brainiac wrote:
Although this is heading off topic, that link is very important reading. I hate to have to say 'I predicted this', but I did, and that's why I bought a 1Ds3 last December. Looks like the 1Ds3 is still about on a level with the D700 and 5D2 at very high iso. ;-)
To me it looks like the 5dII employs less in body noise reduction, noise and sharpness seem to be in balance between these bodies
brainiac wrote:
Although this is heading off topic, that link is very important reading. I hate to have to say 'I predicted this', but I did, and that's why I bought a 1Ds3 last December. Looks like the 1Ds3 is still about on a level with the D700 and 5D2 at very high iso. ;-)
I think this is consistent with what Canon, at least Wesfall, has said. As I remember correctly, with RAW, he said the 5D2 would only be slightly better than the 1Ds3. If the "slightly better" is correct, it might show up in measurememnts, such as 'rawanalysis' or DxO, but not be seen in practical work...eventually we will see.
What is remarkable is the relative pricing.
Not to take anything away from the D700, but comparing it to the 21.Mpixel sensors is a bit of apples to oranges, I think. A fine camera, but "different tool".
It is remarkable that the 5DII image quality is as good as it is given the price, no doubt about that. Perhaps it was just all the hype and build up (and the fact that Canon themselves were claiming the 5DII sensor was superior to anything from them thus far!), but I expected a one to two stop ISO advantage with the 5DII sensor vs. 1DsIII. That is definitely not the case. Looks like the only advantage with the 5DII is in jpg output(thus processing) and not in the sensor itself.
Off topic, but NIkon is going to have a hell of a time selling 8K D3x's with sub 3k Canon 5DII's and Sony a900's producing similar IQ, not to mention the reduced price 1DsIII's at just over 6K!
Eek! You know what I am going to say: this is a 100% crop from a 21 megapixel frame, so forget per pixel detail - it's creaming a 5D for image detail. You need to imagine uprezzing your 5D file to 21 megapixels and then comparing a 100% crop.
If this crop could be more detailed, then I would be inclined to look first at the noise reduction algorithm, and then the lens. When you push up iso on a 1Ds3 (and 5D2), the image detail degrades gracefully - the higher you go, the less detail is retrieved. That's normal.
For 1600 iso at 21 megapixels this is a great result - every bit as good as the class leading 1Ds3. Tariq, I'm afraid that your comment is exactly the kind of "5DmkII less sharp than 5D" comment which I set this thread up to pre-empt. ;-)
I know. My position is different in that I expect technology advances to produce at least the same(or very close) per pixel IQ at higher MP's from the same manufacturer. Anything else is sort of a cop out in my view.
I do agree with all of your other points though. 5D image interpolated up will not match a 5DII image and, as I mentioned earlier, I think that image looks damn good for ISO 1600. At a lower ISO, it would certainly resolve more detail.
Tariq Gibran wrote:
My position is different in that I expect technology advances to produce at least the same(or very close) per pixel IQ at higher MP's from the same manufacturer. Anything else is sort of a cop out in my view.
What I don't understand about the per pixel way of looking at it is that when one is shooting at very high isos, image sharpness is usually constrained by a number of other factors, including lens, aperture, shutter speed, subject movement, and shaky hands. Shooting at 100 iso is a different matter: at low shutter speeds one would hope to see the full benefit of all the pixels, and landscape shooters often shoot on a tripod through special lenses at selected apertures. We all know these cameras don't perform at their best at very high isos, so the question is not whether every pixel counts, but rather, does this camera make good prints in desperate circumstances? That said, per pixel comparison makes little sense at iso 100. One wants to see the difference between the two cameras, and equal magnification is the only fair way to make that comparison. Per pixel analysis just sucks, UNLESS the images have the same number of pixels. I simply don't believe that anyone is capable of reliably judging whether a file has a 29% advantage in resolution. As a result, it's practically impossible to judge whether one camera is better than another that way. The argument that per pixel assessment allows you to judge quality in proportion to the number of pixels the camera has is a total red herring. What we want to know is which camera produces less noise and more detail per square inch of print, and it simply doesn't matter how many pixels it does or doesn't take to achieve that result.
I think it would make more sense to scale down your 5Dii files to the 5D resolution than the inverse, but your point is valid. Higher resolution puts more demand on the optics.
I know we've had this argument before, but I don't think you're going to see those type of improvements per generation. You're setting yourself up for disappointment or you'll save yourself a lot of money by waiting a couple generations between bodies. The sensor quality is rapidly moving towards incremental improvements, not revolutions. Also, the 5dII sensor is basically the 1DS3's with a few polishes. They're saving a ton of money by not updating the sensor and using the prior production knowledge to move this sensor down the price bracket.
If you want "huge" improvements, we'll have to see how the 1ds4 does, or whatever form it comes in. IQ wise, I'm pretty pleased to have a 1ds3 without some bells and whistles for less than half the cost (for a lot of applications at least). Looks like a pretty great cam, all things said and done. That said, I'm perfectly happy with my 5d. If I need huge landscape prints, (and I'm nowhere near a good enough photog to be worrying about that) I'll shoot mosiacs. I don't mind the processing end of the spectrum.
So only by using the best of the best resolving lenses, camera/lens calibration, best computer/printer, outstanding camera techniques, and masterful in Photoshop can one begin to understand the greatness of a 5DII. You guys ought to get a room.
Barry Pehlman wrote:
So only by using the best of the best resolving lenses, camera/lens calibration, best computer/printer, outstanding camera techniques, and masterful in Photoshop can one begin to understand the greatness of a 5DII. You guys ought to get a room.
Yeah...and if your lens is out of focus it also makes it difficult :-)
Barry Pehlman wrote:
So only by using the best of the best resolving lenses, camera/lens calibration, best computer/printer, outstanding camera techniques, and masterful in Photoshop can one begin to understand the greatness of a 5DII.
It's certainly true that with better and better sensors technique becomes more likely to introduce the ceiling on image quality in most real world circumstances. But make no mistake, a technically illiterate first-timer with a good eye can pick up a 5D2 in green square mode and reasonable light and get stunning results from jpeg, printed at a high street lab. You don't need to know anything much about photography or Photoshop to see sometimes the advantages of these cameras. Excellent technique is only required if you want to see the best the camera can do all the time.
By the way, the green square mode will also shoot RAW nowadays not like before for 50D and they even added a CA mode for bigginers but I doubt so they will have the green square mode for 5DII
MountainTop wrote:
They did include "green square mode" on the 5D II - can anyone see a Mark II owner using it?
If I hand my camera to my wife, son, sister, friend, etc., I often put it in the auto (green square) mode; the unavailability of raw files in that mode sometimes convinces me to choose P instead. The 5DII's addition of this feature is a boon to me.
The fact that it does 3.9 fps suggests that the 1Ds3 could have done 7.8 fps if they had wanted to offer that, and that as we all know, these cameras are not the best that they can be in every feature, but are tailored to distinguish themselves from each other in the way that the marketing department wants them to. For instance, the 1Ds3 could have had a 6400 and 12800 iso setting, but Canon knows that pixel peepers and DPReview, in their ignorance, would have been scornful due to the lack of good in-camera NR.
You wouldn't give DIGIC IV some credit for performance in the 5D2?
Not to be argumentative, but what is the point of going for more megapixels, if there is little to no difference on print? What are we gaining besides bigger files to handle and store?
I haven't yet run comparisons between my 5D and my 5D2, but when I compared my various DSLRs with my long-time stalwart Mamiya RZ67:
An 8x10 portrait from my 300D and 10D compared to an 8x10 from my Mamiya, but not at any larger sizes.
A 16x20 portrait from my 20D could compare to a 16x20 from my Mamiya, but not at any larger sizes.
A 20x30 portrait from my 5D can compare to a 20x30 from my Mamiya, but not at any larger size.
From what I've seen of 1Ds3 images, I should be able to get a 30x40 portrait from a 5D2 that compares to my Mamiya...which should enable me to finally retire the Mamiya.
Notice that I've been specifying "portraits" here, which have different resolution requirements from landscapes.
Ran across these examples between the 5D and 5DII. Seems to me the 5D, while not quite up to the 5DII, is putting in a very good showing. I would have expected a greater spread and this, to me, does show that the 5DII is indeed less sharp than the 5D at the pixel level. So much so that even interpolated up, the 5D image sharpness is rather close to the 5DII.
Example showing the superior per pixel iq of 5D which, even when interpolated up, is close to the iq of the 5DII, even with a 50 1.4 at 5.6:
RDKirk wrote:
You wouldn't give DIGIC IV some credit for performance in the 5D2?
Not only is the Digic IV helping the speed of the 5DMKII, you can't expect 2 cards to double the speed. Parallel processing would have to be perfectly efficient to double the speed. Some coordination of the cards is required and in the past Canon has gotten about 1.6X times the speed instead of 2X times the speed. Now there is reason to believe that the Digic IV can process information faster than in the 5DMKII (it does in the 50D), so what we can expect out of the 2 Digic IV chips is that a camera could process about 150 megapixels per second. So as we look to a 1DsMK4, I expect to see an approximately 30 megapixel camera that can shoot at 5 fps. They could, however, make a 21 megapixel camera that shoots at 7 fps.
RDKirk wrote:
Notice that I've been specifying "portraits" here, which have different resolution requirements from landscapes.
I think this is the main point. Portraits are less demanding than shots of distant landscapes, particularly those that contain lots of fine, intricate detail such as foliage and leaves. A 5D landscape image falls apart at 20x30 whereas an image from my film Hasselblad and most probably your Mamiya would not. This is the case where more resolution and MP are needed. Certain other studio work, such as the reproduction of paintings or anything with fine detail will also benefit greatly from more resolution. Portraits are the least demanding of any subject matter in my experience.
