brainiac wrote:
I am also curious about a fair comparison at iso 12800. Banding becomes a problem at such high isos, and I found it more of a problem on my 1D3 than on my 1Ds3. I no longer have the 1D3, so I can't easily do a comparison now.
If I ever had to shoot at 12800 I would get a Nikon D3. But I haven't yet had to shoot an event in a dungeon that didn't allow flash...
ChrisDM wrote:
If I ever had to shoot at 12800 I would get a Nikon D3
...but why? I started this thread with an iso 12800 image from my 1Ds3, which I think is a match for a D3 image at that speed. What makes you think a D3 will do better at that speed than a 1D3 or 1Ds3? Or do you need to shoot jpeg?
brainiac wrote:
...but why? I started this thread with an iso 12800 image from my 1Ds3, which I think is a match for a D3 image at that speed. What makes you think a D3 will do better at that speed than a 1D3 or 1Ds3? Or do you need to shoot jpeg?
ChrisDM wrote:
While I do appreciate the comparison here, I find it academic. These two cameras have very different feature sets (10 vs 21mp, 5 vs 10fps, ISO3200 vs ISO6400, etc) and are clearly intended for different purposes. While it is certainly impressive that the 1Ds3 can nearly hold its own against the 1D3 in terms of high ISO performance, it is the last reason I would pick it up over the other. It's like choosing a truck over a car in a road race simply because it has more horsepower... The 1D3 is the ultimate tool for sports/action/event/wedding etc shooting, not just for its slightly cleaner high ISO performance, but for its 6400 ability, more appropriate file sizes, much faster FPS, etc... But when it is resolution you need, for tasks such as landscape and commercial work, there's no better tool. Makes a good backup wedding camera too...
Once again, neat comparison, but only proves that these cameras are much more different in many more ways other than their 3200 ability. I share my opinion here not to stir a debate, but for the benefit of someone who may be reading this thread trying to decide between the two, from someone who owns and uses both cameras extensively.
A comparison based on features alone is also a bit academic IMHO. There are lots of nature photographers catching fast paced action with the 1Ds3. There are also lots of portrait / landscape photographers that use a 1D3. Nonetheless you are right that both cams were designed for different purposes by Canon. Whatever suits your shooting style, desired output and budget I would say. It is only a good thing that high ISO performance isn't a consideration to choose any of these cams over the other because differences seem to be only marginal. OTOH 5FPS over 10FPS might in some cases be a bigger difference
Daan B wrote:
A comparison based on features is also a bit academic IMHO. There are lots of nature photographers catching fast paced action with the 1Ds3. There are also lots of portrait / landscape photographers that use a 1D3. Nonetheless you are right that both cams were designed for different purposes by Canon. Whatever suits your shooting style, desired output and budget I would say. It is only a good thing that high ISO performance isn't a consideration to choose any of these cams over the other because differences seem to be only marginal. OTOH 5FPS over 10FPS might in some cases be a bigger difference
...Show more →
I agree. Although I shoot mostly landscape I recently bought a 1D Mk III. I don't need the speed but did want all the rest of the pro features. My idea has always been to buy the camera that feels best in your hands. They can all do the job but if it doesn't feel right you won't like using it.
Thanks for the links, Chris. To me they don't show what I suppose you want them to. Lest it seems like I am just refusing to see what is obvious to all these eminent reviewers, I am obliged to explain, in each case, why the comparison is misleading:
1 The shortcomings of the Dpreview test I have spoken of before. First, the images are not resized, so we are not seeing the extent to which noise affects the picture, only the extent to which is affects a pixel. This is misleading since the 1Ds3 has 21 megapixels whereas the D3 only has 12. The D3 needs to show much less noise in order to match the 1Ds3. Also, the 1Ds3 is not tested at iso 6400 and 12800.
2 The LL iso 1600 comparison makes the same mistake. The 1Ds3 image is shown at higher magnification than the D3 crop. Also, zooms of questionable adequacy were used, which handicaps the 1Ds3 even more than the D3. Here are the two crops as shewn on LL, and with the D3 crop resized by me to show why equivalent magnification in crops matters:
D3: http://cyberphotographer.com/lld3v1ds3/d3.jpg
D3 uprezzed to same magnification as 1Ds3: http://cyberphotographer.com/lld3v1ds3/d3up.jpg
1Ds3: http://cyberphotographer.com/lld3v1ds3/1ds3.jpg
The LL reviewer states that the D3 is one stop better on the strength of these crops. which seems wrong to me.
3 The Dpreview forum example shows iso 12800 crops in which the Nikon file has clearly had in-camera chroma noise reduction whereas the 1D3 crop hasn't. Allowed chroma noise reduction, the Canon MK3's can match the D3. The D3 has the great advantage that it obviates post-processing at these speeds, which is important for the press, but with raw and noise reduction processing the D3's advantage isn't clear.
4 The fourth link once again compares chroma-noise-reduced Nikon images against Canon files which have had no such NR. It doesn't test the 1Ds3, and the choice of a live dog as the subject, with varying composition and focus makes this test almost completely opaque.
A proper test between the D3 and 1Ds3 needs to be done with the same Nikon lens, the same lighting and shutter speed. The 1Ds3 and the D3 files need to be processed properly, each to its advantage, probably in DPP and NX. Appropriate chroma noise reduction needs to be applied to each file as it would be in reality, but the Nikon file may not need it since the camera electronics seems to do NR itself. Colour and contrast should be matched as closely as possible between the files. Finally the D3 file should be uprezzed to 21 megapixel. Only then will we see a print-alike comparison between the cameras.
The laziness of review sites like LL and DPR in showing crops at different magnification is terrible and thoroughly misleading. If you are going to link to web sites which show comparisons, please make sure that the tests are done fairly. I didn't set this thread up to consolidate internet mythology. LL and DPR are contributing to the popular delusion.
brainiac wrote:
The laziness of review sites like LL and DPR in showing crops at different magnification is terrible and thoroughly misleading. If you are going to link to web sites which show comparisons, please make sure that the tests are done fairly. I didn't set this thread up to consolidate internet mythology. LL and DPR are contributing to the popular delusion.
From the 1DsIII review just posted on DPReview, it looks like the 1DsIII takes fourth place in the high ISO competition, behind the 5D, D3, and (since it has the identical sensor to the D3) the D700.
Obviously you are in the minority in thinking the results need to be handicapped in favour of the 1DsIII to make the comparisons "fair". When you start thinking that everyone who disagrees with you is delusional, that's called paranoia...
molson wrote:
From the 1DsIII review just posted on DPReview, it looks like the 1DsIII takes fourth place in the high ISO competition, behind the 5D, D3, and (since it has the identical sensor to the D3) the D700.
Obviously you are in the minority in thinking the results need to be handicapped in favour of the 1DsIII to make the comparisons "fair". When you start thinking that everyone who disagrees with you is delusional, that's called paranoia...
Not when all of those people are making a basic scientific error, publicly, in their assessments. Magnification matters. It's a scientific fact. A pixel from a 21 megapixel camera is a smaller constituent of the image than a pixel from a 12 megapixel camera. That's a fact. For that reason, 100% crops favour the lower resolving camera, because noise is being assessed per pixel, rather than per square inch of image. If you don't understand the physics of that, you are in esteemed company, but that doesn't make any of you right. 100% crops should only be compared after the images have been resized, preferably upsized, to the same number of pixels. I've demonstrated the effect twice in this thread, with the DPReview oak tree crops, and the Luminous Landscape eye/spectacle crops above. How do you explain the conflict between what those crops show once they are magnified correctly, and what the reviewers say?
Does it make sense to compare a file with chroma-noise-reduction against one without? These two issues are not paranoid delusions, but reviewing mistakes made on almost every site. These sites are not showing you what kind of results you will get when using these cameras in the real world with a typical raw workflow. There is a delusion going on here, but it ain't me.
Edited by brainiac on Aug 18, 2008 at 01:55 PM GMT
brainiac wrote:
The laziness of review sites like LL and DPR in showing crops at different magnification is terrible and thoroughly misleading. If you are going to link to web sites which show comparisons, please make sure that the tests are done fairly. I didn't set this thread up to consolidate internet mythology. LL and DPR are contributing to the popular delusion.
Actually, this goes back to my original point. Your argument is that for fair comparisons to be made, samples from other cameras must be "uprezzed" to match the file size of the 1Ds3. Once again, perhaps in the purely academic world of argumentative debate this may have some relevance, but what myself, and these reviewers care about is real world relavence... If I'm going to make a 40" print, I'm not going to shoot it at 6400 and I'm not going to use a 1D3. I'm going to use a 1Ds3. That's what it is made for. But if I'm going to shoot a low light event, where I don't need large prints but I do need the ability to shoot at 6400, then I'm going to use the 1D3. That's what it's made for. Plain common sense, once you understand the body of features designed around each camera.
So, back to the relevance. If I was going to race my Jeep against your sports car in a track circuit race, what if I told you that you had to put fat tires on your sports car and fill the trunk with sandbags to make it "fair"? We don't care what's fair, we care about what's the best tool to get the job done. If I'm in a road race, I want a sports car. If I'm off road, I want a Jeep. It's quite simple.
Once again, what these tests, yours and all of the others I linked to, is that these cameras are more dissimilar in many other features than they are in terms of high ISO performance. But then again Canon didn't give the 1Ds3 the ability to shoot natively beyond 3200 anyways, so once again the point becomes trivial. And speaking of trivial: The 5D was the high ISO king last year, this year it's the D3, next year perhaps the 5DII, and probably the year after that the D4, then the 1DMk4, blah blah blah... So are we going to continue to chase and compare, or realize that what matters most is to match the entire feature set of the camera to the tasks at hand? The truth is that for low light work and for standard low light print sizes, 6400 on the D3 or 1D3 is finally "good enough"... Now we photographers can get on with the business of taking remarkable pictures and leave these trivial debates to the "gear hobbyists".
> Your argument is that for fair comparisons to be made, samples from other cameras must be "uprezzed" to match the file size of the 1Ds3. Once again, perhaps in the purely academic world of argumentative debate this may have some relevance, but what myself, and these reviewers care about is real world relavence...
It's not academic at all. The high iso performance of these cameras is highly visible in 10x8 inch prints. The failure to correctly assess high iso performance, and the consequent myth that the D3 is significantly better at high iso isn't academic, it's costing switchers a fortune.
> what matters most is to match the entire feature set of the camera to the tasks at hand
This thread is about high iso and the 1Ds3. Whether reviewers are assessing its high iso performance correctly is relevant here. Whether it can shoot at 10 fps isn't. While there are countless other things to discuss about these cameras, that's probably best done on other threads.
> If I was going to race my Jeep against your sports car in a track circuit race
Last year I made a choice to use D3, 1Ds3, 1D3 or 5D. One of my main requirements is performance shooting people in very low light. I could use any of these cameras to get my work done, and I suspect that's true for many non-sports photographers. They are far more alike than a sports car and a truck.
What I'm saying here is that, although nobody has run a fair race because of preconceptions and a misuse of the stopwatch, if you do, you may well find that the sports car can also win off-road.
DPReview is WRONG when it says high pixel count (i.e., small pixel size) is a disadvantage.
We should put a stop to this fallacy.
High pixel count provides greater flexibility. With a 21 MP sensor, one can either print large at low ISO or resize (e.g. to 12 MP) & print small at high ISO with the same image quality as large pixel sensor cameras.
thw2 wrote:
DPReview is WRONG when it says high pixel count (i.e., small pixel size) is a disadvantage.
We should put a stop to this fallacy.
High pixel count provides greater flexibility. With a 21 MP sensor, one can either print large at low ISO or resize (e.g. to 12 MP) & print small at high ISO with the same image quality as large pixel sensor cameras.
mill4570 wrote:
What settings were you using for the "dust and scratches" on the guitar pic?
I can't remember now, but I think it was a modest d&s on the blue channel, which tends to be the worst. Since then Pixel Perfect has drawn my attention to a much more effective method using layers, which allows the dust and scratches filter only to affect a specific luminosity range.
John Power wrote:
Of course the soon to be released 5D2 will have better high ISO performance that the current "king" at about 1/2 or less the price...
Rather bold to say considering the camera isn't even released yet...
thw2 wrote:
DPReview is WRONG when it says high pixel count (i.e., small pixel size) is a disadvantage.
We should put a stop to this fallacy.
High pixel count provides greater flexibility. With a 21 MP sensor, one can either print large at low ISO or resize (e.g. to 12 MP) & print small at high ISO with the same image quality as large pixel sensor cameras.
They don't say that pixel count or pixel size is a disadvantage, they say pixel density is a disadvantage. And it is... However it isn't the only factor in determining a camera's noise characteristics. It is the most reliable first indicator however. The point is that if you're going to make 40" prints, you need a camera with more pixels, and the potential downside is greater potential pixel density, depending on the size of the sensor. And of course the opposite is also true, if your goal is low noise and high sensitivity, the camera with lower pixel density has the early advantage (Unless you're going to make 40" prints at high ISO, but who does that?). There are of course other factors that come into play, but pixel density is of course relevant.
I do commercial and event photography, so I have a camera with high resolution for my commercial work, and low pixel density for my event work. Each is better suited to its particular task better than the other, based on many factors including pixel density. And yes, I agree that higher pixel density is the more versatile of the two options. If I could have just one camera, it would of course be the 1Ds3. And my "walkaround" camera of choice is the Pentax K20D, and I chose it for many reasons, the first being its much greater resolution over the 40D. Of course its a little more noisy at high ISOs than the 40D, but that's the price I'm willing to pay for having so many pixels... More isn't always better, but it usually is! But no manufacturer has yet to overcome the laws of physics. Some mask it better than others though.
Weather sealing, built-in IS, ultralight ultracool Limited Pancake prime lenses, greater resolution, DNG file format and in-camera RAW development, LCD color calibration, dust mapping, and last but certainly not least: real Auto ISO... The K20D just simply spanks anything that Canon has to offer in this category now. Hopefully the 50D will have half of the incredible features of the K20D, but I doubt it.
