search the site
Canon EOS 1D Mark IV vs. Nikon D3s: ISO Comparison
Canon EOS 1D Mark IV vs. Nikon D3s: ISO Comparison
In one corner the 16.1 megapixel APS-H sensored Canon EOS 1D Mark IV. In the other corner, the full framed, 12.1 megapixel Nikon D3s. We’ve been playing with both cameras now and finally have our ISO shots from each to compare. The winner? No contest…it’s the Nikon D3s.
Order The Cameras
A Quick Tour
 |  |
 |  |
 |  |
 |  |
A Few Words
Don’t get dismayed Canon fans, compared to most cameras, the EOS 1D Mark IV does a very respectable job at high ISO, however the Nikon D3s just takes things to a whole new level, I hate to encourage you to skip past the fine print, but check out the ISO 1600 samples from the D3s, they look a lot like ISO 200 on most other cameras don’t they? It’s really just quite remarkable. Now of course the D3s has two key advantages going for it:
- larger sensor
- fewer megapixels
But in my opinion, Nikon also has chosen a wiser noise reduction path. If you look at the crops, you’ll see a lot of chroma noise in the Canon files, and none in the Nikon files. Instead Nikon leaves behind the luminance noise, the final result is a prettier image–for pixel peepers anyway. ISO isn’t everything, but in the real world shooting scenarios, improved ISO performance has its advantages even when ambient light isn’t that low. For example, while shooting a local high school basketball game, I was a bit hesitant about what I was seeing on the LCD of the Mark IV at ISO 6400, the ISO I needed to get 1/1000 for shooting sports, (at f/2.8 even). So instead I was reducing ISO and slowing shutter speed to 1/800, which resulted in slightly less sharp captures than could have ideally been captured. In similar environments the D3s would allow for the higher ISO setting, which also would have been cleaner than the Mark IV’s lower setting, and I could have shot at a higher shutter speed, the net result would have been sharper, cleaner shots. On the other hand…the Mark IV’s 1.3x crop allowed for a longer lens reach, and bigger files, that can be resized down, or alternatively could be cropped out of. There’s advantages either way, but my takeaway would be if your sports, wildlife, or journalistic photography involves low light environments to a great degree, the D3s should get preferential treatment, otherwise stick with your current system. Nikon shooters get a 1 and a half stop boost in ISO performance and 720p HD video, Canon shooters get a much improved AF, more extensive HD video options, and slight image and ISO improvements.
ISO Comparisons
In general I like to keep these tests at the same equivalent focal length, typically 50mm (and f/8) where possible. Due to lens differences, and processing differences in-camera, it’s important to not judge the sharpness of the results. For both cameras (all three actually) I lower sharpening in-camera to the lowest setting, and no sharpening is done in post. Actual in-camera results from both cameras are sharper at factory defaults. Noise reduction has been turned off in-camera, as well as dynamic range enhancers, so yes this is as bad as these files can look out of camera. One final note, I’ve thrown in the Canon EOS 5D Mark II, which is still putting in a fine showing here. The D3s and Mark IV go all the way to ISO 102,400, where honestly things aren’t all that pretty, in fact the Canon EOS 1D Mark IV’s crops are downright brutal.
Use the image below to orient yourself to the crops, and click each thumbnail for the full 100% view of the crop. Get the full Canon EOS 1D Mark IV Samples here and the D3s samples here. From left to right the cameras are D3s, 1D Mark IV, 5D Mark II. After ISO 25,600, it’s Nikon D3s on the left and Canon 1D Mark IV on the right.
[Edit: For a closer look at how the Canon EOS 1D Mark IV compares with the EOS 5D Mark II, check out our comparison right here.]
Crop 1
 |  |
| ISO 1600 | ISO 3200 |
 |  |
| ISO 6400 | ISO 12800 |
 |  |
| ISO 25600 | ISO 51200 |
 | |
| ISO 102400 |
Crop 2
 |  |
| ISO 1600 | ISO 3200 |
 |  |
| ISO 6400 | ISO 12800 |
 |  |
| ISO 25600 | ISO 51200 |
 | |
| ISO 102400 |
Order The Cameras
Similar Posts:
- None Found
Leave a Reply
About Neutralday.com
Established in 2007 by photographer Patrick Dean, Neutralday.com brings you the latest in photography news and reviews. With insights on photographers, equipment, and how to's, Neutralday.com is focused on and intended for the photography community.
Support Neutralday.com
Canon Instant Savings
Our Most Recent Reviews
Canon EOS 7D In Depth Review
Canon PowerShot G11 In Depth Review
Canon PowerShot S90 In Depth Review
Nikon D3000 In Depth Review
Nikon D300s In Depth Review
Nikon D3s In Depth Review
Panasonic Lumix DMC-FZ35 In Depth Review
Pentax K-x In Depth ReviewRed Hot Pre-Orders
Latest Tweets
resources
Best Sellers at Amazon
flickr
Could you please provide the 5DMII samples as well so I could do a resized comparison? I think it was interesting to se how the 5DMII did compared to the others.
I think I can do that, I’ll dig up the files later today.
I’m currently shooting a D3 and a 5D mark 2. I’m looking for a comparison such as yours that compares the D3 with the D3s. Are you aware of any or could you put one together comparing the higher ISOs.
Thanks,
Len
Len,
I can’t think of one off-hand… though I would say since you’re covered with video via the 5D2, perhaps waiting for the D4 might be better. You’ll only be gaining a stop and half of ISO (some say just one stop) and that’s basically it (unless you have an early D3 with the original buffer–then the D3s provides more buffer). Perhaps DPReview.com?
could you tell us where each camera is placed on those pictures ?
is it D3, 1D, 5D, or D3, 5D, 1D ?
Thanks a lot
Sorry about the confusion, it is the D3s, 1d, 5D2, (from left to right). (Didn’t save my last edit!)
To my eyes, at 1600, 5d2 has the best detail/noise combination. Look at the leaves and you won’t see much details on D3s, and 5d2 visibly has more details. At 6400, they probably equal out. D3s prevails after that. I am surprised to see that 1d4 is not as good as 5d2.
The 1d4 isn’t full frame, unlike the other two cameras.
It will be interesting to see how the 1d4s does!
The crop factor matters if the picture are of the same FOV, which FF has about 70% more light than 1.3Crop sensor (1.3×1.3 = 1.69), that is a 2/3 stop advantage. However, in order to just measure sensor’s single pixel performance, it should use the same focal length with different FOV. I don’t know which way Patrick is using, from the result, I am guessing he uses the same FOV method.
I feel the biggest D3s advantage is that it provides a really good “build-in” NR; it is so good that for most ppl, no NR is needed.
I am in no way an expert, but I feel if at ISO1600 there is virtually no Chroma noise, that only means NR is applied.
Thank you for pointing out the fallacy of evaluating different formats at the same ISO.
The invariant should be the same total light flux on the sensor, for that determins the noise floor.
Using the identical ISO between cameras,the actual difference in flux is 0.8 of a stop, if 1.32X is the Crop Factor.
The test could easily be published with the ISO biased by 2/3 or 1 stop, as you suggest. The 1D4 would appear in a more favourable light. Literally
Too bad Nikon doesn’t have any useful wildlife lenses with VR.
to jesper,
Nikon has a full range of long tele vr lens
http://lmgtfy.com/?q=nikon+telephoto+lenses
There is no holy grail!!!!
When resizing the 5DMkII image crop to match the D3s and apply some NR they are virtually identical in these ISO 25600 samples. Nikon didnt make any breakthrough but merely matched Canon in lowlight ability. You could suspect there is some kind of NR in D3s RAW already applied even though it is switched off but either way I could get the 5DMkII output to match D3s at high ISO.
http://farm5.static.flickr.com/4043/4287330020_3653470559_o.jpg
I just downsized the 5d mKII and it still looks extremely noisy in comparison to the d3s. What did you do to your downsized sample?
Did you add some NR?
what’s the point of that?
any camera’s output will get cleaner with NR in post processing (even the D3s) which you could clean even further since they start up cleaner. Yet that’s another discussion. Yet, the less NR you apply the more fine texture detail, less color bloats and general better image you get.
This is why sensor output is more important than post processing NR in reviewing the true ISO capabilities of a camera and why avoiding NR as much as possible is generaly desirable.
The author wisely chose to compare the base line images because it is assumed you can clean ANY from ANY camera further using ANY number of NR processors.
Wouldn’t the D3s be even better if you applied NR to that image too? You can’t add NR to one image and not to another; wouldn’t be fair. However, it could be interesting and in some aspect more fair to compare the pictures at the same size (downsizing the bigger ones).
I would love to have the “D700-version” of the D3s when it arrives. D700 was about halv the price of the D3, with the same performance.
If I apply NR to D3s image it would look really soft. Suspicion falls to D3s already has NR applied even though camera says it is turned off. Regardless, the fact remains I could get the 5DMk2 output to match D3s so Nikon has only matched Canon in lowlight IQ in its sensor technology.
Why can’t Canon users just accept that Nikon D3s technology is ahead of Canon?
Nikon doesn’t apply NR to their IN-SPEC ISOs (200-12800), as that would be discovered pretty quickly by dxomark and others.
We need more controlled comparisons.
so the nikon image get softer? nonsense.
A D3s image gets cleaner (not softer) and so does the 5D image. The net result is that BOTH get about a 1-2 stop improvement allowing the D3s to retain the same advantage it start with. This is a fact that applies to any two cameras not just nikon and canon. NR = bad which means the less you need to do it, the better.
Your second point is also moot because ALL raws have NR applied to them even when it is OFF. This is a well known fact.
I agree with Simon. Take images taken by the MUCH cheaper 7D (18MP), downsample them and you get images that are about the same as those taken by the Nikon D3s! That’s the power of resolution (see Popular Photography for explaination).
Hehe! Why not go even further and buy a Pentax k-x? Still even cheaper and with enough fiddling maybe you could get similar results. I wouldn’t say it would be the same though. I’ve compared online pics of k-x and D3s, and to me it seems as if k-x at 6400 is as good as D3s at 51200.
Don’t get me wrongo, I love the k-x (my wife has one) and it seems slightly better at high iso than my own 20D, which I also love. But lets face it, the D3s is just simply the best there is at the moment.
I’m a low light photo fanatic (to me the vulgar f word spells “flash”). I’m exited to see what the smaller version (compare D3 / D700) will cost. For me, the D3s is simply way too expensive and too heavy to use.
I don’t understand why Canon’s guy just can’t except that in real-world picture there’re more noise in 1Dmk4 than D3S. and that’s not strange because it FF VS APS-H, Less Pixel VS More Pixel!!.
****(For my own preference, Since I own both Canon and Nikon I found that in the past 2 years Canon ISO performance is not “always better” than Nikon anymore)*****
What you trying to say about cropping or resizing down is already been known for a along time and that is other case.
We just talk about the real world performance, The raw picture that the sensor could get.
And please!! High-ISO performance is really useful in some to many situation
and yes! Nikon picture render its self a little darker than Canon, (I should say Canon is brighter.)
I agree. I’m just an enthusiast willing to be pro. I own a Canon 450D and read many reviews of Canons, Nikons, Pentax, Olympus and Sonys. I’ve noticed the same you did. Canon does not seem to find a good answer to its chroma noise issues and keeps on pushing the pixel count race (which it seems to run against itself). I guess luminance is preferable to chroma noise (the Nikon approach.I think the idea is to get the best output directly from the camera and save time doing pp, and in this respect Nikon took the lead with the D3, D300 series and the D5000. Low light photography needs high-quality high ISO performance and here Canon is falling well behind. On the other hand, Canon’s brighter images are, in my opinion, the result of a widely used technique of shooting to the right (overexposing) to retain detail, which I think is regarded as correct.
clearly a D3x 24mp output file downsampled to 12MP will help noise levels and may match the D3s output. Does that mean the D3x is better? no. it’s much too slow to compete against the D3s.
However the 7D output is much too noisy (noisier still than the 40D according to dpreview). The 5DII has a cleaner output which helps but it’s old AF system and slow fps are no good for action. Thus the only alternative is the 1D4, not a 7D.
Furthermore, the other side to the “power of resolution” is loss in color depth and dinamic range. The 5D, D3s and D3X using full frame sensors giving them far more room for larger pixels which transform in better quality than the 7D could ever attain which crams 18mp in an aps-c format.
http://tinyurl.com/y8lc659
thus the 7D is also not a real alternative to the 1D4 or D3s.
IF canon had put a full frame sensor on the 1D4, it may have stood a chance. Yet an update will not becoming for a good 3 years. The nikon D4 is expected next year.
With all due respect, you can not compare a full frame 12 MP with an APS 16 MP. Size of sensor pixels does matter when comparing ISO. On the other hand, I was curious about the difference in overall luminosity in the Nikon pics, a lot darker than the Canon ones (specially 5DMII). I would say there is about 1 stop, so in practice you are really comparing Canon ISO 25.600 with Nikon ISO 12.800.
+1 You should read the above reply closely. Also, You must compare at the same output size. Pixel to pixel comparisons penalize the sensor with larger absolute pixel-density.
well, the “Lord of Darkness” is the best, but where are the primes for him?
14-24mm rule them all in the wide range.
Tele, nikon too has full range of tele lens.
What do they actually lack off?
Thanks a lot for taking the time to do these tests and post the results.
Despite some people’s comments about the difference in sensor size and resolution, it is very, very interesting to see those comparisons. In fact those differences make the test even more interesting.
It’s very remarkable the image quality achieved by the 5D Mark II, keeping in mind it is a FF 21MP sensor vs. a FF 12.1MP (D3s) and APS-H 16.1MP (1D Mark IV).
It would be VERY interesting to see a comparison on movie/video performance, quality and capabilities too, since it is one of the new feature that will be on every DSLR since now on.
Thank you very much for sharing these tests.
Best regards.
Just to prove there is no fluke here is another from those samples. 5D2 downsized with NR to match D3s
http://farm3.static.flickr.com/2784/4287493700_71dda2db78_o.jpg
They are very similar IQ wise in absolute terms if everything is equal. Now you could argue D3s may have NR applied even though they claim NR is off but I am always been suspcious of manufacturer’s claims.
Also D3s images appears underexposed by a stop which made their samples look less noisey. I have made that allowance by reducing exposure/brightness so on the 5DMk2 images so the exposure look similar in both cameras.
I totally agree.
Many website compare IQ at 100%.
I hope reader consider the fact that if you compare the images at the same resolution, the higher res image would look as cleaner.
I really appreciate if the author to do a resolution downsized comparison as well.
So at the end it is different paths to get images, and in that respect the Nikon is doing better. Though I have a Canon, I feel Nikon is thinking more photographically than its archrival.
This test is not that scientific… the exposure of D3s is different from 1D4 and 1D4’s is different from 5D2. If the apertures used are the same, it means 1D4 uses lower ISO to get brighter images to get the result of higher ISO setting in D3s. So, you should compare 1D4’s ISO1600 vs D3s’s ISO3200 (for example)
Great review!! Hopefully next time you will compare the citrusy flavour of Oranges V Pumpkins. They are both round, both orange, so lets apply the same logic you have here!
You don’t have to be sarcastic. We are all adults.
If you don’t see it useful pls shut T F Up
Thank for the reviewer.
If I don’t need speed, I think 5DII still my choice.
Not that D3s are bad, but I have all the L lens.
No need to be petty and abusive. While the the 100% crops are different because of the resolutions, this review for me and many others, is still very useful.
I see big problems in using the camera jpgs. Many tests revealed NR and sharpening of the jpgs even with turned off features. Especially the 5D markII samples look strongly sharpened. I own one of these and the noise patterns shown here is for sure strongly enhanced by sharpening without NR.
To compare the real ISO performance the samples should be processed from RAW. Further, downsampled samples from the high resolution sensors would be very useful. With my 21MP I can decide whether to use the full resolution and more noise or to downsample with less noise.
It is the weak AA filter. On the same not eyou could argue D3s has turned up NR even when the camera says NR is off because the image looks very soft.
Maybe the samples from 5DmkII are slightly larger so you could see more details?
Dude, 21MP is larger than 12MP
The Nikon seems to be losing quite alot of detail in the shadows. If high ISO is affecting the DR of the camera, i’d say Canon is the better option.
The “s” in D3s means soft!?
Nikon Rules!!
Let’s wait for another comparsion.
I use Nikon & Canon and can believe the 3Ds might be
better in highISO but I can’t trust the comparsion between
5DII an 1DIV)
Do you think Canon is really that stupid to allow
this to happen?
it’s a matter of physics and electronics.
full frame sensor VS cropped sensor.
full frame wins (this is why FF is still big $)
I suspect once results are out for dinamic range and color depth, the nikon will pull even futher. This is not to say canon is stupid. It just means they should have gone full frame.
Wow, thanks for putting this together. I have to say that the D3s at 6400 and 12800 is downright amazing! In all of these tests up to 6400 the 5DII looks better than the 1DIV to me. I’m sure the 1DIV has other advantages, being a real pro body, but for many people the 5DII will be the better camera for the money. I like the detail a little better, even though it has a little noise.
I’d be interested to see a similar comparison of these cameras, but with the noise reduction left on. After all, here you are basically testing the sensor, not the camera as a whole. They build in noise reduction to part of the overall system and for the end user, that image is what counts.
D3s at 12800 still has usable picture!!!!!!!!!!!!
Amazing.
Still what amazes me the most is the performance of 5D mk2, it’s better than 1Dmk4 on 1600 and 6400 ISO.
Thanks for sharing this key comparison
Cheers
First, I’m not sure why the complete fascination with turning off NR. What should matter is the best image quality the camera can deliver. If the NR masks detail that’s a knock against a camera of course, but otherwise the no NR exercises are a bit silly without any comparison of NR performance. Who is going to use high ISO with no NR and sell or display their results that way?
Second, as a few others have pointed out here the Nikon images appear to be underexposed by a stop. This would cancel out one stop from the one and a half the review here claims is the difference in ISO noise performance.
Third, if these were not processed from raw then it is possible that some NR was applied to the Nikon images while not to the Canon. Perhaps that is not the case but it should be made clear exactly how the displayed images were processed.
Finally, this review continues the pixel size mantra that continues to be chanted by Nikon fans. They appear to think that somewhere about 12 MP is the holy grail for sensor size. What this mindset fails to realize is that taking the images from the higher resolution sensor and comparing them with the lower res sensor in a pixel peeping test, without any resizing, puts the higher res sensor at a disadvantage. Of course resizing can introduce additional issues if done wrong, but with good software and proper technique the higher res image could be downsized and become sharper without increasing noise, or perhaps equally sharp while reducing noise.
The bottom line is that the one and a half stop of difference found by this reviewer is likely to disappear. The final images would then subjectively be better or worse based purely on the tastes of the viewers, with no clear advantage as suggested here.
Amen. Even though I’m happy the tester shared his pictures with us I still come to the conclusion that this test wasn’t done properly.
I think there is some denial going on here. canon chose a 16 mp in a cropped 1.3 sensor and that’s why the images are noisier.
a few thing to keep in mind:
1) Nikon ’s 3 year old D3 does better with it’s FF sensor vs the 1.3 crop canon. This is because you just can’t cram so many pixels in a small space and hope good noise performance color depth and dynamic range.
http://tinyurl.com/y8lc659
The canon 5DI would also beat the 1D4 in noise and it is even older. There is just no getting around the full frame advantage.
2) Nikon always waits for canon to go first. Nikon’s D3s is a stop gap measure while the D4 (the 1D4’s true rival) is finished (scheduled for 2011). It will be over 12MP. Thus there is no such thing as a holy grail. Canon does this in-between releases too. They did it for the 1D2n.
3) Nikon took the lead in resolution back in 2008 with it’s 24MP D3X to challenge the 1DsIII’s 21 MP. Thus they do plan to up their game on their upcoming round of cameras.
4) just as nikon released the D3 after canon released the 1D3, there will be a D4 and D4x tid for tad for the 1D4 and 1D4s. Nikon simply moves a about a year after canon does. This is obvious to anybody watching the two companys’ flagship release years.
I owned a 5dMarkII for over a year before selling it and now on the Nikon d3s. These shots reflect my experiences with both cameras. On the 5dMarkII anything above ISO 1600 were unacceptable, to me anyway. With the d3s I can get usable shots at double that ISO. Setting the auto ISO feature to on and a minimum shutter speed of 1/1000s will give excellent shots at sporting events even when the ISO goes as high as 12,800.
The feature you mention is some of many that Canon SHOULD take in account and give us via Firmware Update.
It’s completely possible to achieve, the hardware is there, only some programming is needed, and that means no redesign nor so expensive costs to keep good sales on a camera (5D Mark II) that has HUGE potential.
Sure, it’s not the best choice for sports, but still has LOT more to offer.
Not to mention the movie/video features…
- For those interested on the CINEMATOGRAPHY capabilities, TAKE A LOOK:
http://5dmark2.wordpress.com/2009/12/17/canon-firmware-updates/
We are planning a NEW Campaign soon to request FILMMAKING features… Stay tunned.
canon’s highly compressed mpeg4 video still ways to go, and rolling shutter issues need to be resolved.
both companies (nikon more than canon) need to hurry up and provide RAW video as well as global shutters (or at least faster sweep rates).
the 5D/7D/1D4 video also suffer from interlacing artifacts because the video is not enconded from a full frame capture and scale down. What canon does is capture every so many pixels and then go from 21mp to 1080p. This is because currenty the canon processor is not fast enough to downscale a full 21mp sample on the fly.
however as full of drawbacks as the canon video is, it is pretty good and certainly better than nikon.
Still for real serious applications, both cameras are toys compared to a RED camera, capable of RAW video at higher FPS and up to 2k – 4k resolution. But I guess they are neat toys for boys to play with
You mean Aliasing (not interlaced) artifacts. And honestly it’s not that bad at all as some people say. And you can correct it a bit with sharpness level and filters or other techniques.
Aside that, every camera has “moire” issues (different than aliasing) when you shoot specific pattern-subjects, so that’s a bit out of topic.
I personally guess you don’t have the 5D Mark II, because you wouldn’t call it as “toy” if you had it, even comparing it with RED.
The capabilities of the 5D Mark II are excellent, and surpass RED in low light.
RED are (in many cases) more a myth than a reality… And if you think to buy a fixed lens “cheap” RED, then the differences between these cameras would be really noticeable. If you choose a full equipped expensive RED, the difference would be even more (price wise in this case).
They are indeed different tools, that share some segments, but also differ a lot in some aspects.
The “jello” effect in the 5D Mark II is not that horrible like the Nikon D90 or some other people might say. You need to test it by yourself.
Of course it has limitations, like EVERY camera, but it has GREAT potential, no matter the H.264 compression, which in fact has a great quality, and is not so bad at all as some people may say.
Those are more “3rd party” opinions (or from those who don’t have nor use the camera) than a reality.
Of course it would be more “professional” if it had RAW output, but that’s not a real limitation to use it professionally.
We have heard about Canon interested to provide RAW output for this camera. That’s to be seen yet…
But in any case, if you know the limitations (like EVERY camera has) and know how to deal with them, you CAN get professional results. It’s not only the tool itself, but also the technique and art of the operator/user/director what bring good results.
Great professionals results have been already achieved with the 5D Mark II.
That’s worldwide known, and that’s why we are requesting to Canon to IMPROVE and IMPLEMENT via Firmware Update some really possible FILMMAKING features to this camera, which still has LOT of not exploited potential inside.
Probably we’ll launch new campaign soon…
Canon CAN improve this camera a LOT more.
I thought it had been proven that down-sizing an image had little positive effect on the luminance noise? Or maybe that was just some mis-information spreading around DPR? Hmm I seem to remember being scolded by a pixel peeper the last time I suggested down-sampling and applying a bit of NR… Oh well, I stopped caring which was better the day Nikon got a half decent 1600 and 3200, with the D700… This next level is no longer a race to me, just an improvement…
=Matt=
As some guys mentioned already, the exposure of D3s is a lot different from 1DMk4, and even from D700. I saw a test determining the correct exposure of D3s and in that test, at the same aperture, D3s’s ISO3200 is equal to D700’s ISO2200. So to say that noise level of D3s is more than 1 stops better than D700 is quite a bit of exaggeration, although it was necessary for commercial purpose.
As a long time professional Canon user and cps member who now just switched over to Nikon and the D3s I can honestly say that the D3s is indeed a better low light machine. Nikon asked me to try out the original D3 and I wasnt completely blown away . But the D3s is simply incredible. Whats amazing about it is that the high iso colors still stay saturated, the blacks stay black without being riddled with blotchiness or grey patterns, there’s virtually no noise even at iso 6400, and on top of it the files are very hefty… meaning theres a lot of information that can be sucked out of the shadows and highlights. I’m not saying the 1d MKIV is a bad camera (although the 1.3x crop has to go) but the D3s is simply on another level. The AF also feels faster and more accurate, especially in low light situations.
On the flip side, the D3s files are softer than Canons’ straight out of the camera. But there’s also a huge difference in sharpness when using NX. I found ACR really makes the files soft. But again, the files can take a huge amount of sharpening in comparison to Canon files. Nikon needs a new 135mm to compete with the beautiful 135mm 2.0 of Canon. I will miss that lens. Also, a 35mm prime would be nice.
Just my two cents.
Wouldn’t a better test be with the Canon 1Ds Mk IV? Could you do another test then?
it would be a nonsensical test because the 1Ds is canon’s slow-high resolution body. These are both fast sport bodies.
their lineups for the next 4-6 years will look like this
1D3 -competes-> D3
1D4 -competes-> D3s & (D4 around 2011)
1Ds3 -competes-> D3X
1Ds4 -comptes-> D3X & (D4X about 1 year after D4)
their cycles are off-set by a few years so you’ll always have an older model facing a newer model no matter which brand you get. The real problem is canon didn’t make the 1D4 full frame, and the D4 will certainly be full frame. But there is always a 1D5 3 – 4 years out:)
This is why switching over the changing winds is silly and costly since it’s a game of ping pong.
Amazing performance by D3s! It looks like it’s 3 to 4 stops better than the D300.
Wow, 60 comments and no kind words for my lovely product shots? First thanks everybody for stopping in, it is much appreciated, and the great comments, nearly all constructive, thanks for those too.
I realize that the test isn’t perfect, but it does give one a “sense” of what each can do. As someone pointed out, it isn’t apples to apples, full frame vs crop, but I assume its a comparison folks want to see. Also yes the D3s underexposes, (so does the Pentax K-x, I suspect this is a Sony sensor thing). Even so, it is quite clear out of the camera the D3s is better, you could resize and clean the Mark IV files for sure, but this would involve so many additional variables that I suspect I’d have a 1000 angry comments. I’m still playing with the Mark IV but it remains clear that even at factory default settings it is noisier than the D3s, but still very, very good—only the D3s makes it look less than stellar.
Considering it’s 16.1 megapixel, 1.3 crop Canon vs. 12.1 megapixel, full-frame Nikon, one would expect the Nikon to do better at higher ISO’s. Duh!
I don’t know what the authors here want to make. Most probably advertisement for Nikon.
The crops for the bottle label for the Canon D1 are NOT consistent with the pictures at flickr! Adiitionally the pictures from the Nikon are for sure softened. The Sensor cannot be that soft. You can clearly see that on the feathers.
When looking at the flikr pictures for 100.000 Iso I’m really impressed of the performance of both cameras.
spacey,
I keep hearing I’m advertising for Nikon, but I don’t understand why? I have a Canon 5D Mark II and love it. It just seems pretty obvious that the D3s is better at low light, which isn’t saying anything too profound given the two cameras pixel density.
This is 2 different cameras who can not being compared.
Nikon uses Fuji’s pixel poolish technology. The results gives a chalky rendition. Other point, what about sharpen settings on the Canon?
Both cameras set to no sharpening. None. And none post. From what I’ve seen the with the D90, D300s, and D3s, the Nikons tend to be a bit soft when set this way.
The Nikon looks very very soft and it’s clear that the camera has applied noise reduction and a lot of it. But TBH it’s done it very well. If you look at a few crops you can see the noise marks.
I think the problem is a comparison a very different cameras. The 1d has 1.3x crop. The 5d is aimed at semi-pros and is priced much lower. The Nikon is like the 1ds – a full frame professional camera. Personally I’d take the 5d at high ISO. Downscale it 50% and a little NR (to compensate for the in camera Nikon NR) and I think at ISO 6400, the Canon would have the edge. (In ACR no chroma is visible with most cameras untill ISO 3200 at least anyway.)
But straight from the camera I agree the Nikon wins.
p0ss3s3d,
It is a bit like comparing apples to oranges to pears, still it’s fun to see what’s coming out of the cameras.
The Nikon D3s is amazing! wow!
Very nice test! I’m honestly not too impressed with the 1d4, and think Canon should have gone FF. But, I am not the target market.
The D3s is very impressive, no doubt, but hard to do exact comparisons as its photos look maybe 1/3 to 2/3 of a stop underexposed.
The 5D Mk II looks great.
Thanks for the show. Yes its not a full test and comparison as everyone wants; but at least it you got us on a start here.
As it the IV comes into use, I would be curious how much we can change on the quality.
with any of these would firmware improve the IQ at high ISO or is this a baked in level?
Thanks florian,
Just remember these files are as bad as they can look, in-camera factory settings are cleaner and sharper, and most of do run some NR on our higher ISO images, even when they’re “clean”. As to your point, I’m doubtful any substantial improvements could be made to overall IQ, but I do notice a bug with random “yellowed” captures, as if switching WB from capture to capture. Quite strange, but happens enough to be noticeable.
Curious: this was all done in RAW, right, with the respective converters from Nikon and Canon?
Nikon looks like a winner, the fair price of $ 4000.
Sorry for my English. Nikon looks like a winner. Canon EOS 1D Mark IV should cost no more price $ 4000.
Excuses, excuses, excuses.
So typical of you Canon fanboys.
Small processing in Photoshop. RAW option unfortunately was not.
http://filex.kz/?action=download&id=1264221100788993154
It’s true that the Nikon D3s performs better than the Canon EOS-1D Mark IV at high ISOs, but… the D3s’s samples aren’t too beautiful either!
for both Nikon n Canon also a good camera …
is nothing can argue and nothing is perfect
and D3s is FF and 1dmk4 is 1.3x sensor already different
dont remember if D3s switch to 1.2x the mp will drop …
y not compare like this … for the quality
One thing that stands out is the Nikon images are about ONE STOP under the Canon’s exposure. That would definitely hide more noise in the shadow area for sure. Also the new Lightroom 3 Beta has a brand new imaging pipeline that does an awesome job killing all color noise… which Canon needs for sure. Would like to see raw files processed through the best appropriate raw converter… which ever that might be.
Thanks for the comparison!
John
It does indeed underexpose, I could be as you said part of their intentional processing strategy. I run these test for individual reviews, and though I’m tempted to balance histograms everytime, part of what’s interesting is how every camera handles the same scene.
mmmm, i’m gonna agree that this isn’t a fully executed test. aside from the same resolution issues, the lack of RAW files in this test is kind of crucial. JPG should never be used to compare cameras. No matter how much it is or isn’t processed.
still not sold.
Nikon shutter speed about 20% slower than the Canon. The Nikon is getting more light AND the exposure is still dark. Seems like Nikon is fiddling with the data to get better reviews.
Apparentky the noise reduction of the Nikon can´t be switched off. I find this behaviour a bit strange for a professional camera body. The 1D is the only one of the tree where noise reduction is really switched off. Of course this gives it a even bigger disadvantage in a comparison test with noise reducing cameras than its smaller sensor would give anyway.
So we have to wait until someone presents a similar compariason test using raw photos with identical noise reducion before we get meaningful information about the sensor performance.
Why are all the pix at the same size? If you cropped the samples of the 1D Mk IV there is a 30% blow up. The only possibility for a fair comparison was to use another lens and go 30% closer to the subject.
The noise reduction of Nikon always was stronger and delivered ‘nice’ results but far away from reality. I sold my D3, bugged by the dust on the Sensor, what always was a problem (the same on the D3X) and got a 1D Mk IV a few days ago. The results at higher ISO are extraordinary good, far better than ever believable!
Also the movies at high ISO are exceptional good and in full HD, which is not delivered by Nikon.
I decided to stop reading the boring tests and go out for photos
The one thing Nikon has is no pattern noise in the shadows or high ISO, which makes it very easy to fix in post. Canon has pattern noise, criss-cross color noise patterns, which can not be fully corrected. The new LR3 Beta improves this very much but doesn’t eliminate it.
Why is the images from Canon EOS 1D mark IV much lighter than the ones from Nikon D3s. The background is much darker. And the hole image is much darker. If the images from Nikon D3s should look like the ones from Canon EOS 1d Mark IV you should go up ½ev or even 1 ev. This will mean more Iso. Why is the Nikon images i full screen view so dark.
The Nikon D3s appears to underexpose relative to the Canon EOS 1D Mark IV. We tend to leave the tests as is to show how cameras interpret the same scenes, so one could assume that Nikon is underexposing a bit to a. boost ISO quality and b. protect highlights. Regardless even underexposed the Canon EOS 1D Mark IV has more noise than the D3s.