Rob has today published an extensive first look article on the Mark III.

Nill
~~
www.toulme.net

Interesting! Nikon users take note of the following.



I sometimes love stirring the pot .

I think Canon really made something good here. I mean the ISO performence is fantastic. Now just give me a 1DsMkIII

This does look like a great camera, but they just keep getting heavier -- it's five or six ounces heavier than a D2x, which I thought was already a beast; and with the heavier Canon lenses to capture the same field of view, the difference is even greater. In fact, it's only ~ half a pound lighter than an H3D including digital back. When the D3x come out, I hope they go the other way: I'd like to see them cut a few ounces off.

JC

But still a tad lighter than the 1d2n

It's tough for a landscape shooter to find much pleasure in another (albeit near perfect) 10 megapixel camera.

It wouldve been nice for Canon to make the 1Dmk3 12mp instead of 10, yes, not much difference but marketing wise it gives the body more longevity since there are plenty of 10mp DSLRs out there.

Image quality wise the 1Dmk3 has nothing on the 5D from what ive seen. \

Naturally the 1Dmk3 as a package is impressive, it has many improvements in a lot of areas, lcd, speed, configurability, live view, seonsor clean, wireless capability etc...

So for now Im keeping my 5D and 1Dmk2 combo. Il wait for the 1DsMk3...and wait...

Yes, and Canon is not planing on keeping that camera so long, you will see we will get a new fast camera before the Olympics, perhaps 16MP FF.

I suppose the AF problem mentioned by RG is largely responsible for the slip in shipping dates.

It wasn't clear from his article, but I would guess that you don't need the wireless WFT-E2/E2A to connect a USB drive? This just made field storage a lot easier since I don't need to duplicate PDA's to get field backups (I don't own a notebook) and you could (I assume) read the pictures back from the usb drive to ensure they were captured...

I get decent 17x25's out of my 1d2 - decent, but not great... I think 12 mpx would have been the tipping point for me, but I'll print out some of RG's samples and take a look.

OT a bit, but it was cool to see Calgary (where I grew up) as the location for his shots, I actually recognized the building in downtown in the crane shot!

1Ds Mk III this summer, PLEASE! I'm pathetically drooling ...

If you're willing to come to Toronto to collect, I'll bet you a beer, no announcement before end of Sept and no delivery before the end of Nov.

Oh, c'mon Tim! You're so negative! You know these people always deliver before the expected announce date!

That was wishful pleading and whining, not a prediction, though I wouldn't mind going to Toronto and hang - it's a great city. Unfotunately I have to go keep going to work and show sufficient ROI to my employers so they'll keep me at least until after I've paid for the danged thing, whenever it does come out.

Look at the ISO! That 3200 is amazing and the 6400 isn't bad either.

I hope this signals a shift in the manufacturers away from the megapixel race to perfecting the MP counts already there. (Yes, I know this is blasphey to say on a landscape board).

Landscape shooters are in need for DR as much as resolution, but a lot less in need of clean high ISO.

My guess is that both Canon with its 1ds3 and Nikon with its D3x will increase the MP a bit, but focus mostly enhanced DR and transitions to highlights.

Cheers,
Bernard

No blasphemy at all. Sure, more megapixels is good, but not at the expense of dynamic range, low noise, etc. But of course we'd rather have the low noise, high dynamic range, AND more megapixels ...

New announcements 3rd week of August. Most likely a 30D replacement, and one other body, don't know if it will be a 5D or 1 series replacement.

Given that the 1D MkIII sensor both reduces noise levels and increases resolution by decreasing photosite size, I would instead hope (and expect) that it signals a continuation of the trend to improve both resolution and visible noise levels, not some totally lop-sized obsession with noise levels at high ISO as the most important measure of a camera's virtue.

Surely it is clear that a lot of high end photography has a significant need for more than 10MP, and that for many photographic situations, higher resolution is more important than the extremes of high speed, low-light ISO performance. 10MP does not meet the resolution needs that lead many photographers to often choose high resolution, low speed, low dynamic range films like Velvia.

Indeed, given the need in this sport/PJ oriented camera for high frame rates and the ability to achieve high shutter speeds in sometimes limited light, it is to be expected that its design will lean toward fewer, larger photosites than in most other cameras. This has been a consistent distinction of all three 1D series sensors. It seems clear to me that upgrades to high end models (not to mention some of this year's new advanced amateur models) will to offer increased resolution, to beyond 10MP and even perhaps beyond 20MP, as a major selling point.

My predictions for later this year include 12MP or more with quite good noise levels at up to ISO 1600 in advanced amateur "APS-C" format DSLRs at under US$2,000. In particular, in a Sony CMOS sensor evolving from its now almost three year old D2X sensor. Probably from Canon too, but as usual Canon gives away fewer hints of its plans. And this will soon enough push the resolution asked of entry-level 35mm format DSLR's beyond the 5D's 12.7MP.

I wonder if 20MP brings any real detail advantage over 8-12MP with 35mm format lenses. Last time I was seriously concerned about megapixels (5 years ago) I did quite a bit of reading on the subject and came to the layman's conclusion that 8MP is enough to bring out all useful detail in the vast majority of real-life situations - ie. real subjects, available light, no tripod, non-controlled environment. Therefore I waited until an 8MP dSLR became reasonably priced, and 30D was just that.

I'm sure studio or tripodded landscape photographers can get more than that out of the finest SLR glass, but just how much? 12MP? 20MP? 30MP?

I would much prefer to get more shadow detail out of my 30D. I shoot a lot of long exposures (30+ seconds) in low light, and bracketing and digital blending is a must if I want to get good results.

Just as a reminder of what's available here:

http://www.luminous-landscape.com/essays/back-testing.shtml

compares 1ds @ 11mpx all the way up to a P45 at 39mpx plus scanned LF film

Thanks, had forgotten about that test. To my eyes the difference between the 11 and 16MP Canons is largely academic and only obvious when pixel-peeping. This is especially so in the field - this shoot is obviously the absolute best-case scenario. It certainly doesn't come even close to the environment I shoot in (available light and/or handheld).

I don't know how you shoot or what, but I know that I see a difference between my 5D and 1DsMk2. It always depend on what you want to do with the files, certainly when printing A4 or A3s the difference isn't that big, but go to 24" by 30" or 30" by 40" and you will see a difference. Always as starting point you do it right and use the right stuff.

I really wish for a 24MP camera, with low loise, shooting speed is not important. And more DM is also very welcomend.

You don't need a tripod or a studio to see real world differences between a 8MP and 16MP file. I have the 1DsII and 20D and there is a world of detail in the former that's not present in the latter. Shooting editorial--I always can pick out the lower resolution image--and we're talking full page (a4).

If you shoot with little light, handheld and at ISO 6200, sure. If you shoot landscapes with Leica or Zeiss glass, on a tripod, ISO 100 and print a 17x24 image, I'll take the "academic" 22MP over 11MP any day. I'll even pay $Ks for it. OK, call me a fool .

I don't think a discussion over this sort of thing will go anywhere. This is a landscape site and the majority of users feel a certain way. Now go over to sportsshooter...

I shoot editorial and while I enjoy the resulotion of a 5D or something larger than your average dslr I would rather see DR, shadow resolution/noise, etc first. But, that's because I shoot a certain way, as does everyone.

On the other hand, for shooters who stitch landscape, the difference in megapixels may be less important, an extra few frames won't be too much trouble and stitching time won't be effected due to the smaller files.

That said the same argument would apply to the current 1D mkIIN, at iso 100 and for landscape work the advantages of this camera are not that many unless the 14 bit shows a big difference in RAW.

There are more differences than pixels when comparing the 20D and 1Dsmk2

I think these discussions never mean to go anywhere, but they do provide some entertainment value, and occasionally even some bit of information .

But what you said is absolutely true, select the right tool for the particular job and approach used.

Not really. The images shot in the exact same lighting scheme produce similar images. It's how much detail that resolved that really shows the difference.

Soon this amazing new camera will be as forgotten about as the highly touted Leice M8 which has all but vanished from these discussions. Unless of course it turns out better than the Leice...maybe it wont have that dreaded purple cast which so many people paid $5,000 for.

Hey, I've forgotten about Arca Swiss cameras, but that doesn't prevent you from capturing great images ...

The Canon 1D series can be considered mainstream top-of-the-line, while the M is more of a nitch camera, so I'd say it should have more staying power. But who knows?

There's a law of diminishing returns at work as you increase pixel count. It's quite clear to me that the 12.7mp of the 5D produce a noticeably more detailed image than the 8.2mp of the 20D, assuming same FoV.

It's also clear that the 20D with same lens, same f stop and same shooting position produces a more detailed image than the 5D when the image is cropped to the same FoV as the 20D.

The conclusion is that a 20-22mp upgrade to the 5D would be worthwhile. I would be very doubtful about an upgrade from 12.7mp to 16.7mp simply because so much argument has resulted from camparisons between the 5D and 1Ds2. Ultimately, those who claim that the 1Ds2 produces more detail are correct, if you make a print large enough, but the very fact that there has been so much disputation about it is a clear indication that the differences are subtle.

Generally, I would say the greater the pixel count, the greater the increase needed to get a worthwhile improvement in detail.

In a studio environment (low iso) with prime lenses at optimum appertures and good strobe lights the 1Ds mk2 produces more detail than the 5D, Ive seen it and have zero doubts about this, it isnt a huge diff but its there.

The problem is, even with the 5D you start requiring the use of the best Canon glass at optimum appertures to see the full potential of the camera. Naturally with the 1Ds mk2 is an absolute must. With 22mp 1Ds mk3 you start being close to the limit of the best canon glass at optimum apertures.

Naturally there are other factors like the body itself (features), and other IQ characteristics like noise, dynamic range, smoothness etc

Where's the evidence for this? If you are talking about performance near the edges of the frame or performance at full aperture, then I agree. Most lenses are not good enough for any FF sensor whatever the pixel count. However, within an image circle of around 30 to 36mm diameter, not even a 22mp FF sensor will do justice to even a moderately good lens.

The following 200% crops compare resolution at f8 of 4 different lenses, moving clockwise from top left, the Canon 24-105 IS zoom at 50mm, the TS-E 24mm, the supposedly very fine Canon 50/1.4 and the el cheapo Canon 50/1.8.

The camera used was the Canon 8.2mp 20D, which is effectively a 22mp full frame sensor cropped.

There are qualitive differences in the structure of some of the lines. To my eyes the cheapest lens in this test produces perhaps the cleanest result. At least the more expensive 50/1.4 doesn't appear to have any advantage at f8, compared with the 50/1.8.

However, as regards sheer resolving power, they seem all about equal to me.

[attachmentid=2400]

Yes I was talkig across the frame of a full frame, not the reduced area of the image circle the smaller sensor uses. But its still close to the limit. To prove this you need a trully extensive and expensive test. There are many factors including sample lens variations between even "identical" lenses.

But I haven't used identical lenses. They are all different, with Photodo test results ranging from just 3.3 for the TS- 24 to 4.4 for the 50/1.4, yet they all show remarkably similar absolute resolution, though it's quite clear the 50mm primes are displaying better quality lines, straighter and neater etc.

What's perhaps even more remarkable is that the situation doesn't change much at f16, except it's now clear the 50mm primes resolve just a few more lines than the other two lenses, perhaps half a dozen or so, and the quality of the lines from the other two lenses hasn't improved by stopping down. Who said all lenses are equal at f16?

These relatively small differences between f8 and f16 seem to imply to me that a 22mp FF sensor is not nearly sufficient to get the best out of current high quality prime lenses at their sharpest aperture. Even though no 35mm lens is quite diffraction limited at f8, the better lenses should be able to resolve considerably more than an extra 6 lines at f8 than at f16.

The black spot is approx 10mm from the centre. The 50/1.4 that Photodo tested has a fairly flat MTF response at 40 lp/mm out to 18mm from the centre.

[attachmentid=2403]