We’re not even out of tax season yet, but it’s already clear that the two biggest trends in smartphone imaging in 2023 are big sensors and big resolution. Manufacturers like Vivo and Xiaomi are putting big 1-inch-type sensors in their phones, leaning into the light-gathering and creative benefits of a larger imager. In the other corner, Samsung is squeezing as many pixels as it possibly can onto a smaller sensor, betting on pixel binning and multi-frame processing to make up the shortfall in performance.
So which approach is right? If we were dealing with traditional cameras, it would be more clear: all things being equal, a bigger sensor wins every time. But all things are not equal when it comes to smartphone cameras. They capture multiple frames to boost dynamic range, combine pixels in response to lighting conditions, and automatically combine exposures in near darkness. The days of simply opening and closing a shutter are long gone.
I spent a week shooting with the 200-megapixel Samsung Galaxy S23 Ultra and the 1-inch-type Vivo X90 Pro, and the results were surprising. Sure, a big sensor wins on certain things. Its RAW files are less noisy in low light than the S23 Ultra’s. There’s also shallower depth of field to play with outside of portrait mode. But when I let both cameras make their own decisions about exposure settings and processing — and introduced extremely challenging subjects like a moving toddler — the advantages of a bigger sensor shrank dramatically.
The sensor is just one part of the story
Then there’s something obvious that I was reminded of as I compared images: the sensor is just one part of the story. There’s the lens, the hardware processing the image data, and the software analyzing the scene in front of the camera. A sensor’s theoretical performance is one thing; how all those parts play together with real-world subjects is another.
First, some housekeeping. The Vivo X90 Pro and Samsung Galaxy S23 Ultra both have a lot of rear cameras, but for the purposes of this article, we’re focusing on each phone’s main camera. That’s a 200-megapixel 1/1.3-inch-type sensor with an f/1.7 lens on the S23 Ultra and a 50.3-megapixel 1-inch-type sensor and an f/1.8 lens on the Vivo X90 Pro. Both have optical image stabilization, but only the Vivo bears prominent Zeiss branding on the lens.
Say it with me: a 1-inch-type sensor doesn’t literally measure an inch. The way we talk about image sensor formats has something to do with old TV camera tubes. It’s weird. That said, this 1-inch sensor is still very big by smartphone camera standards. It offers about 1.8x more surface area to collect light than the 1/1.3-inch type sensor on the S23 Ultra.
The more photons, the better
To start, let’s look at a couple of places where the big sensor has the clear advantage. If two cameras are set up with the same exposure, capturing the same subject with the same framing, you’d expect a cleaner image from the bigger sensor. There’s just more surface area to soak up photons. Richard Butler at DPReview explains it way better, but basically, it’s just physics. The more photons, the better.
To help play catch-up, smaller sensors use a technique called pixel binning to group the data from individual pixels together so they act as one big pixel in low-light conditions. In bright light, they act as individual pixels — since there’s enough light to go around, smaller pixels work fine. It’s kind of a have-your-cake-and-eat-it scenario: you get better low-light performance and the flexibility of a higher-resolution sensor. It’s a sensor technology that’s been in use for years, and even Apple got on board with the iPhone 14 Pro, so you know it’s a trend that’s sticking around.
The pixels on the X90 Pro’s Sony IMX989 sensor are physically bigger, and bigger pixels collect more light. But even when you take pixel binning into account, the X90 Pro has the advantage over Samsung here, too. In low light, the S23 Ultra’s sensor is binning 16 tiny 0.6μm pixels to create much larger 2.4μm pixels. The Vivo X90 Pro bins four 1.6μm pixels to make 3.2μm pixels. Both result in a 12-megapixel image (well, 12.5-megapixels in Vivo’s case — Samsung downsizes to an even 12 megapixels).
So yes, the Vivo X90 Pro produces cleaner RAW image files in low light than the Samsung S23 Ultra. Here’s what that looks like at ISO 1600 — there’s clearly more noise in the S23 Ultra’s image.
But! Those are single-frame RAW files, and we haven’t brought computational photography into the mix yet. Putting both cameras in Night Mode evens out the difference quite a bit. If you look closely, there’s still some color noise clinging to areas of fine detail in the S23 Ultra’s image that aren’t visible in the X90 Pro’s. But downsized to the resolution you’d see on a computer or phone screen, they look about the same.
Night mode is all well and good if your subject isn’t moving, but the real test for a smartphone camera is with a moving subject in crappy lighting — kind of a white whale for mobile imaging. That’s where the bigger sensor format could make a real difference.
A faster shutter speed will freeze a moving subject so you get a sharp shot, but that requires a higher ISO so that your overall image is bright enough. Since the bigger sensor produces less noise in high ISO images, the hope is that it would be able to use faster shutter speeds in low light and come up with decent images of moving subjects. Computational photography makes it a little more complicated than that since shutter speed and ISO are kind of fluid concepts when you’re combining multiple frames, but the basic principle of “more light, better photo” broadly applies here.
Sadly, we have not found our white whale just yet. The Vivo X90 Pro and S23 Ultra both very much like to stay at 1/120sec shutter speed when I’m taking pictures of my toddler in the standard camera mode. That’s fine if he’s not moving much, but it’s not fast enough to keep up with him when he’s running from room to room (his default speed).
The bigger sensor’s better baseline noise performance doesn’t seem to help much here, either. The images from the X90 Pro don’t look any more detailed than the S23 Ultra’s — and often, the S23’s images look better, even at a higher reported ISO.
In the scene below, both cameras reported an ISO of about 320 and a shutter speed of 1/120sec. The S23 Ultra image looks flat-out better.
More pixels — or, at least, a more aggressive application of computational photography — looks like it comes out slightly ahead here.
But wait, there’s lenses!
The other other thing about a bigger sensor is that there’s a bigger lens attached. It’s kind of a requirement. It’s also part of the whole better-in-low-light equation since a bigger lens will usually have a bigger aperture. That’s the case with the X90 Pro. On paper, the S23 Ultra seems to have the wider aperture at f/1.7. As far as phone cameras go, a bigger aperture is generally better — it lets in more light and gives a little more depth of field control. But the f/1.8 aperture on the X90 Pro is actually significantly bigger compared to the S23 Ultra’s because the overall lens is larger.
If you factor in sensor format and compare apples to apples, the S23 Ultra’s f/1.7 aperture actually behaves about like an f/6.5 full-frame lens — at least as far as light-gathering and depth of field are concerned. The X90 Pro’s f/1.8 lens is equivalent to f/4.9 on full-frame, which lets in more total light and provides shallower depth of field. If you want to argue about the concept of equivalency, please report to the DPReview forums and tell ‘em I sent you.
That bigger aperture is good news for the X90 Pro, but the bad news is that despite the little blue Zeiss badge on the lens, it ain’t great. It’s sharp at the center, but there’s a clear drop-off in quality just outside of the center portion that looks to me like spherical aberration. There are halos around highlights and general smudginess around out-of-focus areas of the frame that looks like someone dabbed a little vaseline on the lens.
Spherical aberration is inevitable to some degree when you’re projecting an image through curved glass onto a flat plane, but I haven’t seen it like this in any smartphone camera I’ve tested over the past year and a half. Maybe that’s just one of the hazards of being an early adopter of a relatively new sensor format.
In any case, you do get more natural bokeh with the X90 Pro’s camera. It just comes with a side of mildly unpleasant lens aberrations. Putting a big-name brand on your camera doesn’t magically make it better, but Vivo is hardly the first to try it anyway.
Who wore it better?
I’m happy that the 1-inch sensor format is coming to more smartphones. It’s great to be able to get a little more of the background out of focus without flipping into portrait mode. And the RAW data speaks the truth: a bigger sensor is still better for low light noise performance. The photography snob inside of me would prefer a camera with a bigger sensor producing cleaner images rather than a high-res sensor bludgeoning noise into submission with more pixels and processing.
At the end of every imaging pipeline, there’s just one thing that matters: the image
But at the end of every imaging pipeline, there’s just one thing that matters: the image. The X90 Pro is technically better in some ways, but as a total package, I’d pick the S23 Ultra every time. Samsung’s color tuning and HDR aren’t always my favorite, but in most real-world situations, it just performed better than the X90 Pro. Either the extra pixels or savvier computational techniques give Samsung an advantage, and I’ll gladly take that over better high ISO RAW files any day.
Samsung and Vivo are both well aware that a camera is much more than its sensor, and the S23 Ultra seems to have all of those parts that make up an imaging system better synced than Vivo. The X90 Pro might not have fulfilled every promise of a bigger sensor format, but I highly doubt it will be the last 1-inch sensor smartphone I test this year — far from it.
Photography by Allison Johnson / noti.group
[Written in collaboration with other media outlets with information from the following sources]
