Canon 60D--18 MP

[spotmeter]

I see that the new Canon 60D has an 18 megapixel sensor.

What is the size of those pixels, and how many megapixels would a full-frame sensor have at that pixel size?

Daydreaming about the next full-frame Canon. 

[uaiomex]

Little over 36. Sounds nice but I have to ask how many lenses would render that resolution if any? Maybe at center and at optimal fstop, but I'd hate to be thinking or being limited about that all the time instead of concentrating on more important things.
Eduardo

I see that the new Canon 60D has an 18 megapixel sensor.

What is the size of those pixels, and how many megapixels would a full-frame sensor have at that pixel size?

Daydreaming about the next full-frame Canon. 

[spotmeter]

Little over 36. Sounds nice but I have to ask how many lenses would render that resolution if any? Maybe at center and at optimal fstop, but I'd hate to be thinking or being limited about that all the time instead of concentrating on more important things.
Eduardo

Thanks. Now I'm looking forward to the next full-frame Canon even more than ever.

[DarkPenguin]

Little over 36. Sounds nice but I have to ask how many lenses would render that resolution if any? Maybe at center and at optimal fstop, but I'd hate to be thinking or being limited about that all the time instead of concentrating on more important things.
Eduardo

How would you be limited?  Nothing would behave worse with more pixels.  Might not be any better but certainly no worse.  The lighter touch of an AA filter for a higher density sensor should give one more resolution.

[spotmeter]

How would you be limited?  Nothing would behave worse with more pixels.  Might not be any better but certainly no worse.  The lighter touch of an AA filter for a higher density sensor should give one more resolution.

Thanks, Dark. My thoughts exactly. 

My print quality jumped so much going from the 5d to the 5d2 that I am really looking forward to going to the next level.

If Canon doesn't come out with a 30+ MP camera soon, the new Pentax 645D with a 40 MP sensor looks very attractive. Since the sensor is smaller than 645 film, the sensor will grab the best part of each lens.  The only downside is at wide-angle--but they say they are working on this.

Unlike Canon, the Pentax has a dedicated mirror lock-up!

[douglasf13]

The Canon 7D has 18mp and has been out for a year.

[BJL]

The Canon 7D has 18mp and has been out for a year.

I am glad someone mentioned that! The 60D is an exercise in "trickle down sensor economics". I have not read many complaints from 7D users about soft images due to diffraction, or any of the other horrors attributed to increasing the resolution offered at a given format size.

By the way, the area ratio between Canon's EF-S and 35mm format sensors is about 1.6^2=2.56, so with the 7D/60D pixel density, a possible future Canon 35mm format sensor would be about 2.56 x 18MP = 46MP. Close enough to 40MP that pixel peepers could roughly predict how it would compare to the 40MP MF options by comparing 100% crops from the 7D to ones from those current 40MP MF options.

[spotmeter]

I am glad someone mentioned that! The 60D is an exercise in "trickle down sensor economics". I have not read many complaints from 7D users about soft images due to diffraction, or any of the other horrors attributed to increasing the resolution offered at a given format size.

By the way, the area ratio between Canon's EF-S and 35mm format sensors is about 1.6^2=2.56, so with the 7D/60D pixel density, a possible future Canon 35mm format sensor would be about 2.56 x 18MP = 46MP. Close enough to 40MP that pixel peepers could roughly predict how it would compare to the 40MP MF options by comparing 100% crops from the 7D to ones from those current 40MP MF options.

Thanks, BJL, for your calculations.

I'm looking forward even more to a new Canon full-frame.

Anyone out there with a 7D and a 40MP back who would like to post such crops?

[feppe]

The Canon 7D has 18mp and has been out for a year.

And so does 550D which has been out for six+ months. Although I own one I had to check the MP count as that's not high on my feature-needs.

[uaiomex]

fStop range limited. These small pixels will difract sooner than those in the present 5D2.
I came from 20D to 5D and the incremental in resolution was spectacular. From the 5D1 to 5D2 the increment was very moderate. It is nicer to have 21 mp than having only 12, but the law of diminishing returns is now catching up very quickly for FF. If you really want more rez, leaping to MF is the answer. I will be gladder if Canon addresses other more important issues. Please no more banding and griding!!!!
Eduardo

How would you be limited?  Nothing would behave worse with more pixels.  Might not be any better but certainly no worse.  The lighter touch of an AA filter for a higher density sensor should give one more resolution.

[pcunite]

How would you be limited?  Nothing would behave worse with more pixels.  Might not be any better but certainly no worse.  The lighter touch of an AA filter for a higher density sensor should give one more resolution.

Looking forward to this.

[Doug Peterson]

Close enough to 40MP that pixel peepers could roughly predict how it would compare to the 40MP MF options by comparing 100% crops from the 7D to ones from those current 40MP MF options.

That would give you a pretty decent comparison for the middle of the image where very few lenses are challenged, but anyone concerned with edge-sharpness, especially with wide-angles or difficult-to-design-lenses like tilt-shift lenses would gain no information about the frame-edge of a predicted 1Ds IV.

It is nicer to have 21 mp than having only 12, but the law of diminishing returns is now catching up very quickly for FF. If you really want more rez, leaping to MF is the answer.

And rez is not the only aspect of image quality which MF excels at. I'd like to see canon take a series shot at increasing other aspects of image quality like the type/amount of noise present in shadows, especially when pushed, the quality of color especially in mixed lighting situations, the fidelity of skin tones as exposure reaches blown, the quality of their long exposures, smoothness of tonal gradients, shadow color accuracy etc etc etc.

All of these are pretty good today (especially compared to a few years ago), but are still well behind MF cameras even from five+ years ago. When you shoot both (as I do) you miss Canon's High ISO and video when shooting MF and miss MF's stunning image quality and resolution when shooting Canon.

Doug Peterson (e-mail Me)
__________________
Head of Technical Services, Capture Integration
Phase One, Leaf, Cambo, Canon, Apple, Profoto, Eizo & More
National: 877.217.9870  |  Cell: 740.707.2183
Newsletter: Read Latest or Sign Up
RSS Feed: Subscribe
Buy Capture One at 10% off
Personal Work

[BJL]

... with the 7D/60D pixel density, a possible future Canon 35mm format sensor would be about 2.56 x 18MP = 46MP. Close enough to 40MP that pixel peepers could roughly predict how it would compare to the 40MP MF options by comparing 100% crops from the 7D to ones from those current 40MP MF options.

That would give you a pretty decent comparison for the middle of the image where very few lenses are challenged, but anyone concerned with edge-sharpness, especially with wide-angles or difficult-to-design-lenses like tilt-shift lenses would gain no information about the frame-edge of a predicted 1Ds IV.

Indeed: I was thinking about comparing sensor performance and diffraction effects. Maybe I should have said "photosite peepers", meaning those whose main hobby is comparing "per pixel" sensor performance and image quality.

[ErikKaffehr]

Hi,

With improved sharpening we can reduce the effect of the AA-filter, thus increasing apparent sharpness. We also get a lot of jaggies, star case effects and so on.

Increasing resolution past lens resolution gives over sampling and that has always been a god thing. Little is lost and much gained by making pixels smaller. That said, pixel wise image quality will be better with fewer pixels, and DR will be less with smaller pixels.

Best regards
Erik

[ErikKaffehr]

Hi,

It seems that most lenses get diffraction limited around f/8 independent of pixel size. At medium apertures a higher resolution sensor would still benefit.

Also, keep in mind that linear resolution is the square root of "megapixels" so even doubling the megapixels only increases resolution by 40%. It seems that most DSLR still can benefit from resolution increases.

MFDBs will always have an edge:

- Larger sensor collect more photons, which gives less shot noise
- Larger physical pixels will put less demand on lens
- Lenses may actually be better at least in some cases

An observation. There are two variants of DSLR lenses, professional lenses and standard lenses. Professional lenses (like L-series) are often having large apertures and other features which may not be optimal for image quality. It's much easier to build f/2.5 lens than a f/1.4 lens, that's probably one of the reasons that macro lenses are that good. Fewer elements and smaller.

Best regards
Erik

[BJL]

It seems that most lenses get diffraction limited around f/8 independent of pixel size.

Some people talk as if being diffraction limited is a bad thing! Quite the opposite; it means that all the avoidable resolution limits like lens aberrations or the limitations of sensors or film are controlled well enough that resolution is near the maximum allowed by the basic physics of light: the diffraction limit. I dream of sensors and lenses good enough that I am diffraction limited all the time!

Also, I have seen many lenses, including zooms, that look diffraction limited to about f/4 ... at least, their resolution increases one opens up from f/8 to f/5.6 to f/4.

[Ray]

Some people talk as if being diffraction limited is a bad thing! Quite the opposite; it means that all the avoidable resolution limits like lens aberrations or the limitations of sensors or film are controlled well enough that resolution is near the maximum allowed by the basic physics of light: the diffraction limit. I dream of sensors and lenses good enough that I am diffraction limited all the time!

Also, I have seen many lenses, including zooms, that look diffraction limited to about f/4 ... at least, their resolution increases one opens up from f/8 to f/5.6 to f/4.

Well, the last part I agrre with, BJL. A few lenses can be slightly sharper at F4 than they are at F5.6 and F8, but sometimes when this occurs, you find that although the sharpness at F4 is slightly better than the sharpness of the same lens at F8, the sharpness at F8 is not as good as it could be, compared with other lenses which might have been optimised for maximum sharpness at F8, so one is left with the impression that diffraction plays a negligle role at F4.

A lens that is diffraction limited at F4 should deliver twice the resolution, at the same MTF, of a lens which is diffraction limited at F8. Is this not correct?

[ErikKaffehr]

Hi,

My point is mainly that resolution starts to drop significantly at small apertures. But it seems that we are not really in diminishing return with todays pixel sizes. On the other hand I had seen some indications that with many of the newest cameras with the highest pixel density the lens shouldn't be stopped down beyond f/5.6 for maximum resolution.

I essentially agree with both BJL and Ray, but may take a more relaxed view:

- I don't just look at central resolution but resolution across the field. Corners often still improve when stopping down although center has passed maximum.
- In my view you cannot go wrong with f/8. Yes some lenses may be better at f/5.6 or even f/4 at the center, but they still near optimum at f/8
- Ray is correct that about twice the resolution at the same MTF at f/4 compared to f/8, at least in my view
- The classic definition of "diffraction limited" was that the first Airy ring could be clearly seen on optical bench on a single point object (collimated rays of light)

In general I'd suggest that there are benefits of increasing pixel density. The need of OLP filter is reduced so the filter may be thinner. Aliasing will be reduced.

On the other hand, a bigger format will always win, at least for flat objects. It simply collects more photons. MTF for a given feature will be higher decent lens on both sensors presumed.

That said, we had a lengthy discussion about some of the supposed advantages of MFDBs over DSLRs, and it seems that it's not easy to find a theoretical explanation. The resolution MTF at given feature size argument is always there. If you shoot something on MFDB you would use a longer lens and the subject would be imaged on a larger area of the sensor. If a lens of corresponding quality would be used, MTF would be higher. The larger sensor will collect more photons so shot noise will be lower. It's feasible to assume that MF lenses may have some advantages over DSLR lenses as DSLRs are often used with zooms.

Aliasing effects can produce fake detail which are seen as better sharpness.

My guess is that a DSLR with excellent lenses like Canons L-lenses (except wide angles) and some of the Zeiss lenses (like the 21/2. will perform very well on a 40 MPixel camera and that a 40 MPIxel camera will have better image quality than a 21 MPixel camera, if the image is printed at the same size.

An MFDB would probably produce something like 50% higher MTF for small features, so microcontrast would be perceived significantly better. This would probably hold for "normal lenses" like double Gauss designs but not for some of the more exotic design. A Coastal Optics 60/4 APO macro on a DSLR would certainly beat a "Coke bottle" on any MFDB.

Best regards
Erik

[Ray]

That said, we had a lengthy discussion about some of the supposed advantages of MFDBs over DSLRs, and it seems that it's not easy to find a theoretical explanation. The resolution MTF at given feature size argument is always there. If you shoot something on MFDB you would use a longer lens and the subject would be imaged on a larger area of the sensor. If a lens of corresponding quality would be used, MTF would be higher. The larger sensor will collect more photons so shot noise will be lower. It's feasible to assume that MF lenses may have some advantages over DSLR lenses as DSLRs are often used with zooms.

Erik,
I've long held the view that the advantages of the larger sensor are only significant in certain situations, and for certain types of scenes and lighting conditions, often where strobes and/or tripods are needed. As always, the best tool for the job applies. MFDB is not the best tool for all jobs.

To illustrate the point, let's consider an outdoor scene with rather dim lighting, towards the end of the day perhaps, or perhaps a very cloudy and misty day, with some movement in the scene one wants to freeze, and some interesting features in the foreground requiring a reasonable DoF, and some interesting detail in shadows one wants to preserve.

I'm carrying two cameras, a 60D with 100/2.8 macro, and a P65+ with 645 body and 300mm lens (or a zoom lens with 270mm to get the exact FL equivalent).

I decide that the Canon 60D at F4 and ISO 100 will give me the needed shutter speed of 1/400th (or 1/200th, whatever) and sufficient DoF. I'm also pleased with the fact that the new 100/2.8 IS macro is actually sharpest at F4, so I figure I'm going to get optimum results using the 60D for this scene.

However, I also know that the P65+ will capture more than 4x the amount of light and has about 4x the number of pixels (before binning and after cropping the 60D image to the same aspect ratio). Surely the image from the P65+ should be better in all respects. But maybe not. Would anyone care to show such a comparison?

From a theoretical perspective, I've worked out the following.

(1) The 18mp 60D image cropped to the 4:3 aspect ration becomes 15.75mp

(2) The P65+ image in sensor+ mode (binning 4 pixels to one large pixel) is 15mp

(3) The ratio of the diagonals of the two respective sensors is 2.7:1

(4) To get the DoF that the D60 give me at F4, I need to use F11 with the P65 (F10.8 to be precise - 2.7x4)

(5) To get the same shutter speed at F11 that I get with the D60 at F4, I need to use ISO 800 with the P65+

(6) According to DXOmark, the 7D (presumably similar to the 60D in this respect) at ISO 100 has a DR almost 2 stops greater than the P65+ at ISO 800.

Wow! 2 stops (or 1.87 to be precise) is very significant.

It's doubtful that the P65+ image would be sharper than the D60 image considering it was necessary to use F11 for DoF requirements. However, it's possible it might be noticeably better in subtle ways regarding noise at 18% grey, color sensitivity and tonal range. I'd like to see such a comparison on print.

[Ray]

The pixel size is about 7.4 x 7.4 µm. A full-frame sensor would have 21MP.

A full frame 35mm sensor consisting of 60D pixels would be about 46mp, and a full frame 645 sensor such as the P65+ would be about 115mp.