Has Canon fallen hopelessly behind? (cross-posted)

[hjulenissen]

One issue may be, if diffraction at F16 removes the need for an AA filter, is resolution still slightly compromised at F16 if the sensor also has an AA filter? Are the two blurring effects of diffraction and AA filter additive?

Yes. Diffraction ("h1") will enlarge the PSF and so will the AA filter ("h2"). Both together will make for an even larger PSF (convolution of h1 and h2). From the perspective of aliasing vs sharpness tradeoff, there will probably exist an "optimal" PSF, but you are unlikely to find it in a real camera, and even then it will probably only exist for one (or a couple of) aperture(s)/settings.

If you are usually using a small aperture for maximal DOF, where loss of sharpness is the limitation, aliasing will not be an issue, and the presence of an AA filter will only add to the loss of sharpness (however, its effect will probably be minute compared to that of diffraction).

If you are usually using a large aperture, where lense flaws reduce the sharpness, you should experience the same thing.

-h

[Ray]

.....the presence of an AA filter will only add to the loss of sharpness (however, its effect will probably be minute compared to that of diffraction).

What I've found from various testing in the past, using the same lens and same camera with AA filter, is that the effects of increasing diffraction, after appropriate sharpening, is already minute, comparing one f/stop with the next full stop down.

For example, resolution at F16 appears hardly worse than resolution at F11. Likewise, resolution at F11 appears hardly worse than resolution at F8. One has to do serious pixel-peeping to see it. However, I've found that the resolution difference between F16 and F22 is more clearly noticeable.

I was hoping that a D800E used at F16 might produce at least as good resolution as a D800 at F11, and with no more aliasing artifacts than the D800 at F11. If what you say is true, then it sounds like the removal of that additional blurring due to the AA filter will be less significant than the increased blurring due to diffraction when stopping down from F11 to F16.

[Christoph C. Feldhaim]

[....] For example, resolution at F16 appears hardly worse than resolution at F11. Likewise, resolution at F11 appears hardly worse than resolution at F8. One has to do serious pixel-peeping to see it. However, I've found that the resolution difference between F16 and F22 is more clearly noticeable. [....]

Since the diameter of the airy disc is direct proportional to the F-Number your observation makes sense.
For green light it is about  F-Number*1.35=Airy Disc Diameter in µ.

[Ray]

Since the diameter of the airy disc is direct proportional to the F-Number your observation makes sense.
For green light it is about  F-Number*1.35=Airy Disc Diameter in µ.

As I understand, from my layman's perspective, it can be misleading to equate Airy disc size with pixel size in predicting what cannot be resolved.

I get the impression that at most of F/stops we use with the 35mm format, resolution is affected by a mixture of various lens aberrations, as well as diffraction to varying degrees which may or may not be significant in the mix.

At F5.6 and wider, diffraction is still present but may be irrelevant in relation to more serious limitiations on lens sharpness.

At apertures between F8 and F16, diffraction plays a more dominant role but is still not necessarily the limiting factor regarding resolution in all lenses. I have a suspicion that an MF or LF lens which has been optimised for best performance at F16, might still produce better resolution on the same sensor than an average 35mm zoom lens used at F16.

There is an adage that all lenses are equal at F8. However, with the benefits of the pixel-peeping potential of the modern computer and digital camera, we can see that this is not really true. Perhaps the adage needs to be revised along the lines, "All lenses are equal at F22."

From memory, my own tests have indicated that there's a sudden and fairly obvious drop in resolution moving from F16 to F22, which seems greater than the drop between F8 and F16. I imagine there would be an equally large drop moving from F22 to F32, but I never use F32 so haven't addressed the issue.

I get the impression that at F22 and smaller apertures, lens resolution is fully diffraction limited. At wider apertures, F16 to F8, resolution is only partially limited by diffraction, and at apertures wider than F8, resolution is generally limited entirely by lens aberrations. Diffraction is present, but entirely overshadowed by more serious lens aberrations.

Correct me if I'm wrong.

[Christoph C. Feldhaim]

I just wanted to make clear, that since the F-Numbers grow with the square of the EVs the size of the Airy disc also grows with that square.
So its pretty obvious, that increasing the F-number from 16 to 22 has a greater effect on diffraction blur than increasing the F-number from 4.0 to 5.6.

Like this:

relative EV:        7    6    5     4     3     2     1
F-Stop:             4.0  5.6  8    11    16    22    32
Airydisc (green):   3.4  4.7 6.8   9.3  13.6  18.6  27.2    

[Bernard ODonovan]

I've used Canon cameras since 1980. I've always thought it was a technologically advanced and bold company that benefited from a multifaceted corporate environment (medical, opthomological, optical, etc.). And now, it swoops down to present us with (gasp!) a 22MP instead of 21MP 5D3, and (another gasp!) 18MP movie, I mean still camera -- the 1Dx. And the 1Dx is suppose to replace both the 1d and 1Ds series?? I own both the 1D4 and 1Ds3. I actually, except at higher ASA values (yes, I know ISO is the "correct term"), feel my old 1Ds2 had better image quality than my 1D4. And my 1Ds3 does a great job, except my commercial clients are starting to itch for more res in their instore posters. Because Canon is so far behind Nikon (and I'm also assuming Sony at this point), I've thought of buying a Pentax 645D.

Am I the only one who feels Canon has taken it's eye off the ball by thinking all photographers want to really be cinematographers (I was one one in my early days, and even won a Kodak film award, so I know where I'm coming from)? I'm curious about other thoughts...

Nemo

This self confessed Nikon Fanboy who wept with joy when reviewing the new D800 (here - http://www.youtube.com/watch?v=MT6ilCd7CS4 )

Suggested he is considering purchasing the Canon 5D3 (here - http://www.youtube.com/watch?v=8jKspoynL0o )

It depends on what you need the camera for. Each offers something.

If Nikon bring out a 4Dx that will move the goal posts again. Equally Canon will be bringing products out to match, beat or create new markets... Low cost Full Frame will be a big market if done right... Some rumors Canon may bring out an entry level full frame this year...

And for a bit of fun see this:

http://www.youtube.com/watch?v=dm31O9no34A

[dreed]

There was a time, long ago that long gray lenses were the norm at sporting events.....no so much any longer.

I disagree. At the last internationally significant sporting event I went to (Australian Formula 1 Grand Prix), out of the 7 photographers that I saw wearing official bibs, 5 were wielding Canon and 2 were wielding Nikon.

[Ray]

I just wanted to make clear, that since the F-Numbers grow with the square of the EVs the size of the Airy disc also grows with that square.
So its pretty obvious, that increasing the F-number from 16 to 22 has a greater effect on diffraction blur than increasing the F-number from 4.0 to 5.6.

Like this:

relative EV:        7    6    5     4     3     2     1
F-Stop:             4.0  5.6  8    11    16    22    32
Airydisc (green):   3.4  4.7 6.8   9.3  13.6  18.6  27.2    

Surely, if a lens were diffraction limited at F4, such as perhaps the sort of lens that is used on the Nokia Pure View Smart Phone with a 41mp built-in camera, then the increase in F number from F4 to F5.6 would have the same degree of blurring effect as an increase from F16 to F22, ie. 4.7/3.4 = 18.6/13.6.

[shadowblade]

I've used Canon cameras since 1980. I've always thought it was a technologically advanced and bold company that benefited from a multifaceted corporate environment (medical, opthomological, optical, etc.). And now, it swoops down to present us with (gasp!) a 22MP instead of 21MP 5D3, and (another gasp!) 18MP movie, I mean still camera -- the 1Dx. And the 1Dx is suppose to replace both the 1d and 1Ds series?? I own both the 1D4 and 1Ds3. I actually, except at higher ASA values (yes, I know ISO is the "correct term"), feel my old 1Ds2 had better image quality than my 1D4. And my 1Ds3 does a great job, except my commercial clients are starting to itch for more res in their instore posters. Because Canon is so far behind Nikon (and I'm also assuming Sony at this point), I've thought of buying a Pentax 645D.

Am I the only one who feels Canon has taken it's eye off the ball by thinking all photographers want to really be cinematographers (I was one one in my early days, and even won a Kodak film award, so I know where I'm coming from)? I'm curious about other thoughts...

Nemo

I think the Sony Exmor sensor has a lot to do with this. Prior to the Exmor, Canon had a distinct advantage in image quality at any given sensor size. The D700 was a much better camera than the 5D2 in almost every respect, but the 5D2's 21MP sensor was enough of a selling point that many of us - including many people moving from MF film to digital - bought it instead, despite its poorer AF, weather sealing, etc.

Canon has had no answer to the Exmor - ever since its introduction, Nikon has had Canon beaten at every (non-video) turn, in dynamic range, low-ISO noise, high-ISO performance, and now resolution. With nothing to match Nikon/Sony in still photography, Canon seems to have decided to pursue video instead, hoping that superior video performance in consumer-level bodies (often bought to replace a point-and-shoot and used to take both stills and video) will help them cement their lead in that field, even if their pro-level bodies (usually bought either for video or stills, but not both) lag behind the competition.

At this point, Canon is yet to release a true replacement for the 5D2 and 1Ds3, which are now 4 and 5 years old respectively. People bought the 5D2 for its image quality, not its other features - for its time, 21MP was a huge number for a 35mm-format camera, and allowed it to compete against the contemporary D700, despite the latter's better features in every other respect. The 5D3 is not a true replacement - 22MP is only a moderate resolution for a 2012-era full-frame sensor, and many people who bought the 5D2 for its image quality don't need the 5D3's faster frame rate (if they'd needed fast frame rate, they'd have bought the D700 or 1D3 instead). Improved autofocus is always welcome in any camera, though. The 1Dx is not a true replacement for the 1Ds3 - people bought the 1Ds3 for its resolution. Had they not needed it, they'd have bought the much-cheaper, much-faster 1D3 or D700 instead.

Essentially, what Canon has done this round is released one fast action camera (1Dx) and one slightly-slower action camera (5D3) with nothing for those requiring high resolution to make large, detailed prints - completely different from what they offered last round. Nikon, on the other hand, has released one action camera (D4) as well as one high-resolution camera (D800). This is an obvious hole in the Canon lineup - no landscape or studio photographer buying into full-frame digital in 2012 for the first time would consider Canon.

If Canon cannot compete with the Exmor, what they should do to stay in the game is to buy it and use it in a high-resolution body, while developing their own technology. After all, that's what every other camera company which can't keep up sensor-wise does (Nikon, Pentax, Leica, etc.). The Sony NEX-7 sensor scaled up to full-frame would be 54MP, allowing for 150ppi in a 40x60" print. Pair it up with a 16-bit A/D converter, put it in a 5D3 body and Canon could easily trump the D800. Unfortunately, pride may get in the way of this ever happening...

[Keith Reeder]

Canon has had no answer to the Exmor

Frankly it doesn't need an "answer". For the vast majority of photographers, in the vast majority of shooting situations, the qualitative difference between Canon's current sensors and the Exmors is too close to worry about - and then, only really at base ISO.

[shadowblade]

Frankly it doesn't need an "answer". For the vast majority of photographers, in the vast majority of shooting situations, the qualitative difference between Canon's current sensors and the Exmors is too close to worry about - and then, only really at base ISO.

Well, base ISO is where the majority of landscape photography takes place. And dynamic range and resolution are two other critical areas for landscape photographers - areas which the latest Canon offerings are distinctly lacking in.

[Hulyss]

You should all stop fighting and buying an SD1. Not that much MP, very high resolving power. Or at least waiting for the little DP merills

[Keith Reeder]

Well, base ISO is where the majority of landscape photography takes place. And dynamic range and resolution are two other critical areas for landscape photographers - areas which the latest Canon offerings are distinctly lacking in.

Yep, I hear that - but (with respect) landscape 'togs are a small subset of Canon's userbase, and landscape 'togs who are skilled,  serious and demanding enough to really benefit from the base ISO DR difference are a small subset of that small subset.

Don't get me wrong, it would be "better" in an absolute sense if Canon was able to match - or beat - the Exmor sensor, but I stand by my assertion that in reality, for almost all of us it doesn't really matter.

[MatthewCromer]

Canon has had no answer to the Exmor - ever since its introduction, Nikon has had Canon beaten at every (non-video) turn, in dynamic range, low-ISO noise, high-ISO performance, and now resolution. With nothing to match Nikon/Sony in still photography, Canon seems to have decided to pursue video instead, hoping that superior video performance in consumer-level bodies (often bought to replace a point-and-shoot and used to take both stills and video) will help them cement their lead in that field, even if their pro-level bodies (usually bought either for video or stills, but not both) lag behind the competition.

Canon's DSLRs are worthless for P&S style video though.  Only Sony offers a decent solution for that in their dSLT cameras.

[shadowblade]

Canon's DSLRs are worthless for P&S style video though.  Only Sony offers a decent solution for that in their dSLT cameras.

I guess they're no worse than Nikon in that regard, at least... but these are meant to be primarily still cameras!

[BJL]

Canon's DSLRs are worthless for P&S style video though.

I guess they're no worse than Nikon in that regard, at least... but these are meant to be primarily still cameras!

Indeed: the greatest lack in the 1D C is not having an EVF, so that:

Only Sony offers a decent solution for that in their dSLT cameras.

except that I would add Panasonic to that list, and maybe even Olympus with the apparently good in-body video image stabilisation with all lenses of the OM-D E-M5.  With Canon's recent heavy emphasis on video and video+stills combinations in its SLRs, isn't it time to follow Sony [SLT, NEX] and Panasonic [GH2 and "HD ready" lenses] in offering SLR-like cameras with EVF's? It could be a hot-shoe accessory for a traditional SLR if the OVF is still wanted.

[shadowblade]

Indeed: the greatest lack in the 1D C is not having an EVF, so that:except that I would add Panasonic to that list, and maybe even Olympus with the apparently good in-body video image stabilisation with all lenses of the OM-D E-M5.  With Canon's recent heavy emphasis on video and video+stills combinations in its SLRs, isn't it time to follow Sony [SLT, NEX] and Panasonic [GH2 and "HD ready" lenses] in offering SLR-like cameras with EVF's? It could be a hot-shoe accessory for a traditional SLR if the OVF is still wanted.

Hybrids are, by their very nature, a compromise - not the best set of features for stills, and not the best set of features for video. Therefore, they are usually most attractive to the lower end of the market, in the consumer-oriented lines. Those at the higher end are usually either stills photographers or videographers, not both; the rare people who shoot both normally want optimal quality in each, i.e. not a hybrid.

[Petrus]

Those at the higher end are usually either stills photographers or videographers, not both; the rare people who shoot both normally want optimal quality in each, i.e. not a hybrid.

I might be an example of a press photographer / photojournalist who has voluntarily drifted to video also during the last 10 years of my 30+ year career. At the moment I use Canon 5D2 as my main stills camera and Canon XF305 as the favorite video camera. I also use 5D2 for video, depending on the style needed. I DO NOT shoot both stills and video at the same event. Photo assignments are photo only, video assignments are video only + sound, which is my other expertise also as a hobby. Many photographers have to deliver both video and stills, that is true, I do not envy them at all. We make "slide show" type presentations and videos if needed for the web. If there is enough time, then video is shot also, but rarely (a week in Afganistan type situations).

It seems that many posters (not particularly here, but generally) do not understand that there are several different styles and types of video cameras for a good reason and there are many different styles and needs of shooting video, just as there are many different kinds of stills cameras and different requirement for shooting stills. Putting a zillion pixel sensor in a point and shoot camera or cell phone does not make it a great sports or architectural camera, putting a video capture software into a DSLR does not make it an ENG or documentary video camera. These "hybrids" can be really successful like 5D2 has proved and the new versions from all major players are even better, but really they are not hybrids in the sense that video and stills are utilized at the same time (like in cars electric and combustion engines are used at the same time), but that they can be used as great all-round still cameras, and also as video cameras for SPECIFIC TYPES of video productions. Namely these video DSLRs suddenly were capable of movie-like results at a fraction of the cost or size of real movie cameras*. They also make possible to shoot video while on photo assignment, but the usability is in reality severely restricted compared to "real" video cameras like the XF305 I mentioned.

If I shoot a video (tv commercial for example) which is preplanned and scripted and has no synched audio, I use DSLR, because it is compact and free and quality is great and I can play with DOF. If I shoot an interview with audio I rather use XF305, if ENG style stuff I only use XF305, because it is much more ergonomic, has working AF, long fairly fast 18x zoom, good audio with phantom power and has actually better video quality than 5D2 (4:2:2 versus 4:2:0).  So what I am getting at is that it somewhat annoys me when people discuss and write about the new video capabilities of new cameras and do not really know what video production means in each case and why comparing and dissing different solutions is quite pointless when the uses and needs are totally different. Like saying that why buy a $700 Makita router when you can get a Black&Decker power drill set for $150.

*) Still there are good reasons why real cine lenses cost $12000 and up and same quality similarly specced autofocus auto aperture still camera lenses cost less than 10% of that.

[BJL]

Hybrids are, by their very nature, a compromise - not the best set of features for stills, and not the best set of features for video. Therefore, they are usually most attractive to the lower end of the market, in the consumer-oriented lines.

That makes sense overall, but the Canon 1D X and now the $15,000 1D C push hybrids way beyond the normal "consumer" price range, and are being marketed by Canon as for professional usage. I can see that for some journalists who need to single-handedly bring home both video and stills, as Canon says at http://www.dpreview.com/news/2012/04/14/Canon-4K-who-is-it-for

"Because of the EOS 1D C’s ability to act as a film and stills camera it also has an ideal application in journalism, allowing a single reporter could produce high quality video and capture stills for broadcast. Overall the unique feature set of the EOS 1D C means it will be suitable for diverse shooting scenarios."

What mystifies me more is things like the absence of an EVF from that $15,000 camera! combined with the claim that:

"The EOS-1D C will also be used by the film production industry professionals, possibly as a ‘B’ camera and particularly in situations where a conventional camera, even one as small as C300, won’t fit. We also believe it will be used by independent cinema productions as a standalone camera."


P. S. to Petrus and others who have used cameras like the 5D2 for professional video work:
How do you deal with the lack of an EVF, or of a video-camera sized twistable LCD? Are there accesories that fill the gap, or can you do with just using the fixed 3" rear LCD?

[stevesanacore]

I too don't really understand the 1D C camera unless the image quality is vastly superior to it's competition. As far as shooting video with the 5D, I always use a 7" LCD monitor mounted on or near the camera. I also almost always use a jib arm or dolly when shooting, so there is ample room to mount the monitor. Zacuto also make a few good options which I use. But personally I really don't see any cinematographer wanting to use any DSLR over a RED Scarlet or Epic, Arri Alexa or Sony F3, etc. once you get into the past the $10K price point.