Sony RX10 III Image Quality


Color

Saturation & Hue Accuracy
Slightly below average mean saturation levels and 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. Click for a larger version.

Saturation. The Sony RX10 III pushes some colors like dark blues, dark reds, dark greens and purples, but not as much as most cameras, and it undersaturates lighter greens, yellow and aqua tones moderately. The RX10 III's mean color saturation of 106.3% (6.3% oversaturated) is a bit lower than average these days, but the camera generally produces realistic colors in its images. You can of course tweak saturation to your liking, or choose a non-default color mode. Most consumer digital cameras produce color that's more highly saturated (more intense) than found in the original subjects. This is simply because most people like their color a bit brighter than life.

Skin tones. Here, the Sony RX10 III did reasonably well, producing fairly natural-looking Caucasian skin tones that were a bit on the warm, yellow side with Auto white balance. Manual white balance produced more pleasing, pinkish results. 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 Sony RX10 III shifts aqua toward cyan, cyan toward blue, orange toward yellow, and yellow toward green by small to moderate amounts, but most other shifts are very 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, and we don't find the resulting color objectionable. The RX10 III's handling of yellows and yellow-orange colors is one of its weaknesses: Yellows are rendered closer to a yellow-green, and significantly undersaturated as well, making them look a little dingy. With a mean "delta-C" color error of 6.3 after correction for saturation, overalll hue accuracy is a bit below average. Hue is "what color" the color is.

See full set of test images with explanations
See thumbnails of all test and gallery images

Sensor

Exposure and White Balance

Indoors, incandescent lighting
Auto setting produced warm results, though Incandescent and Manual white balance worked well. Average exposure compensation required.

Auto White Balance
+0.3 EV
Incandescent White Balance
+0.3 EV
Manual White Balance
+0.3 EV

Indoors, under normal incandescent lighting, color balance was warm and reddish with the Auto white balance setting. Results with the Incandescent setting were pretty good, just a touch yellow. The Manual setting was the most accurate, just slightly on the cool side. (Note: The RX10 III also has a Kelvin Temperature White Balance option, however we did not test that mode.) The Sony RX10 III required +0.3 EV exposure compensation which is about average required for this shot. (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.)

Outdoors, daylight
Realistic colors and very good exposure accuracy under harsh lighting.

Manual White Balance,
+0.3 EV
Auto White Balance,
0 EV

Outdoors, the Sony RX10 III performed well for its class. +0.3 EV exposure compensation was required to keep the mannequin's reasonably face bright in our "Sunlit" Portrait shot, which is actually a bit less compensation than most cameras require. Contrast is a quite high as you might expect under such harsh lighting, but despite the bright appearance, the camera did a good job, holding on to the majority of highlights. (And at default exposure, almost no highlights were blown.) Auto color balance performed fairly well in our "Sunlit" Portrait shot, but Manual white balance produced more pleasing, pinkish skin tones. The Far-field shot with Auto white balance has natural-looking color, that's just a touch cool. Default exposure is perhaps just slightly underexposed but with very few highlights blown, and while there are some very deep shadows, detail in the shadows is surprisingly good for its class. Deep shadows do have moderate amounts luminance noise as well as some discoloration, but that's to be expected. Overall, very good performance in harsh lighting, especially considering DRO was off for these shots.

See full set of test images with explanations
See thumbnails of all test and gallery images

Resolution
~2,500 to ~2,600 lines of strong detail.

In-camera JPEG:
Strong detail to
~2,600 lines horizontal
In-camera JPEG:
Strong detail to
~2,500 lines vertical
ACR converted RAW:
Strong detail to
~2,600 lines horizontal
ACR converted RAW:
Strong detail to
~2,500 lines vertical

An in-camera JPEG of our laboratory resolution chart revealed sharp, distinct line patterns up to about 2,600 lines per picture height in the horizontal direction, and to about 2,500 lines per picture height in the vertical direction. Some may argue for more, but lines begin to merge and aliasing artifacts start to interfere at those resolutions. Complete extinction of the pattern didn't occur until about 3,100 to 3,200 lines. Adobe Camera Raw produced similar results though color moiré was much more visible. 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.

See thumbnails of all test images

Sharpness & Detail
Crisp images with good detail, but area-specific noise reduction and sharpening can produce an overprocessed look at higher ISOs.

Good definition of high-contrast
elements here with only minor
sharpening haloes.
Subtle detail: Hair
Noise suppression tends to blur
detail in areas of subtle contrast.

Sharpness. The RX10 III produces fairly crisp, sharp looking images without heavy sharpening halos, but Sony's area-specific noise reduction and sharpening algorithms can lead to some unnatural or slightly crude looking results at higher ISOs. 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 good detail for the class of camera, with moderate levels noise suppression in the darkest areas of the mannequins's hair, and almost no chroma noise. Quite a few individual strands are smudged together in areas of low contrast at base ISO, but performance here is not bad considering the size and resolution of the sensor. 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 noted above the Sony RX10 III produces sharp, crisp and clean images, but fine detail can suffer as a result of aggressive default processing. Compare a base ISO in-camera JPEG to an Adobe Camera Raw conversion below to see what we mean.

Base ISO (100)
Camera JPEG, defaults
RAW via Adobe Camera Raw

In the table above, we compare an in-camera JPEG taken at base ISO using default noise reduction and sharpening (on the left) to a matching RAW file converted with Adobe Camera Raw 9.1 via DNG Converter 9.6 using default noise reduction with strong but tight unsharp masking applied in Photoshop (in this case 300% USM with a radius of 0.5 pixels and a threshold of 0).

As you can see, ACR produced quite a bit of additional detail that isn't present in the JPEG from the camera. Fine detail in the mosaic crop for example is more realistic and refined-looking from the ACR conversion, while the in-camera JPEG is much cleaner and more contrasty, but isn't quite as detailed. This is also true of the fabric crop, where ACR was able to resolve much of the thread pattern in the red-leaf swatch which the camera's JPEG engine presumably treats as noise, and ACR also does a better job reproducing fine detail in the pink fabric. But as is usually the case, more noise can be seen in the RAW conversion particularly in flat areas as shown in the bottle crop, thanks to ACR's light default noise reduction. You can of course apply stronger noise reduction (default ACR NR used here) to arrive at your ideal noise versus detail tradeoff. Color is improved as well, with ACR producing slightly warmer and more accurate colors while removing much of the yellow-to-green shift seen in the JPEG bottle crop. Bottom line: You can do noticeably better than the camera with a good RAW converter, provided you're willing to apply your own noise reduction and sharpening to taste.

ISO & Noise Performance
Very good high ISO performance for its class.

Default High ISO Noise Reduction
ISO 64 ISO 100 ISO 200
ISO 400 ISO 800 ISO 1600
ISO 3200 ISO 6400 ISO 12,800

ISO 64 (extended) and ISO 100 produce clean, very detailed, crisp images containing almost no chroma noise. ISO 200 shows a very minor drop in image quality as noise reduction ramps, but fine detail is still quite good. ISO 400 shows an additional small step down in detail with very good overall image quality, but stronger sharpening attempts to compensate, making luminance noise more visible in some areas. At ISO 800, luminance noise is more noticeable, accentuated by fairly aggressive sharpening, but chroma noise is still welll-controlled and fine detail is still good. ISO 1600 is noticeably softer with more artifacts thanks to stronger noise reduction and more visible luma noise, but chroma noise is still very low. At ISO 3200, fine detail takes a larger hit, and the camera's aggressive processing produces images with a somewhat crystalline look in some areas. Image quality drops off very quickly from here, with ISOs 6400 and 12,800 looking more like impressionistic paintings than photos, with an almost hammered looked to flatter areas. Chroma noise in the form of diffuse purple and/or yellow blotches is also visible in darker midtones and shadows.

Overall, though, high ISO performance remains much better than average for a long-zoom camera and very similar to the RX10 II's, though perhaps just a touch noisier. We're of course pixel-peeping to an extraordinary extent here, since 1:1 images on an LCD screen have little to do with how those same images will appear when printed. See the Print Quality section below for our evaluation of maximum print sizes at each ISO setting.

Note: We used to shoot this series at f/4 because of the relatively low light, but we now shoot it at f/5.6 or f/8 for 1"-type and larger sensors, as lens performance well away from center where we take the above crops is often not optimal at wider apertures. The added depth of field for a scene with this depth is also a better compromise than the potentially slightly sharper but shallower focus depth that a larger aperture would produce.

Extremes: Sunlit, dynamic range and low light tests
Pretty good dynamic range for its class. Good low-light performance, capable of capturing bright images in near darkness.

0 EV +0.3 EV +0.7 EV

Sunlight. The Sony RX10 III did fairly well under the deliberately harsh lighting in the test above. To keep facial tones reasonably bright, +0.3V compensation was required, which led to some clipped highlights in the mannequin's shirt and flowers, though not as many as we usually see from smaller sensors. Some may prefer the +0.7 EV setting for its brighter overall exposure, but we found too many highlights were clipped. Detail is quite good in the shadows at +0.3 EV, though very deep shadows are a little grainy with some discoloration, though noise is fairly fine-grained. Very good results in harsh lighting for a long-zoom camera, but consider using fill flash in situations like the one shown above; and it's better to shoot in the shade when possible.

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.)

Face Detection
Aperture Priority, 0 EV
Face Detection Off
Aperture Priority, 0 EV
Face Detection On
Auto Mode,
0 EV

Face Detection. Like most cameras these days, the Sony RX10 III has the ability to detect faces (up to 8 in a scene), and adjust exposure and focus accordingly. As you can see from the examples above, face detection improved exposure in Aperture Priority at f/5.6, producing a slightly brighter exposure than without it. Full Auto mode performed even better. It selected a larger aperture of f/4 while using a faster shutter speed of 1/125s, and automatically applied DRO (see below) to reduce overall contrast.

Outdoor Portrait DRO Comparison
DRO
Setting:


Auto
(Default)


Off

Level 1

Level 2

Level 3

Level 4

Level 5

Dynamic Range Optimization is Sony's name for their dynamic range enhancement technology. DRO divides the image into small areas, analyzes the range of brightness of each area, and adjusts the camera's image processing parameters accordingly to make the best use of the available dynamic range. Auto DRO is enabled by default on the Sony RX10 III. You can also set the level manually, from 1 ("weak") to 5 ("strong"), or turn it off. As one would expect, DRO is only available for JPEG files.

The above thumbnails and histograms show the effects of the various levels of DRO on our "Sunlit" Portrait shot with no exposure compensation. Mouse over the links on the right to load the associated thumbnail and histogram, and click on the link to visit the full resolution image. As you can see from the thumbnails and associated histograms, increasing DRO progressively boosts shadows and midtones while essentially leaving highlights alone, though boosting shadows does make noise slightly more visible. The Auto setting did a pretty good job overall, and the five manual levels give quite a bit of control over the effect.

Above, you can see the effect of DRO settings on our Far-field shot. The default Auto setting produced a nicely balanced exposure, despite the harsh lighting. A useful feature. (Apologies for the change in framing in the "Off" shot.)

Far-field HDR Comparison

High Dynamic Range. The Sony RX10 III's HDR mode takes three images in rapid succession, one nominally exposed , one underexposed, and one overexposed, then combines them into one high dynamic range JPEG automatically. Lighter areas from the underexposed image are combined in-camera with darker areas from the overexposed image to produce an image with compressed tonal range. The camera then saves a single composite image, as well as the nominally exposed image. The overlaid images are micro-aligned by the camera, but it can only correct for so much movement. If it can't micro-align successfully, an icon indicating HDR capture failed will appear. For best results, the scene should be completely static. There is also a manual mode where you can select from 1 EV ("weak") to 6 EV ("strong") difference in exposures.

Mouse over the links above to load the associated thumbnail, and click on the link to visit the full resolution image. (Again, apologies for the change in framing in the "Off" shot.) As you can see, the Auto setting did a decent job, similar to the 3 EV manual setting. Normally, we'd expect the higher the manual setting, the more highlights are toned-down and shadows opened up, however the 4 EV setting seemed to take a step back. As you can see, though, higher settings can produce flat and unnatural results. And watch out for ghost images and other artifacts from subject movement during the capture sequence, as can be seen in some of the shots above. (It seems the RX10 III attempts to avoid ghosting by including a moving subject from only one of the frames, however this isn't always entirely successful.)

Dynamic Range Analysis (RAW mode)
While we once performed our own dynamic range measurements based on in-camera JPEGs as well as converted RAW images (when the camera was supported by Adobe Camera Raw), we've switched to using DxO Labs' results from their DxOMark website. As technology advanced, the dynamic range of modern high-end cameras in some cases exceeded the range of the Stouffer T4110 density scale that we used for our own measurements. DxO's approach based on RAW data before demosaicing is also more revealing, because it measures the fundamental dynamic range of the sensor, irrespective of whatever processing is applied to JPEGs, or to RAW data by off-the-shelf conversion software.

In the following, we use DxO's "Print" dynamic range results, which are scaled based on camera resolution. As the name suggests, this scaling corresponds to the situation in which you print at a given size, regardless of how many megapixels the camera might have. (In other words, if you've decided to make a 13x19 inch print, that's the size you're printing, whether the camera's resolution is 16 or 300 megapixels.) For the technically-minded, you can find a discussion of the reasoning behind this here on the DxOMark website. Also note that DxO Labs uses a signal-to-noise (SNR) threshold of 1 when defining the lower boundary of acceptable luminance noise in their dynamic range measurements, which corresponds to the "Low Quality" threshold of the Imatest software we used to use for this measurement.

Here, we decided to compare the RX10 III (in orange) to its sibling, the RX10 II (yellow), as well as to its closest competitor, the Canon G3X (red). You can always compare other models on DxOMark.com.

As you can see from the above graph (click for a larger image), the two Sonys have nearly identical dynamic range, which isn't a surprise since they share the same or very similar 1"-type 20-megapixel BSI sensors. Dynamic range for the RX10 III and RX10 II peaks at about 12.6 EV at base ISO, falling to a minimum of 6.8 EV at ISO 12,800.

Dynamic range is similar to the Canon G3X as well, however the Sony's lower base ISO of 100 gives it a slight advantage when it comes to peak dynamic range, with the Canon offering about 12.3 EV at ISO 125. This is a very minor difference, though, which may not be discernible in real-world shots. Minimum dynamic range is also a bit lower, at about 6.2 EV, though measured ISO is actually much closer to 12,800 than the two Sonys. (Also note that the Sonys apply some light noise reduction to raw data at their top two ISOs, as indicated by the smoothed data points.)

Overall, excellent dynamic range performance from the RX10 III for a 1"-type sensor. Click here to visit the DxOMark page for the Sony RX10 III for more of their test results and additional comparisons.

  1 fc
11 lux
1/16 fc
0.67 lux
1/16 fc
No NR
ISO
100

2s, f2.8

30s, f2.8

30s, f2.8
ISO
3200

1/15s, f2.8

1s, f2.8

1s, f2.8
ISO
12800

1/60s, f2.8

1/4s, f2.8

1/4s, f2.8

Low Light. The Sony RX10 III performed well in our low-light tests thanks to its relatively fast lens, capturing bright images at the lowest light level (1/16 foot-candle), even at the lowest native sensitivity setting (ISO 100). As expected for a 1"-type sensor, luma noise is a little high at ISO 3200, but fairly fine-grained, while chroma noise is well-controlled. The RX10 III's highest single-shot ISO of 12,800 is quite grainy with noticeable noise reduction artifacts, but that's no surprise.

Color balance is pretty good with Sony RX10 III's Auto white balance setting, just a touch cool, even at highest ISO and lowest light level. We didn't notice any significant issues with hot pixels, pattern noise or heat blooming. (Some hot pixels can be seen with long exposure noise reduction disabled (rightmost column), but that's normal.)

LL AF: The Sony RX10 III's AF system was able to autofocus unassisted on our standard low-contrast AF target down to -0.9 EV (just above 1/8 foot-candle) which is not bad, but it was able to autofocus down to -3.3 EV with our high-contrast AF target. With the AF assist lamp enabled, the RX10 III could autofocus in complete darkness on both of our low-light AF targets.

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.) For such applications, you may have better luck with a digital SLR camera, but even there, you'll likely need to set the focus manually. For information and reviews on digital SLRs, refer to our SLR review index page.

Output Quality

Print Quality
Excellent 24 x 36 inch prints at base and extended low; a nice 11 x 14 at ISO 1600 and a good 5 x 7 at ISO 6400.

ISO 64/100 prints look terrific at 24 x 36 inches, with crisp fine detail and pleasing color tones throughout. Wall display images are fine at an even larger 30 x 40 inches, but for critical printing at this resolution and sensor size, 24 x 36 inches is the largest suggested size.

ISO 200 images look quite nice at 20 x 30 inches, with virtually no issues to report. 24 x 36 inch prints are fine for less critical applications, but there is a bit of noise in a few flatter areas of our Still Life target apparent at this size.

ISO 400 yields a 20 x 30 inch print that also passes our good grade for prints. There is now a trace of noise seen in a few shadowy areas of our target behind the bottles, but otherwise fine detail and color reproduction are quite good at this size.

ISO 800 is generally where the corner starts to turn for image quality delivered by 1-inch type sensors, as there's simply too much visible noise apparent even in 16 x 20 inch prints to warrant our good seal. A reduction in size to 13 x 19 inch prints does the trick, although there is a bit of contrast detail now lost in our target's tricky red-leaf swatch, but this is quite common.

ISO 1600 delivers a good 11 x 14 inch print. There are a few mild and fairly typical issues with noise in flatter areas of our target and not much contrast detail remaining in our red-leaf swatch, but otherwise the print has good fine detail and full colors present at this size.

ISO 3200 prints are similar at 8 x 10 inches as the 11 x 14 at ISO 1600. There are a few of the same minor issues, but it's still a fairly good print all around for this sensitivity.

ISO 6400 produces a 5 x 7 inch print that just passes our good ranking. Anything larger is simply too noisy and muted to be usable.

ISO 12,800 prints are acceptably good at 4 x 6 inches, although just barely. It's a tad on the muted side, but still manages enough fine detail to pass the grade.

The Sony RX10 III delivers print sizes we'd expect based on the performance of both its predecessor and most 1-inch type sensored cameras in general. You can expect high-quality 24 x 36 inch prints at base and extended low settings, and then a typical size reduction occurring predictably after about ISO 800. Even ISO 12,800 allows for a reasonable 4 x 6 inch print, while good 8 x 10 inch prints are possible all the way up to ISO 3200.

 

The images above were taken from our standardized test shots. For a collection of more pictorial photos, see our Sony Cyber-shot DSC-RX10 III Photo Gallery .

Not sure which camera to buy? Let your eyes be the ultimate judge! Visit our Comparometer(tm) to compare images from the Sony Cyber-shot DSC-RX10 III with those from other cameras you may be considering. The proof is in the pictures, so let your own eyes decide which you like best!



Editor's Picks