Original Link: https://www.anandtech.com/show/2680
First Look: Canon 5D Mark II
by Wesley Fink on December 4, 2008 3:00 AM EST- Posted in
- Digital Camera
The Canon 5D Mark II update to the original Canon 5D had been long expected and rumored since 2006, just a year after its introduction. However, when Canon finally announced the 5D Mark II about 2.5 months ago, Canon still managed to shock the industry. As we mentioned in Canon Announces 21.1MP 5D Mark II, most expected a solid update to 16MP or so with performance similar the current 5D - updated to include the current sensor cleaning, a larger LCD, dynamic range expansion, and hopefully an even better ISO range.
It was not until just before the announcement when some specs were leaking that everyone realized Canon would use an updated version of the 21.1MP sensor used in their $8000 1Ds Mark III, along with the latest Digic 4 processing and a 10-stop ISO range that extended from ISO 50 to ISO 25600. To both significantly extend the sensor resolution and extend the ISO range to beat the champion 12MP Nikon D3/D700 was something no one really expected.
Since that announcement, we have all been wondering if the full-frame 5D Mark II was for real - and it's been a long wait. Yesterday we finally received a production Canon 5D Mark II and we have been busy since. We couldn't wait to take a closer look at ISO performance and noise, and compare the performance of the 5D2 to the 24.6MP Sony A900, which is the current resolution champ, and the Nikon D700/D3, which are the current ISO champs. While we were at it we also took a closer look at performance compared to the original Canon 5D, which developed quite a reputation the last three years for color accuracy, high resolution, and low noise.
Many pages could be written on the handling and features of the Canon 5D2, and they will be written by us and others. For now, we are trying to answer the question of whether the 5D2 was worth the wait and the hype. Is performance improved in the update to the 5D? Does the incredible 10-stop ISO range compare favorably to the current Nikon D3/D700 and does the 5D2 become the new ISO champ? Did Canon find a way in the 5D Mark II to simultaneously increase resolution to an impressive 21.1MP and increase the ISO range in a manner that would do the legacy of the 5D proud?
After the somewhat disappointing performance of the Canon 50D skepticism runs high. The 15.1MP 50D was expected to take the resolution crown in the APS-C market, but instead it basically tied with the 12MP D300/D90/Sony A700. This is certainly not a poor performance by any means, but it did not match Canon's marketing claims, and it brought into question Canon's claims about the performance of their new sensors.
We hope in the next few pages to shed some light on the answers to these questions that are on the minds of anyone looking for a full-frame DSLR in the suddenly fully populated $3000 category. Is the Canon 5D2 everything we had hoped it would be?
Full-Frames and APS-C Compared
The Canon 5D Mark II is the update to the camera that created the $3000 full-frame DSLR market - namely the Canon 5D in 2005. Canon had this market all to their self for the first two years, but during the past year Nikon and Sony have both introduced several models to compete in the full-frame DSLR market.
The 5D2 sensor is approximately the size of a frame of 35mm film, which is 24x36mm, and resolution is 21.1MP - a significant increase from the original 5D at 12.8MP. The new Canon 5D2 compares very well to the reolution of the recently introduced Sony A900, which at 24.6MP is the highest resolution currently available in a full-frame camera. Compare this to most other DSLR cameras today, where the sensor size is closer to APS-C size. The smaller APS-C sensor used in most other cameras results in the lenses appearing to be 150% to 200% longer than the marked focal length. On the Canon 5D Mark II and other full-frame DSLR cameras lenses behave exactly as they would on 35mm, with no crop factor.
In the computer world, smaller and smaller traces mean higher density, more transistors, and generally better and faster performance. However, the digital sensor is not a digital device; instead, it's an analog device that gathers light and converts it into a digital signal. Thus the reverse is true in sensors in that larger sensor size is almost always better, with everything else equal. As you can see in the chart below, the APS-C sensors aren't even half of the area of a full-frame sensor. With a range of 28.1% to 42.4% of full-frame size, there is clearly a lot more information that can be potentially captured with a full-frame sensor.
DSLR Sensor Comparison | ||||
Camera | Effective Sensor Resolution | Sensor Dimensions and Area | % of Full-Frame | Sensor Density (MP/cm2) |
Olympus E-520/E-3 | 10 | 13.5x18 2.43 cm2 | 28.10% | 4 |
Canon Xsi | 12.2 | 14.8x22.2 3.28 cm2 | 38.00% | 3.7 |
Sony A350 | 14.2 | 15.8x23.6 3.72 cm2 | 42.90% | 3.8 |
Pentax K20D | 14.6 | 15.6x23.4 3.65 cm2 | 42.20% | 4 |
Canon 50D | 15.1 | 14.9x22.3 3.32 cm2 | 38.40% | 4.5 |
Sony A700, Nikon D300, Nikon D90 | 12.3 | 15x23.5 3.66 cm2 | 42.40% | 3.3 |
Nikon D700/Nikon D3 | 12.1 | 24x36 8.64 cm2 | 100% | 1.4 |
Canon 5D | 12.8 | 24x36 8.64 cm2 | 100% | 1.5 |
Canon 5D Mark II | 21.1 | 24x36 8.64 cm2 | 100% | 2.4 |
Canon 1Ds Mark III | 21.1 | 24x36 8.64 cm2 | 100% | 2.4 |
Sony A900/Nikon D3x | 24.6 | 24x35.9 8.61 cm2 | 100% | 2.9 |
The last column in the chart is the one that tells the story most accurately, however. Here the resolution of the sensor is divided by the sensor area to yield a sensor density. The lower the density, the larger the individual pixel size, and the more information that pixel can gather - all else being equal. There are a few surprises here, such as the Sony A350 being essentially the same density as the Canon XSi, and the new Canon 50D having the highest density of any current DSLR camera.
The last column does put into perspective the true potential of the full-frame sensor and sheds some light on the true meaning of Canon's 21.1MP sensor and Sony/Nikon's 24.6MP A900/D3x sensor. At 2.4MP per cm2 the 5D2 still exhibits a lower density and theoretically better high ISO performance than any current APS-C DSLR. This is very much at odds with the ridiculous claims many on the web are making about full-frame cameras going too high in resolution. In fact, sensor density on the 5D2 is significantly lower than the 10MP Canon 40D, which has a density of 3.1.
The point is that any issues the 5D2 may have with noise are not the result of pixels being "too small". All else being equal the high ISO noise should be at least as good as an 8 to 10MP Canon sensor. Where the new Canon does suffer is in comparison to sensor density of other full-frame sensors. In that metric the Canon 5D2 has about 65% more pixels per cm2 than the Nikon D3/D700 and Canon 5D, and keeping up in high ISO performance with those cameras would be quite a feat. The Canon 5D Mark II is, however, slightly lower in resolution than the Sony A900 and Nikon D3X that feature 24.6MP sensors, although the difference between 21.1 and 24.5/24.6 has to be considered negligible.
Canon 5D Mark II vs. Canon 5D
The Canon 5D is something of an imaging legend, so any examination of the 5D Mark II update has to begin with a comparison to the current 5D. Our comparisons are based on a similar technique used in our review of the Sony A900. Namely, we did same size 150x250 actual pixel crops as one means to compare to the 5D. However, the area covered with the 5D2 21.1MP sensor is about 65% greater than the 12.8MP sensor in the 5D.
Any who have printed large images from a digital file will immediately understand that a larger noisy image when printed smaller often looks like it has much lower noise. This is why we often say the noise level would be good enough for small prints but not for big enlargements. The noise becomes more apparent as the image size is increased. To better compare noise in the same image area, we ran a second set of crops that attempt to cover the same image area. With the metrics of the 5D and 5D Mark II that means a crop of 190x317 pixels for the "same image" comparison, which is then downsized to 150x250
That is the reason for the two crops for the Canon 5D Mark II. The regular view is a pixel-level 150x250 crop, while the 0.6x view is about 65% more pixels adjusted for the same image area as the 12.8MP crop form the 5D. We will leave it to you to decide which is the more relevant of the two crops for the 5D/5D2 comparison, and of course you can also view the original images by clicking on any crop.
All images are captured using a 2-second shutter delay on a tripod in the same location. The manufacturer's 50mm f/1.4 prime lens is used in all cases at an aperture priority setting of f/4.0, some three stops down from their rated speed. All images are processed with the in-camera JPEG processing with high ISO noise reduction set to the low setting. Light is provided by a 100W tungsten bulb, and white balance on all cameras is manually set to tungsten.
The base range of the original 5D is 100 to 1600 ISO with expansion to 3200. Canon appears to have succeeded very well in matching and surpassing the low noise and superb resolution of the original 5D. In fact the 5D2, with a base range of 100 to 3200, is every bit as good in that range as the 5D in its 100-1600 range. Most will have no real problem shooting in the 50-3200 ISO range without much regard to noise. That range is a justified option in the Auto setting on the 5D2.
The 6400 and 12800 ISO options are certainly usable in most circumstances, but the noise increases as you go up from ISO 3200. By 25600 the Canon 5D2 is still producing amazingly sharp images, but noise has reached a point where output should be limited to smaller prints. Further noise reduction processing could improve the image but there is usually a trade in image softness for the reduced noise.
Finally, this is a first look so tests are limited to in-camera JPEG images. We plan to do further comparisons shooting RAW with post-processing as we look more closely at the 5D Mark II.
We will leave to you whether the actual pixels or actual image areas are the better comparison for resolution and noise, but certainly the results are a bit different. Regardless of how you look at it, though, the comparisons to the 5D are truly exciting. Canon seems to have achieved their goal in improved image quality and extended ISO range in comparison to the original 5D. We next compared the 5D2 to the other two full-frame DSLRs in its class.
Canon 5D2 Full Frame vs. Nikon D700/D3 vs. Sony A900
The Canon 5D was the first "popular" priced full-frame DSLR introduced at the $3500 mark for the body only about three years ago. The 5D Mark II just started showing up at dealers on Monday, and the first units began arriving from web etailers on Tuesday, December 2. However, the market is no longer Canon only as both Nikon and Sony have recently introduced models to compete in the $3000 full-frame DSLR market.
The A900 was introduced just a couple of months ago and sells for $3000. It is currently the highest resolution sensor in its class at 24.6MP. The 5D2 has slightly lower resolution at 21/2MP. Nikon also just announced a 24.5MP D3x that will ship later this month, but the new Nikon is in a completely different class with an $8000 price tag.
Nikon introduced the pro-targeted D3 about a year ago with a full-frame $5000 12.1MP sensor. The D3 claim to fame was the widest ISO range ever seen in a DSLR, with a range from ISO 100 to ISO 25600, a 9-stop ISO range. Mid-year 2008 Nikon moved this wide-range full-frame down to the $3000 D700 to compete with the Canon 5D, as well as the coming 5D2 and Sony A900.
Again, all images are captured using a 2-second shutter delay on a tripod in the same location. The manufacturers' 50mm f/1.4 prime lens is used in all cases at an aperture priority setting of f/4.0, some three stops down from their rated speed. All images are processed with the in-camera JPEG processing with high ISO noise reduction set to the low setting. Light is provided by a 100W tungsten bulb, and white balance on all cameras is manually set to tungsten.
The D3 and 5D2 are both no brainers to ISO 3200. I wouldn't hesitate to use either of them for anything up to ISO 3200. 6400 and 12800 are both good with more noise, but still usable for most things. 25600 on either is more for small prints, although I suspect DXO, Noise Ninja, or perhaps other noise reduction programs can do wonders at that speed. They certainly do with RAW processing - the Sony A900 high ISO images that look noisy unless RAW processed. Resolution of the 5D2 is better than the D3/D700, but not quite up to the Sony A900 at regular ISO. I suspect at high ISO with in-camera JPEG the Canon will be equal or better than the Sony.
The Sony seems to be fine up to ISO 1600, but above this noise really starts to intrude. It is not the wide ISO wonder of the 5D2 or the D700, but it is definitely the highest resolution image at lower ISO ratings. The 5D2 is close in resolution, however - much closer than the D700/D3. Recently we have also seen some professionals getting extraordinary high ISO Sony results with RAW post-processing with programs other than Adobe Camera RAW. They claim ACR is quite poor at processing noise in the A900 images and there are much better tools. We are experimenting with their suggestions for A900 RAW post-processing.
Looking Ahead
Looking at the Canon 5D Mark II for a day hardly provides enough data to reach conclusions regarding the 5D2. However, the initial results are truly impressive. Unlike the exaggerated claims for the 50D, the 5D Mark II appears to perform as Canon claims. Resolution is as good or better at the same ISO as the original 5D, but the range is extended somewhere between one and three stops in a sensor with 65% more pixels.
ISO performance appears to match the current ISO champion Nikon D700 up to 3200, and the 5D2 reaches lower to ISO 50. We would not hesitate to use either camera for most anything in this 50/100 to 3200 ISO range, which is saying quite a lot. At ISO 6400 there is some noise in both the 5D2 and D700 images, but they are basically equivalent. At 12800 and 25600 the Nikon D700 appears to have somewhat lower noise, but the image also appears softer, which is often a processing tradeoff for reduced noise. This range of the top two ISO speeds needs to be examined in more detail so we understand what is going on and what tradeoffs can and can't be made with each camera to lower noise.
When you also consider that the 5D Mark II has almost 75% more pixels than the Nikon D700, the above comparisons become much more remarkable. To say that a 21.1MP camera is the equal in low noise to a 12.1MP over most of its range is high praise. To see both reach to ISO 25600 with more similarities in performance than differences is icing on the cake.
Our initial testing still points toward the Sony A900 as the resolution champ at normal ISOs, but there is no doubt that the Canon 5D2 is much better at high ISO performance while achieving resolution nearly as good as the A900. It should also be pointed out that if you want high resolution shooting combined with a brisk continuous shooting speed, the Sony at 5FPS matches the just-announced $8000 Nikon D3x at 5 FPS, That is much better at high ISO than the 5D2 specification of 3.9 FPS. Still, it is almost as if the 5D2 combines what is best about the Sony A900 and the Nikon D700 in a single package.
It is too early to draw final conclusions, but there is much to like in what we have seen so far with the Canon 5D Mark II. The high performance we have seen over a 10-stop ISO range will make the rest of the 5D2 testing journey even more fun. Users looking to upgrade from the original 5D will almost certainly be pleased.