Exclusive: the vivo X60 series will be the first with pixel shift technology

20 March 2021
Pixel shift can be found on some DSLRs, but this is the first time it will be used on a smartphone.

Sort by:

aquinoes, 22 Mar 2021Forgot to say previously: this pixel shift technology on Ol... moreNot totally true. The first 2 iterations of pixel shift on Olympus camera, required tripod; however since EM1x and EM1 Mk III it's possible to have it also hand held (even if at lower resolution). So my guess is on mobile phones will work in a similar fashion

  • aquinoes

aquinoes, 22 Mar 2021When properly done, it really succeeds in restoring much mo... moreForgot to say previously: this pixel shift technology on Olympus cameras makes mandatory the use of a TRIPOD.
May be on Vivo X60 the gimbal stabilizer acting on that tiny sensor can stabilize as well as an EVIL camera on a tripod, and may obtain good results in good light conditions... but the minimum movement makes the technology useless.
Best attempts based solely on 'image registration', as on Pixel phones or Photoshop, in opposite to precise subpixel hardware controlled alignment, aren't unable to elevate resolution as Olympus has achieved (on tripod).

  • aquinoes

Anonymous, 21 Mar 2021A few years ago, Dpreview even posted an article that showe... moreWhen properly done, it really succeeds in restoring much more detail: https://www.dpreview.com/reviews/olympus-om-d-e-m5-ii/4


Olympus sensors are the smallest of all the camera manufacturers. But their cameras stand out for their OIS. So Olympus used its capacity to finely shift the sensor to achieve an effective resolution 4x greater that of the unshifted sensor.

  • Anonymous

Anonymous, 21 Mar 2021A few years ago, Dpreview even posted an article that showe... moreThough one needs to say that Photoshop can only align demosaiced raw files as far as I know.

  • Anonymous

Anonymous, 21 Mar 2021As I said, when Night Sight is slightly sharper than the de... moreA few years ago, Dpreview even posted an article that showed that when you align multiple raw files at subpixel level with Photoshop in good light conditions, the result isn't sharper than a sharpened single frame raw file. It's just less noisy and has a tiny bit less artifacts. See dpreview.com/reviews/image-comparison/fullscreen?attr369_0=20_image_stack_average&attr369_1=standalone_raw_oversharpened&normalization=full&widget=603&x=-0.28709151925237925&y=0.18589480724373086
Only when you also sharpen the super resolution result, there can be a tiny advantage regarding sharpness.

"Note: pixel shift shouldn’t be confused with sensor shift, which is an image stabilization technique (an alternative to OIS)."
Sensor shifting is indeed used for image stabilization (yes alternatively to OIS.) However, the pixel-shift technology to improve high-resolution images is also derived from the same technique of sensor shifting. It's actually sensor-shift technique that eventually shifts pixels in a frame. Means you really didn't need to add that line on the end...

  • Anonymous

Nick Tegrataker, 21 Mar 2021Not exactly sure what's happening to your units but I ... moreAs I said, when Night Sight is slightly sharper than the default mode at 1x in certain light conditions, this does not mean at all that this is due to a pixel shift super resolution algorithm. That's a wrong interpretation. Night Sight chooses a longer exposure time per frame, combines more exposures (total exposure time is longer, which leads to more light and less noise) and uses a lower amount of software noise reduction. These are factors that are extremely(!) important regarding sharpness.
So the only proper comparison is a perfectly exposed (exposed to the right) single frame - dng file in perfect light conditions and then there isn't a significant difference. There isn't even a significant advantage when the zoom is compared with an old Google phone like the 26mm equivalent Nexus 5x in perfect light conditions. And also the YouTube video that I mentioned shows no zoom advantage over the 26mm equivalent iPhone 12, even though the Pixel is equivalent to 27mm.
Also note that there are even additional factors, Google uses machine learning for the zoom, which adds fake detail at pixel level while making contours smoother. I think that a lot of people mistake Google's machine learning details with super resolution.
So, the interpretation is very important as there are a lot of factors that play a role, therefore only a comparison with a single frame dng file is acceptable.

  • Anonymous

Are you sure you would be able to tell the difference between pixel shift and straight out sharpening with object edge detection on such a tiny sensor with crappy optics? I'm not.

  • Wereweeb

"Yeah, they copied that from Apple"
-Someone, in 10 years when Apple adopts this

  • lonewalker

first android phone though

Anonymous, 20 Mar 2021No, I have a Pixel 3a and I had three different Pixel 3a an... moreNot exactly sure what's happening to your units but I guess we can agree to disagree. I prefer to believe what I saw with my own eyes, including a bunch of 1x Night Sight samples taken by Pixel 3 during the day clearly resolving more details with less false colours and pattern distortions than regular 1x samples. However, it's obvious that you're not seeing what I'm seeing and vice versa, and I'm not here to convince you that my own fact is superiour to your own fact especially given that you're an owner of multiple Pixel phones.

  • Anonymous

Anonymous, 20 Mar 2021No, I have a Pixel 3a and I had three different Pixel 3a an... moreAnd regarding the iPhone 12 vs Pixel 5 YouTube comparison from Irina Tech: Keep in mind that the Pixel 5 has a 27mm field of view whereas the iPhone 12 has 26mm, so even though the Pixel 5 has a slight hardware advantage here, its zoom doesn't really perform better.

  • Anonymous

Anonymous, 20 Mar 2021No, I have a Pixel 3a and I had three different Pixel 3a an... moreAlso note that it depends on how noisy the light conditions are and which exposure the camera chooses. If the camera chooses a too short exposure time (for example in order to protect the highlights), then even good light conditions will be noisy and will benefit from less noise due to more frames. Google's zoom was advertised by avoiding demosaicing, but it actually mainly works by reducing the noise and by using machine learning, so it's not really better than traditional digital zoom in perfect light conditions, it is hard to see a significant difference when it is compared with a crop of a well exposed single frame dng file in perfect light conditions.

YUKI93, 20 Mar 2021Lol, don't expect to see a DSLR camera sensor on a pho... moreI wouldn't be so sure, retractable optics can help, so can bent sensor I guess. Not to mention there is some light bending tech coming(which alone would add medium format capability without former tech).

  • Anonymous

Nick Tegrataker, 20 Mar 2021I've seen dozens of full size photo samples from Pixel... moreNo, I have a Pixel 3a and I had three different Pixel 3a and I also tried Pixel 3 and Pixel 4 in a store and I looked at dozen camera reviews with full resolution sample images. It does not work in perfect light conditions. For example, see the Pixel 5 vs iPhone 12 YouTube camera test from Irina Tech.
Note that you can't simply compare 1x photos with 2x photos. Even if the 2x photos have slightly more detail than 1x photos in certain light conditions, it does not mean that it works. The 1x photos can suffer from more noise reduction(so they are not as sharp as they could be) and Google uses machine learning and the exposure time of the 2x photo can be different, too. Furthermore, I said that the zoom works in worse light conditions as they combine more frames when you zoom, which leads to less noise.

Anonymous, 20 Mar 2021Huawei can't get colors properly, all of their images ... moreRYYB is one of ways to improve photography, it being new its nothing surprising it came with some problems, which however are being solved and with every generation it gets better, actually last one is close to the best, even if its not perfect and guided with color sensor, it being no mirrorless ff, using multiple sensors, stacking etc. it expected.
Low light is nowhere near limits, even with smaller sensor you can get more if used right tech, what's more there is more development here in phones than in premium camera segment, ryyb/ color sensors/ stacking/ no IR filter sensors/ AI features etc. Basically its not just about size here but about tech - new better sensor and better processing(stacking and the likes) will wipe floor with old ones if you compare performance of the same size - of course compared to FF its getting just closer, but so is with size... I can also mention that still nobody (I believe) used full spectrum sensor for low light, not to mention full spectrum without any rgb/ryyb filter, these 2 features could multiply its low light performance also bringing it over mirrorless as those would not stop you from also using ''visible colors sensor'' as opposed to hw redesign of mirrorless(like to full spectrum).
Considering HDR its matter of sw, but for jpeg its better to have more content than less, fine tunning would be great but there is still option to disable it. If you can shoot raw it would solve all your problems, but not every phone can shoot raw with all cameras (for ordinary user), from what I know, only Vivo and Sony can. Back to this, I hope I see kind of blend of samsung's hdr raw and vivo's hi res stacking, having it all - hi res hdr raw in all fov's.

YUKI93, 20 Mar 2021Lol, don't expect to see a DSLR camera sensor on a pho... moreThey could have already done that, but as usual, niches/minorities are left aside, partially because of users complaining about any idea different from the mainstream stuff, also because of people buying the phones regardless if they are good or not.

Erin, 20 Mar 2021Dumping everyhing into a phone is lazy technological innova... moreIt is still much better than removing useful features that people do want and do use actively though.


Something many do not know.

The values 1/3", 1/2.6", 1/1.7", 1/1.2" are about the pixel area only .
Not overall size of sensor.

Samsung sensors have more " bezels" than sony sensors. This means they take more space with same pixel area.


Ice Universe has posted that GN2 is largest cmos for mobile in 2021.
The imx800 being 1" was fake story.

That was obvious. With so many rear cameras (also ToF/laser AF), there is no space enough for 1".

1/1.2" has 85mm2, while 1" has 123mm2. ( both 4/3 ratio).
That's 44,7% difference.
Too much space required.