Probably. Canon is making noises about having upgraded the sensor, but I expect it to be barely tweaked. So far 5D2 high iso performance doesn't look much better than what I get from my 1Ds3. Excellent, and on a par with the Nikon D3/700.
One well respected journalist, Vincent Laforêt, who used the 5DII for a full week-end reported on his blog absolutely outstanding high ISO performance, and he is familiar with Canon 1Ds. I do not repost the link, as it has already been posted multiple times.
So for sakes of argument - at a medium/large size print of 16 X 24 of the same subject, same position, same lense, same light - how much difference in detail/tonality would there be in an image from a 5D as compared to a 1DsIII as the sensor should be close to the 5DII? Without knowing which image came from which camera, would there be enough of a difference to tell?
It appears that many/most (all) of Brianiac's doubters are writing (speaking) theoretically. I can tell you that Mr. B is right because I routinely shoot the 12 MP Nikon D3, the 10 Mp Mk III and my 21 Mp 1Ds Mk III right next to each other.
I process and print the images side by side form only RAW, and after a 100K images, I can tell you that can and do adjust both CR2s and NEFs to my own standards (color balance, contrast sharpness etc. etc.) so it is not the default file settings that make the difference.
There is a difference and it is not theoretical it is real-often subtle-but real. This is not to say that my D3 and MK III won't make fine 24x36" prints (and yes something will/must up-rez even 21Mp files to print that large).
philber wrote:
One well respected journalist, Vincent Laforêt, who used the 5DII for a full week-end reported on his blog absolutely outstanding high ISO performance, and he is familiar with Canon 1Ds. I do not repost the link, as it has already been posted multiple times.
Frankly, considering the dog's breakfast that most reviewers and many photographers have made of assessing noise, high iso, and detail, because they fall into the trap of comparing crops from 21 and 12 megapixel cameras without resizing for equal magnification, I reserve my doubts about Vincent the forest's experience of using a 1Ds3 at iso 12800, and his ability to fairly assess that performance.
SoundHound wrote:
........................... I routinely shoot the 12 MP Nikon D3, the 10 Mp Mk III and my 21 Mp 1Ds Mk III right next to each other.............................................
There is a difference and it is not theoretical it is real-often subtle-but real. ................
This post actually says something. Real experience not bogged down in ranting on about theory and blasting away at others.
george malamis wrote:
So for sakes of argument - at a medium/large size print of 16 X 24 of the same subject, same position, same lense, same light - how much difference in detail/tonality would there be in an image from a 5D as compared to a 1DsIII as the sensor should be close to the 5DII? Without knowing which image came from which camera, would there be enough of a difference to tell?
The tone is more like the 1D3, i.e. noticeably smoother than a 5D, but the detail is noticeably finer than the 5D too. Most people will be disappointed by the increase in detail - it is an incremental improvement, but many will be subliminally impressed by the tone. In a 16x24 print it isn't hard to distinguish between the two cameras, AS LONG AS THE SUBJECT MATTER AND TECHNIQUE/SHARPNESS ARE APPROPRIATE TO SHOW THE DIFFERENCE. There will be cases where the difference is hard to see, and cases where it is obvious. 5D prints seem to have less texture, because the smaller pixels of the 1Ds3 give images a very nice, very fine grittiness. It's hard to describe this, but it is a very pretty effect, and many skilled printers try to introduce slight grain into very large 5D prints. There is less need to do this with 1Ds3 files, as prints often have a lovely etched look.
The only valid comparison is an equally sized print made from each camera at native resolution. I've seen your up-res posts about differing resolution SLRs and find it amusing that you want to introduce interpolation into the lower res file to make it LESS clean to match the noise characteristics of the higher res camera.
Scientific method, bro? Actual critical evaluation of captured data without error/manipulation from a software package added in?
Next time I go to my favorite restaurant, I'm going to add a ton of salt to the food so it's on equal footing with MacDonalds.
brainiac wrote:
I reserve my doubts about Vincent the forest's experience of using a 1Ds3 at iso 12800, and his ability to fairly assess that performance.
Some people just can't see the forrest for the trees...
Alex 1.0 wrote:
The only valid comparison is an equally sized print made from each camera at native resolution. I've seen your up-res posts about differing resolution SLRs and find it amusing that you want to introduce interpolation into the lower res file to make it LESS clean to match the noise characteristics of the higher res camera.
Scientific method, bro? Actual critical evaluation of captured data without error/manipulation from a software package added in?
Next time I go to my favorite restaurant, I'm going to add a ton of salt to the food so it's on equal footing with MacDonalds.
It is my opinion that uprezzing a file more than say 10% or 20% isn't destructive to actual pictorial detail at all. It's destructive to the sharp boundaries between pixels, but we specifically don't want to be able to see those when printing, so clearly compromising those nice sharp edges isn't going to affect image quality. It is important to understand the difference between pictorial detail and sharpness. Sharpness can be enhanced with a filter; detail can't. When people complain that uprezzing a file is somehow destructive to the picture data, they are typically complaining about the pixel boundary sharpness, not the image detail.
Below is an example of a file uprezzed using nearest neighbour to twice the baselength, and the same file uprezzed using bicubic which is supposedly non-conservative of detail. Both crops are shown at 200% so that you can really assess how damaging it is to uprez:
Bearing in mind that pixel edges in the first crop will not be resolved in the print, hopefully, it's worth looking carefully at the legibility of the text on the box label. Has the legibility of the text been compromised by uprezzing? If not, then how destructive is uprezzing bicubically?
If you'll forgive me for bringing Fourier analysis into what should be a matter of common sense, there are sound mathematical reasons why uprezzing by a significant margin is non-destructive to image data at frequencies less than the frequency of a pixel.
I've said it before, and I'll just have to go on saying it until sanity returns to the web: uprezzing a file does not turn a unicorn into a pantomime dromedary.
...to return to the topic, how much less legible is the text in the 2nd image? If each file were printed, do you think you would be able to discern the shortfall in legibility of the text in the 2nd crop?
I don't think I could, and consequently I agree with Fourier, and the established physical science of signal processing, rather than what some dude said on the web. Have no fear of uprezzing. It won't damage your picture beyond all recognition, if at all. Don't take my word for this, try it yourself. You can even try adding a bit of sharpening after the uprez to see if any perceived loss was picture detail from the sensor or just sharpness which you can add with a filter.
More importantly, the extent of any damage, if there is any, is negligible, whereas the effect of comparing 100% crops at different magnifications is very noticeable, and easily enough to cause you to make the wrong decision.
Always uprez the lower rez file, or both, to the higher rez file's dimensions, or bigger, before comparing any two cameras, if they have different numbers of pixels.
Alex 1.0 wrote:
The only valid comparison is an equally sized print made from each camera at native resolution. I've seen your up-res posts about differing resolution SLRs and find it amusing that you want to introduce interpolation into the lower res file to make it LESS clean to match the noise characteristics of the higher res camera.
Scientific method, bro? Actual critical evaluation of captured data without error/manipulation from a software package added in?
Next time I go to my favorite restaurant, I'm going to add a ton of salt to the food so it's on equal footing with MacDonalds.
Producing and equally sized print from each camera at native resolution will introduce error/manipulation from the printer software won't it? In addition, you are adding another variable...the potential resolution from the printer itself. I agree that a valid comparison should look at the printed output, but it's not the whole story.
Both methods of comparison are legitimate and can tell you enough about a camera's noise and detail characteristics to favour one or other. If printing is your ultimate purpose in taking the picture, then a very large print, although expensive, is a slightly better method of making the comparison. If you generally take pictures for display on the web, then there isn't a great need to compare by printing. But if you do want to compare 100% crops on screen... you know what I'm going to say.
