Canon EOS 5D Mark III Image Quality
Saturation & Hue Accuracy
Good saturation levels with excellent hue accuracy.
|In the diagram above, the squares show the original color, and the circles show the color that the camera captured. More saturated colors are located toward the periphery of the graph. Hue changes as you travel around the center. Thus, hue-accurate, highly saturated colors appear as lines radiating from the center.|
Skin tones. When white balance is set manually, lighter Caucasian flesh tones shot with the Canon 5D Mark III appear natural, with appropriate saturation levels and accurate color, just slightly on the pinkish side. Darker skin tones show only a small nudge toward orange. Good results here as well. Where oversaturation is most problematic is on Caucasian skin tones, as it's very easy for these "memory colors" to be seen as too bright, too pink, too yellow, etc.
Hue. The Canon 5D Mark III's hue accuracy is excellent, much better than average. There are the usual shifts in cyan toward blue (actually quite small compared to most cameras), red toward orange, and orange toward yellow, but all are fairly minor. (The cyan to blue shift is very common among the digital cameras we test; we think it's a deliberate choice by camera engineers to produce better-looking sky colors.) Average "delta-C" color error is only 3.92 at base ISO (smaller numbers are better), and remains quite low across the ISO range. Hue is "what color" the
The Canon EOS 5D Mark III offers a total of nine saturation settings, four above and four below the default saturation. This covers a very wide range of saturation levels, about as wide a range as you're likely to find photographically useful, apart from special effects that are arguably better achieved in software. As it should, the Canon 5D Mark III's saturation adjustment affects only the saturation, leaving the contrast of the image unaltered. (In some cameras, saturation tends to affect contrast, and vice versa.) The fine steps between settings mean you can program the camera to just the level of saturation you prefer. Well done.
|Saturation Adjustment Examples|
The table above shows alternate settings including the default as well as the two extreme saturation settings. Click on any thumbnail above, then click again to see the full-sized image.
| See full set of test images
See thumbnails of all test and gallery images
Exposure and White Balance
Indoors, incandescent lighting
Auto and Incandescent settings both struggled with household incandescent lighting, though Manual white balance produced a much more neutral image. Slightly higher than average exposure compensation required.
|Auto White Balance
|Incandescent White Balance
|Manual White Balance
|2,600 Kelvin White Balance
Indoors, under incandescent lighting, the Canon EOS 5D Mark III's Auto and Incandescent white balance settings both struggled, resulting in strong reddish and yellow casts respectively. Unfortunately, this is quite common among cameras we've tested, but disappointing nonetheless, especially for a premium model. The Manual setting produced nearly accurate results, though just slightly cool and a touch green overall. The 2,600 Kelvin setting which matches our lights produced similar results to Manual, also a touch cool with a very slight blue tint. The Canon 5D Mark III required a positive exposure compensation of 0.7 EV for this shot, which is slightly higher than the +0.3 EV average among the cameras we've tested. (Our test lighting for this shot is a mixture of 60 and 100 watt household incandescent bulbs, a pretty yellow light source, but a very common one in typical home settings here in the U.S.)
Color and saturation are very good, though a tendency towards slightly cool color balance and slightly high contrast under harsh lighting. About average exposure accuracy.
|Manual White Balance,
|Auto White Balance,
Outdoors, the Canon EOS 5D Mark III tends toward a slightly cool color balance, though overall color is generally very good. We prefer the slightly pinker skin tones from the Manual white balance setting above left, but Auto WB is pretty close. The Canon 5D Mark III performed about average in terms of exposure, requiring the typical amount of positive compensation we're accustomed to seeing among digital cameras. The Canon 5D Mark III's default contrast is a little high, as most users prefer, producing some washed-out highlights and dark shadows under the deliberately harsh lighting of our "Sunlit" portrait test shown above left, though the camera's contrast, Auto Lighting Optimization and Highlight Tone Priority settings do help with this difficult lighting quite a bit. See below for examples of this. The Far-field shot (above right) is also a touch cool, and default exposure a touch dim, though very few highlights are blown. Shadows in the trees and shrubs however are very deep and plugged.
Very high resolution, 2,400 ~ 2,500 lines of strong detail.
|Strong detail to
2,400 lines horizontal
|Strong detail to
2,400 lines vertical
|Strong detail to
2,500 lines horizontal
ACR Converted RAW
|Strong detail to
2,500 lines vertical
ACR Converted RAW
Our laboratory resolution chart shows strong detail with distinct line patterns down to about 2,400 lines per picture height horizontally and to about the same vertically from in-camera JPEGs. Extinction of the pattern occurred just past 3,400 lines horizontally and vertically. Adobe Camera Raw converted .CR2 files show slightly more resolution than the in-camera JPEGs, perhaps 100 lines more in both directions, but complete extinction of the pattern was extended to the limits of the chart in the horizontal direction, and to about 3,800 lines in the vertical direction. While ACR was able to extract more detail, it also produced more color moire than JPEGs from the camera, especially in vertical lines. Use these numbers to compare with other cameras of similar resolution, or use them to see just what higher resolution can mean in terms of potential detail.
Sharpness & Detail
Very good sharpness with loads of detail, though some sharpening artifacts are visible. Some minor detail loss to noise reduction processing even at base ISO.
Sharpness. The Canon EOS 5D Mark III's 22-megapixel sensor captures excellent levels of image detail when coupled with a good lens, though some edge-enhancement artifacts such as sharpening halos are visible around high-contrast edges at default settings, as shown in the crop above left. (The above crop of our "Pine" shot was taken with a Canon EF 50mm f/1.8 II lens which is very sharp when stopped-down to f/8.) Pros shooting JPEGs may want to turn in-camera sharpening down a few clicks, and sharpen in post. Edge enhancement creates the illusion of sharpness by enhancing colors and tones right at the edge of a rapid transition in color or tone.
Detail. The crop above right shows some minor detail loss due to noise suppression at base ISO. You can see some smudging of lower contrast and darker strands of hair, but performance here is still very good given the resolution. (The crop above of the hair taken with our very sharp Sigma 70mm f/2.8 reference lens.) Noise-suppression systems in digital cameras tend to flatten-out detail in areas of subtle contrast. The effects can often be seen in shots of human hair, where the individual strands are lost and an almost "watercolor" look appears.
Raw vs In-Camera JPEGs
As mentioned above, the Canon 5D Mark III does an excellent job of capturing sharp, detailed JPEGs when coupled with a sharp lens, but as is usually the case, slightly more detail can be preserved while producing fewer sharpening artifacts by carefully processing its raw files.
In the table above, mousing over a link at the bottom will load the corresponding crop in the area above, and clicking on the link will load the full resolution file. Examples include in-camera Fine JPEG, raw file processed through Canon's Digital Photo Professional (DPP) software using default settings, and finally, a raw file processed through Adobe Camera Raw (ACR) version 6.7, then sharpened in Photoshop using 250% unsharp masking with an 0.3 pixel radius.
Canon's DPP software produced images similar to in-camera JPEGs, perhaps with a touch more detail but also showing obvious sharpening halos. Raw images processed through ACR and then moderately sharpened in Photoshop contain improved detail compared to in-camera and DPP-processed images, with less obvious sharpening halos. Users wanting to extract the most detail from their 5D Mark III will definitely want to shoot raw and use a good raw converter.
ISO & Noise Performance
Very good detail versus noise to ISO 1,600, with good performance up ISO 6,400.
Default High ISO Noise Reduction
|ISO 50||ISO 100||ISO 200|
|ISO 400||ISO 800||ISO 1,600|
|ISO 3,200||ISO 6,400||ISO 12,800|
|ISO 25,600||ISO 51,200||ISO 102,400|
The Canon EOS 5D Mark III's high ISO performance is excellent up to ISO 1,600 in the relatively dim indoor lighting used here. Sure, there's a slight loss of detail to noise reduction already at base ISO, and a gradual loss of fine detail as ISO increases, but remaining detail is still very good to ISO 1,600. At ISO 3,200 we begin to see some moderate detail loss as stronger noise reduction blurs subtly contrasting detail, as well a touch of chroma noise in darker areas, but results are still quite good. At ISO 6,400 noise grain becomes coarser, blurring stronger, and chroma noise more apparent, but there's still a lot of detail left to work with. Noise and the effects of noise reduction really become apparent at ISO 12,800, with much stronger blurring and more noticeable grain. At ISO 25,600 and above, fine detail really deteriorates quickly, with obvious chroma blotching and high pixel noise exacerbated by relatively strong sharpening, especially at ISO 102,400. Saturation also drops off above ISO 12,800, probably in an attempt to mitigate chroma noise.
Overall, though, a very good performance. See the Print Quality section below for our evaluation of maximum print sizes at each ISO setting.
A note about focus for this shot: We shoot this image at f/4, using one of three very sharp reference lenses (70mm Sigma f/2.8 macro for most cameras, 60mm f/2.8 Nikkor macro for Nikon bodies without a drive motor, and Olympus Zuiko 50mm f/2.0 for Four Thirds and Micro Four Thirds bodies). To insure that the hair detail we use for making critical judgements about camera noise processing and detail rendering is in sharp focus at the relatively wide aperture we're shooting at, the focus target at the center of the scene is on a movable stand. This lets us compensate for front- or back-focus by different camera bodies, even those that lack micro-focus adjustments. This does mean, though, that the focus target itself may appear soft or slightly out of focus for bodies that front- or back-focused with the reference lens. If you click to view the full-size image for one of these shots and notice that the focus target is fuzzy, you don't need to email and tell us about it; we already know it. :-) The focus target position will simply have been adjusted to insure that the rest of the scene is focused properly.
Extremes: Sunlit, dynamic range and low light tests
Very high resolution with strong detail, though somewhat limited dynamic range. Highlight Tone Priority, Auto Lighting Optimization, and contrast adjustment options do a good job of dealing with harsh lighting. Excellent low-light performance.
|+0.3 EV||+0.7 EV||+1.0 EV|
The Canon EOS 5D Mark III produced moderately high contrast with some washed-out highlights and deep shadows under the deliberately harsh lighting of the test above. The mannequin's face was too dim at the default 0 EV and +0.3 EV exposure settings, and far too many highlights were blown at +1.0 EV, so we preferred the image with +0.7 EV exposure compensation overall. This still resulted in some clipped highlights in the shirt and flowers, though, an indication of somewhat limited dynamic range compared to recent APS-C and full frame competitors. Noise is fairly low in the shadows in all but the deepest tones, though detail is a bit soft indicating some active noise reduction is taking place in the shadows.
Because digital cameras are more like slide film than negative film (in that they tend to have a more limited tonal range), we test them in the harshest situations to see how they handle scenes with bright highlights and dark shadows, as well as what kind of sensitivity they have in low light. The shot above is designed to mimic the very harsh, contrasty effect of direct noonday sunlight, a very tough challenge for most digital cameras. (You can read details of this test here. In actual shooting conditions, be sure to use fill flash in situations like the one shown here; it's better to shoot in open shade whenever possible.)
We really like it when a camera gives us the ability to adjust contrast and saturation to our liking. It's even better when those adjustments cover a useful range, in steps small enough to allow for precise tweaks. As was the case with its saturation adjustment, the Canon 5D Mark III's contrast setting meets both challenges very well.
|Contrast set to lowest,
|Contrast set to lowest,
At its lowest contrast setting, the Canon 5D Mark III did a good job of preserving highlight detail by allowing a lower exposure for our "Sunlit" portrait, while maintaining fairly natural-looking skin tones, and bringing out shadow detail. It also helped with the dark shadows in our Far-field shot. Overall, very good results here, with the contrast adjustment working on both highlights and shadows.
|Contrast Adjustment Examples|
The Canon 5D Mark III's contrast-adjustment control offers a very broad range of control in usefully fine gradations, and does a good job of adjusting contrast without affecting color saturation in the process. (As noted earlier regarding saturation adjustment, something that not all cameras manage to do.)
Highlight Tone Priority
The Canon 5D Mark III's Highlight Tone Priority (HTP) option did an excellent job of preserving highlight detail, as shown below. (Mouse over the Off and On links to load the corresponding thumbnail, histogram and crops.)
Highlight Tone Priority, +0.7 EV
to reveal noise.)
Both shots above were captured at +0.7 EV (the default exposure had very few hot highlights to correct), the only difference being that HTP was enabled for the second shot which necessarily increases the ISO to 200; part of how HTP works. The result is evident in the histograms and crops above, clearly showing the superior highlight preservation when HTP is enabled, while shadows were only minimally impacted. As you can see from the shadow crops, an increase in noise is however the price you pay when ISO is boosted from 100 to 200. (Note that levels were highly boosted in the shadow crops to reveal the increase in noise.) Except in the very deepest shadows, though, overall noise is low enough at ISO 200 that this is really a negligible tradeoff.
Auto Lighting Optimization
Like other recent EOS models, the Canon 5D Mark III offers three selectable levels of Auto Lighting Optimization (ALO), plus Off. In Auto exposure mode, ALO is automatically enabled at the Standard level. All four shots below were taken with the same default exposure settings.
Mouse over the links on the left to load the associated thumbnail and histogram, and click on the links to load full resolution images. As you can see, ALO has the effect of shifting shadows and mid-tones in the histograms to the right, brightening shadows and indeed most of the image without clipping too many additional highlights, making it a useful feature for high contrast shots like this. ISO is not boosted for ALO so increased noise is not an issue, though it may be slightly more visible in shadows that have been boosted significantly.
The Canon 5D Mark III has a new High Dynamic Range (HDR) capture mode where the camera takes three images (underexposed, normal, and overexposed) in quick succession and combines them in-camera into one image. If performed properly, this method should result in much higher dynamic range, without the additional noise penalty that comes with boosting sensitivity when using the HTP option. (In fact, it can reduce shadow noise by combining shadows from the overexposed shot.) There are four HDR settings available: Auto, ±1 EV, ±2 EV, and ±3 EV. There are also five Effect modes: Natural, Art standard, Art vivid, Art bold, and Art embossed.
Far-field HDR Examples, 0 EV
The above images were shot using the default Natural mode. Rollover the links to compare the available exposure settings, and click on the links to access full resolution images.
As you can see, dynamic range is improved, however the resulting HDR images are still a bit dim, and are also a little soft compared to a single exposure, likely because we forgot to disable Auto Align as recommended by Canon when shooting on a tripod. (Auto Align appears to crop the image, then interpolate the image back to full size.) Also, watch out for ghosting and other artifacts such as partial elements when the scene is not entirely static (look in the lower-left corners of the full-resolution ±2 EV and Auto shots to see what we mean).
Still Life HDR Art Effects
While not a high dynamic range scene, above are samples of the various HDR Art Effects available taken of our Still Life shot.
Just like most point & shoot cameras these days, the Canon EOS 5D Mark III has the ability to detect faces in Live View mode, and adjust exposure and focus accordingly.
|Face Detection, 0 EV|
Face Detection: Off
Face Detection: On
As you can see from the examples above, it works, as the center image with face detection enabled is better exposed for the face (though still slightly dim) than the left image where face detection was not employed, though the camera did boost ISO to 200. Scene Intelligent Auto (right image) did even better without Live View mode, though, producing the best overall exposure. Auto mode also selected a much wider aperture (f/4), making the depth of field shallower to help isolate the subject from the background, though the eyes are out of focus. (When shooting portraits at wide apertures, it's best to manually place the focus point over an eye.)
A key parameter in a digital camera is its Dynamic Range, the range of brightness that can be faithfully recorded. At the upper end of the tonal scale, dynamic range is dictated by the point at which the RGB data "saturates" at values of 255, 255, 255. At the lower end of the tonal scale, dynamic range is determined by the point at which there ceases to be any useful difference between adjacent tonal steps. Note the use of the qualifier "useful" in there: While it's tempting to evaluate dynamic range as the maximum number of tonal steps that can be discerned at all, that measure of dynamic range has very little relevance to real-world photography. What we care about as photographers is how much detail we can pull out of the shadows before image noise becomes too objectionable. This, of course, is a very subjective matter, and will vary with the application and even the subject matter in question. (Noise will be much more visible in subjects with large areas of flat tints and subtle shading than it would in subjects with strong, highly contrasting surface texture.)
What makes most sense then, is to specify useful dynamic range in terms of the point at which image noise reaches some agreed-upon threshold. To this end, Imatest computes a number of different dynamic range measurements, based on a variety of image noise thresholds. The noise thresholds are specified in terms of f-stops of equivalent luminance variation in the final image file, and dynamic range is computed for noise thresholds of 1.0 (low image quality), 0.5 (medium image quality), 0.25 (medium-high image quality) and 0.1 (high image quality). For most photographers and most applications, the noise thresholds of 0.5 and 0.25 f-stops are probably the most relevant to the production of acceptable-quality finished images, but many noise-sensitive shooters will insist on the 0.1 f-stop limit for their most critical work.
JPEG. The graph at right (click for a larger version) was generated using Imatest's dynamic range analysis for an in-camera Canon 5D Mark III JPEG file with a nominally-exposed density step target (Stouffer 4110). At default camera settings and base ISO, the graph shows 11.8 f-stops of total dynamic range, with 7.8 f-stops at the "High" Quality level. These are pretty good numbers numbers for JPEGs, though the high quality score was quite a bit lower than the best full frame models. Compared to the Nikon D800/D800E, the 5D Mark III scored significantly lower at the High Quality level (7.8 vs 9.11 f-stops), but better in total dynamic range (11.8 vs 10.7 f-stops), however much of that difference is due to differences in default tone curves, etc. Note though that this measurement has a margin of error of about 1/3 f-stop, so differences of less than 0.33 can be ignored.
Raw. The graph at right is from the same Stouffer 4110 stepchart image captured as a raw (.CR2) file, processed with Adobe Camera Raw using the Auto setting then tweaking from there. The Canon 5D Mark III's raw file scored 0.7 f-stop more in total dynamic range (12.5 vs 11.8 f-stops), but the score at the highest quality level increased only 0.07 f-stops from 7.8 to 7.87, which is an insignificant improvement and below average these days. This is due to the somewhat high noise in the 5D Mark III's darker shadows, as indicated when comparing the bottom-left Pixel Noise plot in the Imatest chart. The Nikon D800/D800E for instance managed significantly better scores of 9.98 f-stops at the highest quality level, and 13.3 f-stops total dynamic range. See below for a comparison of shadow noise between the Canon 5D Mark III and Nikon D800/D800E.
|Canon 5D Mark III||Nikon D800/D800E|
The primary reason the Canon 5D Mark III doesn't score as well as the D800/D800E in dynamic range tests is shadow noise. As mentioned above, shadow noise is much higher from the Canon 5D Mark III, which limits usable dynamic range. Above, we have 100% crops from steps 32 to 34 of our Stouffer stepchart raw shots at ISO 100, converted with Adobe Camera Raw 6.7 with noise reduction turned all the way down. The "Unboosted" crops have had no levels adjustment, while "Levels Boosted" have had Photoshop's mid-tone level slider adjusted equally from 1.0 to 4.0 for both cameras, to expose the noise lurking in the darker tones. (The chart has a total of 41 steps with step 41 being the darkest, so these aren't even the darkest steps).
Roll-over the "Levels Boosted" text to see just how much cleaner the Nikon D800/D800E is compared to the Canon 5D Mark III in deep shadows. (Keep in mind you'd likely never boost exposure this much in real-world photos. This was done here just to illustrate the difference in dark shadow noise between the two cameras in similar conditions.)
Low Light. The Canon 5D Mark III performed very well on the low-light test, capturing bright images at the lowest light level (1/16 foot-candle), down to the base ISO of 100, though ISO 50 shots are a little dim at the lowest light levels (limited by the longest non-bulb mode shutter speed of 30 seconds).
As expected, noise increases as ISO goes up and light levels go down, but noise remains fairly well-controlled to ISO 12,800. There were a few hot pixels at low ISOs and low light levels with noise reduction turned down, as well as some bright pixels at higher ISOs, but nothing unusual. Some faint banding can be seen in darker areas at very high ISOs, especially with noise reduction turned down, but that's not unusual either, and shouldn't be a problem for properly exposed subjects. Some heat-blooming can also be seen in the lower-right corner at very high ISOs, but that's also not unusual at such high sensitivities.
Color balance was pretty good, just a touch cool with Canon 5D Mark III's Auto white balance setting even at high ISOs, though at lower light levels white balance shifted slightly towards magenta.
When using the optical viewfinder and phase-detect AF, the Canon 5D Mark III's autofocus system was able to focus on the subject down to below the 1/16 foot-candle light level unassisted with an f/2.8 lens. Note that the Canon 5D Mark III does not have an AF assist lamp (it uses the one provided on most flash units). In Live View mode using contrast-detect AF, the Canon 5D Mark III was also able to focus down to below 1/16 foot-candle unassisted, which is excellent.
As always, keep in mind that the longer shutter speeds here demand the use of a tripod to prevent any blurring from camera movement. (A useful trick is to just prop the camera on a convenient surface, and use its self-timer to release the shutter. This avoids any jiggling from your finger pressing the shutter button, and can work quite well when you don't have a tripod handy.)
How bright is this? The one foot-candle light level that this test begins at roughly corresponds to the brightness of typical city street-lighting at night. Cameras performing well at that level should be able to snap good-looking photos of street-lit scenes.
NOTE: This low light test is conducted with a stationary subject, and the camera mounted on a sturdy tripod. Most digital cameras will fail miserably when faced with a moving subject in dim lighting. (For example, a child's ballet recital or a holiday pageant in a gymnasium.) Digital SLRs like the Canon 5D Mark III do much better than point & shoots, but you still shouldn't expect a quick autofocus lock with moving subjects.
Good quality 30 x 40 inch prints at ISO 100/200; ISO 3,200 shots still looked good at 16 x 20; and ISO 51,200 made a good 4 x 6.
ISO 200 shots also looked great at 30 x 40 inches.
ISO 400 shots look spectacular at 24 x 36 inches.
ISO 800 images started to show a very slight indication of softness due to noise suppression, but all major elements still looked quite good at 24 x 36 inches.
ISO 1,600 images were a little softer in the red leaf swatch, but other than that, these images still looked good at 20 x 30 inches.
ISO 3,200 shots started to show a little more loss of detail in reds and other low-contrast elements, and shadows started to show a little more luminance noise at 20 x 30, but prints looked better at 16 x 20 inches.
ISO 6,400 images lost more detail in the red swatch, and luminance noise got a little darker and more prominent. 16 x 20 inch prints are still usable, but we preferred 13 x 19 inch prints.
ISO 12,800 images had shadow noise that was reasonably controlled at this size, and fine detail was good. For rendering of reds, we preferred the 8 x 10 inch print here.
ISO 25,600 prints a nice 5 x 7.
ISO 51,200 shots were usable at 5 x 7, but really better at 4 x 6.
ISO 102,400 shots were just a little too fuzzy in the shadows to be called usable at any size, and is best avoided altogether.
Overall, the Canon 5D Mark III did quite well in our Print Quality test, with all but the highest ISO setting able to produce a good quality print at sizes people commonly use (and some uncommonly large ones, too). Even ISO 51,200 made a good 4x6-inch print!
Testing hundreds of digital cameras, we've found that you can only tell just so much about a camera's image quality by viewing its images on-screen. Ultimately, there's no substitute for printing a lot of images and examining them closely. For this reason, we now routinely print sample images from the cameras we test on our Canon Pro9000 Mark II studio printer, and on the Canon Pixma MP610 here in the office. (See the Canon Pixma Pro9000 Mark II review for details on that model.)