Image Quality

The Pixel 2 is capable of seriously good image quality, out of camera.

ISO53 | 1/1500 sec| F1.8

Assessing the image of quality of a device that heavily utilizes computational photography, like the Pixel 2, is a bit different than it is for a traditional camera. The camera's exposure and processing decisions vary from scene-to-scene, so assessing image quality from one static (studio) scene only tells you so much. As such, this image quality page is going to be a mix of both studio and real-world analysis, relying more on the latter.

Studio scene

Below, we framed the Pixel 2 up and pointed it at our studio scene using the default camera app, allowing it to choose its exposure values for daylight (10 EV) and low light (3 EV). While we could have used Lightroom CC to shoot JPEGs with exposure values dialed in, we opted against this because the result would be files that don't make use of HDR+ and therefore are not representative of the best of what the Pixel 2 can offer.

Daylight
ISO 50 | 1/209 sec | F1.8

Low light
ISO 411 | 1/30 sec | F1.8

Our daylight sample shows well-judged color and excellent detail throughout much of the frame. There is some noticeable distortion throughout the image even after all the auto distortion correction, but you'll only really ever notice this if you're shooting architecture. Detail is sharp corner to corner, indicating an excellent lens. Skintones are a bit under-saturated and green-shifted compared to the actual prints on our studio scene (more on this later), but overall these are impressive results (see how the iPhone X performed on our studio scene).

Considering how much smartphone cameras have historically struggled in low light shooting, these results have us very impressed

Low light JPEG image quality is also excellent, thanks to the fact that the camera is essentially behaving like one with a sensor nine times its size. Noise is well-handled, without blurring away fine detail, even in the very dark areas of the scene. Considering how much smartphone cameras have historically struggled in low light shooting, these results have us very impressed. Plus, the fact that the Pixel 2 defaulted to 1/30 sec in low light means this level of image quality can be achieved when shooting hand-held, or even people in low light.

The iPhone X needs to drop its shutter speed to 1/4s to achieve similar results in our low light scene, a shutter speed largely impractical for many types of photography (see how the iPhone X does with our low light scene when shot hand-held with a bit of movement added to our scene, which forces it to use the same 1/30s the Pixel 2 uses).

Here's a base ISO DNG file for reference. Click through to see a daylight, base ISO shot that's been run through our standard studio scene process.

Daylight image quality

Punchy but not overdone colors and good detail are par for the course - shot on Pixel 2 XL. Photo: Rishi Sanyal

ISO 60 | 1/1000 sec| F1.8

Google has clearly gone to great lengths to ensure that photos taken with their flagship phone are ready to be shared the moment they are taken, without a need for post-processing. And for the most part our real world experience both mimics that philosophy and confirms what we saw in the studio scene above. HDR+ JPEG's shot in daylight offer well-judged saturation, impressive detail and excellent dynamic range.

The default auto white balance mode tends to settle on color temperatures that are generally accurate, but this can often overcorrect color casts, causing warm scenes to take on a cool appearance. Skin tones are, for the most part, well-judged when shooting in daylight, but some will consider them rather subdued and lacking in saturation. We noted that the more tone to a subject's skin, the less saturated and green-shifted it appeared relative to real life.

Portrait mode handled the complex lighting in this scene extremely well, thanks to HDR+. Skintone of our subject here appear desaturated and green-shifted compared to reality or competitors' renderings though. Photo: Rishi Sanyal

ISO 68 | 1/120 sec| F1.8

Interestingly, we know the Pixel 2 has the machine learning capability to identify human faces, yet it does not tone down the 'clarity' (local contrast) of faces in high contrast scenes. HDR multi-capture of wide dynamic range scenes often requires aggressive tonemapping (lifting of shadows, darkening of highlights) that can make faces look, well, crunchy. As if you had increased the 'clarity' slider in ACR. It would be helpful if Google decreased local contrast of faces after such aggressive tonemaps. For example, have a look at our subject's face in this Pixel 2 XL image vs this similar iPhone 8 Plus image. Or the image of father and daughter inside the ferry a few images below.

For more daylight (and low light) samples, head to our Pixel 2 sample gallery.

Low light image quality

Low light is the Achilles' heel of smartphones, but the Pixel 2 comes the closest of any phone I've used to offering a semblance of good low light image quality. Noise is impressively controlled (thank you HDR+) and color looks accurate, not washed out.

The low light image quality from the Pixel 2 left me impressed.

ISO 268 | 1/60 sec| F1.8

Chosen shutter speeds also appear to be context sensitive. In the image above, despite having OIS and being able to use slower speeds, the camera opted for a 1/60 sec shutter, fast enough to mostly freeze any subject movement. Even in scenes dimmer than the one above, the Pixel 2 rarely drops below 1/30 sec. The camera is confident that the 9-frame image averaging can make up for shorter exposure time, giving the camera a fighting chance at freezing human or moving subjects even in low light.

The Pixel 2 rarely misses focus, thanks to Dual Pixel AF and HDR+

In comparison, an iPhone X can blur your subject by using too slow a shutter speed or, if it does detect motion and use 1/30s, often results in a smeary, watercolor painting-like image due to the lack of as aggressive image averaging. And that's assuming the iPhone doesn't miss focus, which the Pixel 2 rarely does thanks to Dual Pixel AF that also benefits from HDR+. There were times though when the Pixel 2 focused on the background because of its heavily center-weighted focus priority when it momentarily loses a detected face.

While we appreciate Google's use of shorter exposures to freeze motion, there are instances - like static nightscapes - where the the camera would benefit from dropping to even slower shutter speeds if no motion were detected. Its OIS can certainly handle it. And the fact that it doesn't allows the iPhone X to catch up in our low light studio scene and in static nighttime comparisons like this one by The Verge.

JPEG processed in Google Photos

Out of camera JPEG

The candid above was only possible because of the nearly instantaneous snap-to-focus of Dual Pixel AF as the pair turned around briefly to face the camera (the moment quickly vanished). But remember that undersaturated, greenish skintone issue we mentioned earlier? Portraits shot indoors in low light are even more prone to suffer from this. The first image in the above roll-over is one we've processed to taste in Google Photos to have more pleasing skin tones, while the second is an out of camera JPEG.

Portrait mode

Note: when using portrait mode there is a ~1.5x digital crop factor resulting in a 42mm field of view. If you're the kind of person partial to wide angle portraits, I'm sorry. This crop also costs some image resolution. Also note the somewhat unflattering increased local contrast on the faces of our subjects, due to the aggressive tonemap of a wide dynamic range scene. Photo by Rishi Sanyal

ISO 52 | 1/120 sec| F1.8

Shallow depth of field is another area cellphone cameras have never been able to compete with their dedicated camera cousins. But now thanks to portrait mode, that traditional camera advantage is further chipped away.

We already discussed in brief how portrait mode works on the Features page. And if that doesn't leave you satisfied, you can read more about the cool/geeky science behind it on Google's blog or in our 'Pixel 2 vs iPhone 8 Plus Portrait Mode' shootout.

Remember when phones couldn't take photos like this? Shot on Pixel 2 XL. Photo by Rishi Sanyal

ISO 382 | 1/60 sec| F1.8

Overall, we found portrait mode handled most subjects and situations remarkably well. It can occasionally be over-aggressive with the blur, especially around glasses and unexpected objects on human subjects. We were particularly impressed though at how well it deals with even incredibly complex patterns, like hair without making a head look 'cut out'. It's not always perfect, with a few errors here and there, but for the most part the Pixel 2's Portrait mode is the most convincing we've seen. And as time goes on, we expect improvements to be made in this area, thanks to machine learning.

However because portrait mode is tuned to human subjects, you'll occasionally end up with images like the one above, where both subjects on very different planes are in focus, but everything else is not. While this looks unusual to the well-tuned photographer's eye, most people will just see a nice photo with two faces in focus.

Portrait mode isn't just for human subjects. It also works on paint-splattered sawhorses.

ISO 54 | 1/136 sec| F1.8

Of course, portrait mode works on both human and non human subjects. In both cases, the camera creates a depth map and then applies a progressive blur to the image. This helps mimic actual depth of field blur and ensures out of focus areas closest too the subject aren't too jarringly blurred.

Selfies in portrait mode

A selfie shot in portrait mode at Seattle's infamous gum wall.

ISO 86 | 1/24 sec| F2.4

The front-facing camera can also be used in portrait mode with pleasing results. Separation is a tad sloppier than when using the rear camera, though. This is because subjects are separated from the background based on 'segmentation' only, not depth mapping. There is also no progressive blur.

Portrait mode using rear camera

Portrait mode using front camera

The above roll-over demonstrates the difference in blur between the front and rear-facing cameras. Notice how much harsher the area between me and the start of the blur is when using the front camera, and how the rear camera preserves the eyeglasses.

Oddly, in many situations we found the colors from the front and rear cameras to be different, with the front camera tending to offer a more desaturated, magenta-shifted image.

Raw capture

A Raw conversion, shot using the Lightroom CC app and processed on desktop in Adobe Camera Raw.

ISO 50 | 1/190 sec| F1.8

Using third party apps it is possible to shoot DNGs with the Pixel 2. I'm an Adobe CC subscriber and used the LR CC app to unlock Raw capture. Also unlike other smartphones - that make offloading Raw files the equivalent of knowing the average airspeed velocity of an unladen swallow - all DNGs shot on the Pixel 2 automatically upload to Google Photo just like JPEG files.

Until third-party Raw apps can take advantage of HDR+, we'd suggest you stick to the default camera app.

In the roll-over bellow, the first image is the out of camera JPEG, shot using HDR+. The second image is one I converted in Adobe Camera Raw. The results of the Raw conversion are OK. But the amount of time it takes to process hardly seems worth it. Not to mention, DNG files, even those shot in daylight, are very noisy. This is to be expected given the size of the sensor, and its this sensor limitation that HDR+ is designed to work around. Until third-party Raw apps can take advantage of HDR+, we'd suggest you stick to the default camera app.

Out of camera JPEG

DNG converted in ACR

That said, given that the Pixel 2's Visual Core opens up HDR+ to third party apps, keep an eye out for new innovative software that uses and iterates upon the computational approaches Google has developed.