« The Canon 5D Class of Camera | Main | Thom Hogan's Nikon D300 Guide »

Thursday, 27 March 2008


The new 5D will cost around $3200.

Mike the 12 Mp Rebel is $900 with lens. the 10 Mp 40D is $1,200. Their very good at what they do (I know the rebel hasn't been shipped yet) but sample images look great even at 30X40, so on to full frame which for most photographers is not really needed anymore. Most lenses tend to lose after 12 to 14 Mp's So if we use the 40D's pixel pitch we are at 16MP with the 5D?. Not a bad place to be and any more and they kill the 1DsMKIII. Even at this it has to beat all other aspects of the D300. Lets not forget the 1DMKIII. I think Canon may be caught between a rock and a hard place with to many good cameras on the market. If they come out with a less expensive killer full frame 5D? I think $2.995 would be a good starting point. It also has to beat out or come close to the Nikon D3 features. Tough nut to crack. I think at this point their painted into a corner. Pro's will always pay the price and amateurs will always complain. OK $2,495 but it has to have everything everyone wants.

Dear Carl,

I'm not quite sure what you mean by "Most lenses tend to lose after 12 to 14 Mp's" but if you're suggesting that lenses can't routinely make use of larger numbers of pixels, that doesn't match up at all with my experience with film camera lenses.

Merely decent 35mm format lenses have no trouble rendering 100 line pair per millimeter around their optimum apertures. Very good (by no means the best) lenses can render 200.

Using the very crude metric of two pixels per line pair (wrong in so many ways, but let it slide) that's around 35 megapixels for the merely decent lens and 150 megapixels for the very good lens.

I am not remotely suggesting your typical, even demanding, photographer needs anywhere near that number of pixels to get superb results. We have ample experimental data to the contrary.

But, should one choose to acquire more than 12-14 megapixels, they're going to be able to take advantage of them (unless they're buying crappy lenses to go with them).

~ pax \ Ctein
[ please excuse any word salad. MacSpeech in training! ]
-- Ctein's online Gallery http://ctein.com
-- Digital restorations http://photorepair.com

Yeah, re: Carl Leonardi's comments -

I see NO reason why Canon have to make sure that the 5D successor has to compete with both the Nikon D300 and the D3! Why? Nikon chose to make two camera models that do not directly compete with anything on the market, as did Canon with the 5D nearly three years ago. It's worked out brilliantly for both companies! So why start now???

I'm not saying that people won't be choosing between these models, they will, but that does not mean that the camera models have to compete head-to-head! In fact, it would be better if Canon chose to make the 5D MkII do something useful that nothing else on the market does, even if it was at the expense of some other feature. Lateral thinking and all that... :-)

Ctein, this is true at f/5.6 to f/11 on better lenses but on full frame cameras even the best start to get soft on the edges.
I'm putting my 16 to 35mm L Canon lens on e bay at this very moment because it's as soft as cotton on the edges.
Even the new one isn't that hot until you get to f/8 and it's only $1,600. As I understand it the 10 to 22mm lens out performs it on the 1.6 size sensors at $750 dollars. Now if the new 200 mm was not $6,000 I might think of getting it.
I use this as a guide, if it isn't all purple is not good for FF.

I concur with Stephen's speculation that the price will be around the $3,000 - $3,200 spot.

That's not really the key question. No, the big question is -when- will Canon introduce the camera and how many new tweaks they'll build into it. The consumer end of the economy is definitely headed south, with the worst effects of this episode perhaps 9-12 months away. If current indicators are to be trusted (and there are some really unprecedented scares in there) the market for a $3,000+ camera will be much smaller by 2009 than it is now. So will Canon squeeze the new model out sooner, perhaps with fewer new gizzies to sop-up the last bull market grease drippin's?

If I were advising them (which I am not) I would suggest that they take into account the high customer satisfaction with the current 5D. Specifically, I'd introduce a "5Dn", featuring perhaps 3 out of a potential 8 significant advancements, at around $2,800 - $3,000. I would postpone introduction of a "6D" until late 2009 or 2010 depending on economic conditions.

Meanwhile, I'd consider introducing a new genre of p&s in the $500 bracket in late 2008 or early 2009. This line would be an evolution of the G series designed to both short-circuit the (bizarre) Internet cult forming around those Ricoh cameras as well as serve those who harbor M8 lust. Both of these trends sprout from a strong, sincere desire for a high-quality, small, medium-high performance camera that's different from anything available today. It's well within Canon's ability to address this opportunity very soon if they want to do so.

Dear Carl,

I understand your concerns, but expecting corner-to-corner sharpness from a wide angle zoom isn't realistic over any wide range of circumstances. That may be the kind of lens you mostly need but honestly it's not a good choice if one wants maximum sharpness. Except maybe at optimum aperture *and* optimum focal length. Which kind of defeats the point of having a zoom lens [ smile ].

I had little problem finding a bunch of lenses on that website (cool interactive graphs!) that would give good corner-to-corner sharpness with 20+ megapixel cameras. Not of the sorts you're looking for, but that's a different matter.

Even in the sorts you're looking for, the central area is still going to show substantially better sharpness with more megapixels.

~ pax \ Ctein
[ please excuse any word salad. MacSpeech in training! ]
-- Ctein's online Gallery http://ctein.com
-- Digital restorations http://photorepair.com

My prediction? There won’t be a 5D replacement in the near future.

This is largely based on some stuff Thom Hogan has been writing recently. Take a look at the following price points (from B&H):

Canon 40D: $1,140
Nikon D300: $1,800
Canon 5D: $2,200
Canon 1D Mark III: $4,330

[Side note: $1,140 for the 40D? I almost fell off my chair. For my fellow Europeans, that is only Euro 723. Unbelievable.]

It would seem that producing a full-frame sensor is much more expensive than producing a crop-sized sensor. So the question becomes: does Canon make more money by creating a model to fit between the 5D and the 1D Mark III, or would it make more money by creating a no-holds-barred crop sensor camera to compete with the Nikon D300? I would bet on the latter (side note #2: Nikon must be making a KILLING on the D300).

Relative to what it would cost to produce a new full-frame model, I don’t think Canon would get much of a return. I think the 5D is really more of a defensive product, designed to prevent other camera manufacturers from even thinking about entering the product category. Why try to develop a cheap-ish full frame camera when Canon already has one that is selling for a [relatively] paltry $2,200? Under this line of thinking, it isn’t worth Canon’s while to update the 5D, but it can’t afford to take it out of the lineup for competition and prestige reasons, so it just lets the 5D continue to putter along. It’s still a good camera. It’s still the only sub-$4,000 full-frame camera available new. Why go through the expense and hassle of upgrading it? I’m not being snide or flip, this is what I actually believe.

In order to really move this market segment forward, Canon would need to have a breakthrough in sensor production costs, and it would have to have a product that offers something compelling relative to the 40D (or the hypothetical D300 competitor mentioned above), yet that doesn’t make the 1D Mark III look like an overpriced dud.

It seems to me that the only way to do that would be to take something like the 1D Mark III sensor and put it into a body that had slower frame rates and was otherwise less advanced than the 1D Mark III. Perhaps less robust construction and weather sealing and a less advanced autofocus system. Starts to sound a lot like the current 5D, doesn’t it?

BTW, the 5D originally cost $3,300. I wouldn’t expect its successor (whenever it does arrive) to be significantly below $3,000.


"Merely decent 35mm format lenses have no trouble rendering 100 line pair per millimeter around their optimum apertures. Very good (by no means the best) lenses can render 200."

I agree and disagree. Even the best Leica glass doesn't exceed 125-170 lines, but a good quality lens should not have a problem hitting 80-110 at it's optimum aperture range. The big question of course is if that performance only holds in the center or across the frame. I think the poster was refering to the Nyquist limit and all of that stuff. But I agree with Ctein that most lenses should be able to handle 16MP, without much trouble, especially on a full frame chip.

One of the few lenses I have ever seen to hit over 200 lines was an Angenieux lens for broadcast and film production. I hit a true 250 lines, but also cost more than a Honda Accord.

I'd heard that one of the chief reasons that full frame DSLRS were so costly was that the price of silicon is more or less a given. 1.6 crop DSLRS get more sensors out of a given circle of silicon, hence they can be much cheaper. I'd heard that a full frame dslr sensor would never get below $1000 for the manufacturer, and this set the price of full frame dslrs.

I think I read that on Luminous Landscape.

So does anyone know if there is any truth to that? I literally have no idea what I'm talking about here, that's just something I read that seemed to make sense of some prices.

For me the real reason I got the 5d was that I had a Rebel XT and found the viewfinder was just not useable. I actually got claustrophobic using it. I was at the point where it was no fun or excitement to take pictures with that camera. I decided I'd either quit photography or get a camera with a better viewfinder. I found myself on a small pile of money (a very rare occasion) and bought the 5D and the 50mm 1.4.

Overall it's been a really good experience having that camera. I definitely formed a bond with the camera that I never would have with the Rebel. It was nice having the best camera in the world for a while. It made me push myself in the "well, I have no excuses anymore" way.

But I'm disgusted by the price of the camera, and by the afformentioned (by me) abysmal Canon repair service. And the price gouging on things like batteries. I certainly don't feel like a valued customer in the slightest. And glaring oversights like the lack of weather sealing on the 5D really soured me to Canon. I took my camera out once in a strong mist (extremely light rain) and it fritzed out and didn't work for 4 long and miserable days. I thought I had a $3000 paper weight. That kind of stuff seems like it shouldn't happen.

Sorry guys (Carl & Ctein) ...
"The idiot" (me) needs you guys to go slower...

1) The 35MP and 150MP maximum resolution was on a "DX" sensor?
2) Wouldn't diffraction be too significant for a 35MP sensor at f8 and f11? (Actually, wouldn't anything above f4 be a problem for 35MP and f2 for 150MP?)

mcananeya, This is a good point why continue the 5D, I also feel that the 1.3 sensor pushed to 14 megapixels with live view and 8+ frames/sec selling around $2,250 would be a great camera. But read below.
Lenses,megapixels,Nyquist Frequency, output print size and on infinitum. Every camera out there can make great photos, but every photographer out there will not make great photo's no matter what camera and lens he or she is using.
If Canon does come out with a new 5D it has to in the same low price range as the 40D-- with 16 megapixels, live view, 8 frames/sec at $2,250 and drop the 1D MKIII when they are all sold. It will be the lame duck camera.


I managed to dig up the Thom Hogan posts on this subject. Start here: http://forums.dpreview.com/forums/readflat.asp?forum=1030&message=26522548 and then follow his comments in the rest of the thread. Thom isn't the Wizard of Oz with a crystal ball, so his word isn't gospel, but a lot of what he has to say makes sense (to me at least).


Does the 5D class have to employ a full frame camera - could you not have a camera offering an APS sensor in the same class?

It seems to me Nikon's D300 and the Cannon 5D are in the same class. To my way of thinking "class" is determined by the whole feature set and priceing ...not just sensor size.



I think that is basically the key question. Nikon and Canon sort of went different ways on this subject around the time when the Nikon D200 and the Canon 5D first came out. The D200 was a solid body with a ton of features but with a sensor that some felt wasn't so great. The 5D had (and still has) a solid sensor with what some felt was a not-so-great body.

Nikon seems to have addressed most of the D200's flaws by improving the sensor in the D300, adding new features (like anti-dust shaking, LiveView, etc.) and further improving the speed and responsiveness of the camera.

So how can Canon update the 5D? If they upgrade the body to the level of the Nikon D300, they will have a mini-version of the Canon 1D Mark III that will presumably cut into a significant chunk of their 1-series sales while reducing margins.

They could put in a new, better sensor, but I suspect that if they had a better sensor, they would have put it into the 1D Mark III. That is why I said the only option would be to put the 1D Mark III sensor into the next 5D. They could make a few other changes in terms of features, but they would have to keep them pretty minimal. The end result would be more of a 5D Mark II than a 6D. I'm not sure it would be worth it.

I tend to agree that Canon would be better off making what I called the "no-holds-barred crop sensor camera" to compete with the Nikon D300.


Dear Harry,

If you're talking about on-film resolution, I'm right with you. Over all my years of film resolution testing, a handful of times I managed to get 160-165 line pair per millimeter on 35mm film. The biggest limiting factor was actually variations in the position of the film from frame to frame! To get those kind of on-film resolutions, you're talking about a depth of focus that in some cases is less than the thickness of the film emulsion, which means there is no margin for error whatsoever in the mechanical focus.

What I was talking about might be better referred to as "aerial resolution" -- what the lens is capable of projecting, unconstrained by the limits of film resolution and focus errors. If you include those limits and consider what it takes to get 160 line pair per millimeter on film, it is very clear that the lens is doing much better than 200 line pair per millimeter, aerial.

Personally, I don't think it's a big question (in the context of this thread) whether the lens holds that only in the center or across the entire field. Carl is absolutely right that most lenses can't maintain this kind of quality corner-to-corner, except under optimum settings. But until we can buy sensors that have smaller pixels in the center of the field and larger ones at the edges, it's not germane. The question was whether one could take advantage of more pixels to get sharper photographs and the answer to that is an unequivocal yes.

I am truly, truly impressed that you ever hit 200-250 line pair per millimeter in any format. That is an astonishing achievement, and I bow to your prowess. You have th3 m4d l33t skillz [ grin ].

~ pax \ Ctein
[ please excuse any word salad. MacSpeech in training! ]
-- Ctein's online Gallery http://ctein.com
-- Digital restorations http://photorepair.com

Dear Paul,

My considered opinion would be that that $1000 price limit is wrong. The larger sensor is more expensive because it consumes more real estate (fewer chips per wafer and a higher percentage of defective chips -- a double whammy). But the cost of the silicon itself is not a large part of the sensor cost (it is for solar cells). There is currently a shortage of high-quality silicon; demand has outstripped manufacturing capacity. That problem should go away by 2010, as more production comes online. Silicon *is* relatively expensive now, but that doesn't have a big effect on the full-frame cameras.

Market considerations are a different matter. If there is a consumer demand for full-frame DSLR cameras, I can confidently predict that the price of those will decline to under $500, current dollars, over time. Possibly considerably under. The real uncertainty is not the ability to drive down the manufacturing cost but whether there is sufficient demand for such a camera. If it turns out not to matter to most photographers whether or not they get "full frame," then there's simply no manufacturer gain by driving the cost that low. (Note that the kind of folks who hang out here are not representative of most photographers. This is a mass-market issue.)

To make an analogy to film cameras, there's a reason you didn't see many inexpensive medium-format camera systems. The few that existed were typically niche products. Often curiosities and often of lesser quality. The reason was simply that photographers who demanded medium-format quality were willing to pay substantially more for the cameras and the folks who wanted cheap cameras by and large didn't care about getting medium-format.

~ pax \ Ctein
[ please excuse any word salad. MacSpeech in training! ]
-- Ctein's online Gallery http://ctein.com
-- Digital restorations http://photorepair.com

Dear Jeff,

1) Carl and I are talking about 35mm-format sized sensors. Those pixel counts were also not supposed to indicate any kind of "maximum resolution" but simply the number of pixels that has VERY rough parity with the capability of good 35mm-format lenses.

2) The crude rule for diffraction- limited resolution is 1600/f# = number of line pair per millimeter. Lenses don't reach 100% of this resolution normally, but it's easy for a decent lens to hit 100 line pair per millimeter at f/8 (where the diffraction limit would be 200 line pair per millimeter). Really good ones can still hold it to f/11. (And as Carl so correctly pointed out, we are talking about the center of the field, here. Corner performance is much more sensitive to the particular lens design.) If you're trying to squeeze 200 line pair per millimeter or more out of your lens, you really don't want to be stopping down below f/5.6, as a matter of practical course.

The rules really haven't changed from what they were for film. Wide open, your lens is likely to be low in contrast and the difficulty of achieving accurate focus means it likely won't produce very sharp photographs (although the aerial resolution of the lens may be quite high). As you stop down, the contrast improves and the sharpness improves. At some point, usually somewhere between f/4 and f/8, diffraction starts to dominate and the sharpness starts to go down again. But it will still be good enough to make use of very high pixel counts, if you're working for that kind of quality.

~ pax \ Ctein
[ please excuse any word salad. MacSpeech in training! ]
-- Ctein's online Gallery http://ctein.com
-- Digital restorations http://photorepair.com


I had a brain fart. It was $1000 for the silicon for a sensor for a MF back. Which seems more likely, maybe?


Dear Paul,

Don't have current spot prices at hand, but my recollection is 12" wafers are running a few hundred bucks, so cost of silicon for a medium format sensor would be in the tens of dollars (you can pack circa 15 on a wafer, depending on the precise format).

I'll see if I can ferret out a firmer price later.

pax / Ctein

If it's like a technology company I used to work for, pricing goes something like this.

Just (and I mean just) before the announcement, the head sales guys go into a huddle and consider all of the following (in no particular order): prices of competing (and anticipated competing) models, volume of the current model still in the channel, how does the new model fit within the company's range (now and in the future), what manufacturing volume can they do in the short/medium/long term, what current exchange rates are, etc etc. What they spent on R&D and cost to manufacture are largely irrelevant.

I wouldn't worry. Canon have people that get paid to figure all this out.

"I am truly, truly impressed that you ever hit 200-250 line pair per millimeter in any format. That is an astonishing achievement, and I bow to your prowess. You have th3 m4d l33t skillz [ grin ]."

Hey Ctein -

Perhaps I wasn't being very clear, but I have not personally hit 200-250 lines with the Angenieux during testing. Sorry about the confusion.

But, I have spoken with people who have used and tested this lens (I'll see if I can dig up the model number, it's been a while...) and they are flabbergasted by it's performance. It's amazing what optical designers can do, when they aren't constrained by the size of the lens and eventual price tag of the product...


I think that Canon likes having a full-frame body in the market space currently occupied by 5D because its buyers probably also buy a lot of expensive pro "L" glass where they actually make their money. The 40D buyer may buy some but is also just as likely to buy a third-party f/2.8 standard zoom and call it a day.

The comments to this entry are closed.



Blog powered by Typepad
Member since 06/2007