Megapixel Discussion
« Reply #149 on: June 25, 2013, 06:01:24 PM »