Camera Architecture: Huge New Sensors

As we move on to the camera evaluation of the S20 series, I think it would be good to recap the new sensor architectures that Samsung is deploying in its new flagships. To say that this is the biggest camera hardware upgrade that Samsung has ever embarked on is a bit of an understatement, as the new modules on the S20 series, and in particular the S20 Ultra, have seen some fundamental shifts in terms of their designs and specifications.

Samsung Galaxy S20 Series Cameras
  Galaxy S20 Galaxy S20+ Galaxy S20 Ultra
Primary Rear Camera 79° Wide Angle
12MP 1.8µm Dual Pixel PDAF



 
79° Wide Angle
108MP 0.8µm DP-PDAF


3x3 Pixel Binning to 12MP
8K24 Video Recording
fixed f/1.8 optics
OIS, auto HDR, LED flash
4K60, 1080p240, 720p960 high-speed recording
Secondary
Rear Camera
76° Wide Angle
(Cropping / digital zooming telephoto)
64MP 0.8µm

F/2.0 optics, OIS

8K24 Video Recording
24° Telephoto
(5x optical magnification)
48MP 0.8µm

2x2 Pixel Binning to 12MP
F/3.5 prism optics, OIS
Tertiary
Rear Camera
120° Ultra-Wide Angle
12MP 1.4µm f/2.2
Extra
Camera
- Time of Flight (ToF) 3D Sensor

Starting off with the elephant in the room, that’s the new S20 Ultra. Last year we first saw talk of Samsung LSI introducing a new 108MP camera sensor that had been developed in collaboration with Xiaomi. Although closely related to that design, the HM1 sensor is a bit different to the HMX sensor that’s employed in the Xiaomi phones. What’s special about these new sensors is their sheer size: at 1/1.33”, it’s over double the sensor area of previous generation units found in past Galaxy phones. A lot of people will criticize the 108MP count as being a gimmick, but in light of the huge new sensor the actual pixel pitches aren’t exactly all that smaller than what we’ve seen from previous generation high-megapixel sensors from the last year or two, still falling in at 0.8µm.

Source: TechInsights

What’s rather peculiar about the HM1 sensor in the S20 Ultra though is that this isn’t just another quad-Bayer sensor like those seen in other designs over the year, but rather a “Nonacell” design, with the color filter array covering up 9 subpixels. Just like quad-Bayer designs, the new nona-Bayer  sensor is able to demosaic chroma information to be able to actually achieve the advertised 108MP resolutions. Physically, chroma resolution still is only 12MP on the sensor, and that’s something that you’ll want to keep in mind when we’ll later on investigate the 108MP samples of the phone.


Source: iFixit

As seen in the above teardown shot, the 108MP unit dwarfs the other sensors in the phone, but this does come with some complications. On the optics side, Samsung has now opted for a fixed f/1.8 aperture and has dropped the dual-aperture f/1.5-2.4 system we’ve seen on the S9 and S10 series. Because the new sensor is so huge, it actually becomes an issue to design proper optics that actually fit into the z-height of the phone. Even though the S20 Ultra has a thick camera bump now, it still has to make do with a smaller aperture optical lens than previous generations. The new bigger sensor also has another side-effect, that being a shallower depth of field when focusing on objects. I do find it a very big pity that Samsung opted to drop the dual-aperture system as that would have been the perfect fit for such a big sensor design and essentially eliminate any potential drawbacks on the optics side. Unfortunately, as it’s gone, I do expect the optics to not perform quite as optimally as we’ve seen on the S10 series.

Below the 108MP sensor, we find another unique design that’s exclusive to Samsung at this moment, and that’s the new periscope design telephoto module. Such modules were pioneered by Huawei as we first saw the P30 Pro last year bring it to the mass market, however Samsung goes way beyond what any other vendors is currently brining to the market. The optical magnification of Samsung’s design isn’t too special, only reaching a factor of 4x and a resulting 24° field of view or 103mm equivalent focal length. What is special though is that Samsung has crammed in a full 48MP IMX586 into the module, going far beyond the smaller 12 or 8MP sensors that are being deployed by other vendors such as Huawei. It’s the usage of such a big sensor that lies perpendicular to the body of the phone that actually forces the S20 Ultra to have such a huge camera bump on the back – as it’s being limited by the thickness and the footprint of the telephoto camera module.

To square off the trio of cameras on the S20 Ultra, we see a new ultra-wide-angle module. Samsung here opted to go for a lower megapixel count sensor at 12MP compared to 16MP on the S10 series, but the pixel pitch increases from 1.0µm to 1.4µm, which should allow it go achieve better per-pixel sharpness and low-light capturing abilities. It remains quite wide at a 120° viewing angle, just a tad tighter than 123° unit on the S10 series.

When looking at the different viewing angles of the S20 Ultra, we see some big discrete steps between what the different sensors are able to natively capture at their fullest. The main sensor on the S20 Ultra is actually a tad wider than what Samsung is marketing, and it produces 25mm equivalent focal length images, a little wider than the 26mm of the S10 series and the other S20 phones. Of course, the telephoto lens will have a small field of view of only 24°, but that’s precisely what allows it to achieve such high magnification levels, also thanks to the high resolution captured within this frame.


Source: ITSub

While the S20 Ultra’s camera design is fascinating, the design that actually excited me a lot more when the phones were first introduced is the camera trio that’s found on the regular S20 and S20+. Here Samsung is using a completely different approach that’s pretty much unique in the industry.

Starting off with the main sensor, this is seemingly a pretty straightforward design that really only differentiates itself through the fact that’s it’s now a bigger sensor falling in at 1/1.76”. The resolution still is 12MP and it’s a regular Bayer sensor, so pixel pitches now grow to 1.8µm. The optics remain similar to the S20 Ultra’s, coming with a fixed f/1.8 aperture lacking the dual-aperture system and of course also includes OIS. The Ultra-Wide-Angle is also the same as on the S20 Ultra.

What’s really exciting about the S20 and S20+ is the “telephoto” module. The weird part is that this isn’t a telephoto module at all, but it’s an actual secondary wide-angle lens that’s only slightly tighter than the main camera. The sensor here is also large at 1/1.76”, but it comes a 64MP resolution with 0.8µm pixels. I actually haven’t heard confirmation of the color filter array of the unit, whether it’s quad-Bayer with remosaic or whether it’s a true 64MP Bayer sensor.

Why Samsung is able to call this a 3x telephoto module is that when cropping a 1:1 12MP picture out of it, it does end up at a 3x magnification in relation to the main camera sensor. The question you’re of course posing, is why would Samsung go for such a camera configuration? The first answer, possibly the most obvious one, is 8K video recording. As the main camera unit’s 12MP native resolution isn’t sufficient for 8K video recording, Samsung needed to find a way to include this into the regular S20 series as well – and obviously if you had an actual telephoto module for this then you’d end up with a pretty useless setup. Going for a secondary wide-angle module kills two birds with one stone, as you have one module being able to serve as the 8K video recording unit as well as taking advantage of the high-megapixel count of the unit to be able to achieve respectable crop-zooming.

I was most excited about this setup, more-so than that of the S20 Ultra, because it opens up a lot of possibilities in terms of sensor fusion and computational photography that’s physically just not possible on the S20U. Without spoiling the camera evaluations too much, this also means that the S20 and S20+ have quite high quality zoom capabilities in the 1-3x range as well, and it doesn’t behave as a “normal“ telephoto module at these intermediate levels.

Battery Life: Good (60Hz) to Average (120Hz) Camera: Daylight Evaluation
Comments Locked

137 Comments

View All Comments

  • toyeboy89 - Friday, April 3, 2020 - link

    I'm really amazed in the fact that the iPhone XR is still beating snapdragon 865 in GFXBench in both peak and sustained performance. I am hoping the OnePlus 8 has better sustained performance.
  • TMCThomas - Friday, April 3, 2020 - link

    Amazing review! Always wait for this one before getting a new samsung. And I won't be getting any of the s20 phones. For me the kind of feel like "beta" phones. The 120hz which is not quite ready for 1440p yet, the underutilized 108mp camera, the space zoom which is blurry, the camera hole still being there the big camera bump and so on. I think all these features and more could be way more refined with the next galaxy s which I'll be waiting for. Also the poor exynos 990 performance especially the GPU part is just unacceptable to me. Especially with it probably being a lot better next year, so I'll skip this year
  • wheeliebin - Friday, April 3, 2020 - link

    Thanks Andrei, really good review!

    I have read many users complain about extra crazy post-processing on the S10/S20 series when there is a face detected in the frame. i.e. the phone will apply an aggressive 'smooth skin' filter that you can't disable unless you shoot RAW. I was hoping that your review might touch on this however there were no people in your example shots so perhaps you didn't get a chance to experience the problem. I wonder if you have heard of this issue and can replicate it yourself with the S20 range?
  • anonomouse - Friday, April 3, 2020 - link

    Hi Andrei, did you also run the bandwidth and MLP sweeps from previous reviews? Last year you noted the Snapdragon 855/A76 had peculiar behavior in the L1, and it would be also interesting to see if there are any MLP changes in both the SD865 and the Exynos.
  • anonomouse - Friday, April 3, 2020 - link

    Also, any idea why the new scores for these in 403.gcc seems to be worse than their previous generation products? In particular the score for the SD865 in these S20s is substantially worse than the SD865 score from the QRD preview article.
  • Andrei Frumusanu - Saturday, April 4, 2020 - link

    Yes I know. I don't know why that happens. I also got a V60 now and the scores there are higher, I'm wondering if there's something with Samsungs shared libraries.
  • anonomouse - Sunday, April 5, 2020 - link

    What type of compile flags are used for these binaries? Are they the same for all of the tested binaries (or even same binary on each given platform)? Are LTO or PGO used (and if not why not)?

    I'm also not convinced of this statement from the article:

    "I had mentioned that the 7LPP process is quite a wildcard in the comparisons here. Luckily, I’ve been able to get my hands on a Snapdragon 765G, another SoC that’s manufactured on Samsung’s EUV process. It’s also quite a nice comparison as we’re able to compare that chip’s performance A76 cores at 2.4GHz to the middle A76 cores of the Exynos 990 which run at 2.5GHz. Performance and power between the two chips here pretty much match each other, and a clearly worse than other TSMC A76-based SoCs, especially the Kirin 990’s. The only conclusion here is that Samsung’s 7LPP node is quite behind TSMC’s N7/N7P/N7+ nodes when it comes to power efficiency – anywhere from 20 to 30%."

    Both the energy consumed and the performance scores for both of these A76's seem to also very closely track the "mid" 2.43Ghz A76's on the TSMC-fabbed SD855 - all of which have similar L2's and similar frequencies, but possibly differ significantly (to the point of being suboptimal on latency) on the memory hierarchy and SoC beyond that - which greatly affects many of the SPEC workloads. All of these may also have implementation targets. Given this, is it really conclusive that the Samsung process is truly 20-30% worse in energy efficiency? Granted, things will probably not look pretty next year when TSMC is on a true 5nm and Samsung is not.
  • Andrei Frumusanu - Monday, April 6, 2020 - link

    The test is just -Ofast without any other addition. LTO wasn't/isn't in a good state on the Android NDK - it's something to look into in maybe a new binary revision.

    As for the 855 figures, well, that's also on an earlier 7nm. HiSilicon did a lot better in terms of they physical implementation. If not against N7, 7LPP clearly has a disadvantage against N7P/N7+.
  • Andrei Frumusanu - Saturday, April 4, 2020 - link

    I'll add them in, that test takes a whole day and I needed the phones doing battery tests and other stuff.
  • dad_at - Saturday, April 4, 2020 - link

    Again, your S10+ Exynos results in pc mark are false as of 2020. In performance mode I easily get 9500 work 2.0 overall, about 9600 in browser bench, 21K in photo editing. PC mark in general is inconsistent, irrelevant benchmark, not representative of actual performance in daily usage. The same about these ancient SPEC synthetics. No one uses these for performance evaluation now.

Log in

Don't have an account? Sign up now