MediaTek's latest Dimensity phone chippery is 25 per cent faster than its predecessor, supports 200MP cameras Taiwanese chip flinger MediaTek has introduced two new additions to its Dimensity SoC lineup, both targeting higher-end smartphones. The Dimensity 1200 and 1100 are both manufactured using TSMC's 6nm process and use the same GPU, Arm's Mali-G77. Across the board, there's support for dual-sim 5G connections, Wi-Fi 6, and …
When not watching changing content (eg while looking at texts) even 60Hz is more than is needed. It would be nice if the refresh rate automatically dropped to the minimum for the panel when there was no change in the content being displayed - this might eke out a bit longer battery life.
That's what is being worked towards. I agree it would make perfect sense to drop to 1 Hz when your phone is just sitting there with nothing updating on the screen, and only jump to a higher rate when things are changing.
Same thing for PCs, once you have VRR there's no reason it should only be used for games to get high frame rates. While I'm typing this post the only things changing on the screen are the letters I'm typing and the seconds on my clock changing. 1 Hz would be too slow but 20 Hz would be fine.
> Both chips are capable of powering QHD+ (circa 3200 x 1440) displays at a smooth 90Hz refresh rate. Things get slightly bonkers when you drop the resolution down to FHD+ (roughly 2400 x 1080), with the Dimensity 1100 supporting refresh rates of 144Hz, while the Dimensity 1200 is capable of powering 168Hz screens.
> This feature is likely only of interest to those using their phones for competitive multiplayer gaming.
I know there's competitve games on mobile, but will refresh rates really make any perceptible differences? With your fingers obscuring lumps of the screen - and all the auto-aim mechanisms which have to be in place due to the low precision of touch-screen controls - I would have thought that the impact of upping the screen refresh rates would be so small as to be unquanitfiable.
(Saying that, this Nvidia page makes a lot of claims for how higher FPS affect gameplay. OTOH, they're literally in the business of selling faster and more expensive hardware to drive faster FPS, so I suspect their claims will be somewhat... embellished, shall we say.
https://www.nvidia.com/en-us/geforce/news/what-is-fps-and-how-it-helps-you-win-games/)
is anyone likely to be doing serious competitive gaming on a mobile phone?
> In normal use, you'd crank that way down to conserve battery power, with 90Hz offering the perfect compromise between display speed and battery consumption.
Really?
I see a fair number of reviews from people enthusing about how upping the refresh rate makes things "buttery smooth". But I've never been convinced it makes a measurable difference.
And it doesn't take much searching online to find some (non peer-reviewed) blind testing which shows that even at best, very few people notice the difference between 90Hz and 60Hz.
https://www.androidauthority.com/90hz-smartphone-display-test-1065904/
https://www.phonearena.com/news/90Hz-vs-60Hz-display-refresh-rate-oneplus-7-pro-do-people-see-difference_id117148
https://www.phonearena.com/news/60Hz-vs-90Hz-refresh-rate-test-can-people-tell-the-difference_id121479
So, yeah. I can't help but suspect that the ability to drive screens at a million Hz is being done more for marketing purpose than anything else. And it'll sadly be turned on by default in most smartphones, thereby driving down battery life and wasting energy for no good purpose whatsoever.
(Though to play devil's advocate for myself, I'm guessing the high refresh rates can be useful when driving 3D displays, since there you have two displays to render and 3D viewing is notoriously more sensitive when it comes to refresh rates. But that's pretty much a niche case)
Anyhow, that's my "I don't see a need for it, so no-one else must have a need for it either" grumpy post of the day. Maybe I should have some more caffeine before going back on the interwebs again...
More pixels at this level never means better pictures, given the objectively crappy lenses and tiny sensors fitted to phones.
There's a very good reason why a typical full frame SLR standard (50 mm) lens of any quality is (and always has been) around 60 by 50 mm in size and consists of maybe six or seven elements in multiple groups. Furthermore, as all sensors fitted to phone cameras use a Bayer filter, real resolution is a lot lower than that implied by pixel count, and high pixel count sensors small enough to fit in a phone camera (limited by optical geometry and phone body thickness) are extremely noisy in low light levels. Unfortunately these issues are dependent on physics, so they're immutable.
Phones include a lot of fancy math processing to disguise these faults, but that means the image is much less a record of reality than a piece of constructed art. However higher numbers attract higher prices, which is all phones are really for to the vendor.
When I first got my Lumia 1020 the high pixel count (given good light levels) was really something which I found useful. However, an update (Denim I think) messed with the image processing and after that any full resolution photos start to look like "art" when you zoom in too much. Real shame because the camera used to be one of the main features of the 1020 :-(
> When I first got my Lumia 1020 the high pixel count (given good light levels) was really something which I found useful
The 1020 was definitely ahead of it's time. I was curious about the camera, so I picked one up from CEX at the tail end of the Windows Mobile era, and was pleasantly surprised by how good it was as a phone - for all that it never caught on, Microsoft's "tiles" UI was a decent alternative to the Android/iOS ways of working.
Unfortunately, I was never as impressed with the camera. I don't know if it had the Denim update (probably - I generally make sure my phones are running the latest software), but the photo quality was definitely disappointing; it was slow to focus and definitely didn't like poorer light conditions. Which is a shame, as I'd hoped that the "pixel binning" approach would offer a way to compensate for low lighting.
Still, what comes around goes around, and 7 years later we're pretty much at the limit of what can be done with physical mobile-phone camera tech. But we also have a lot more processing power to spaer, so it's time once again to see if software can compensate for the hardware limitations...
Alas, the results haven't been amazing so far. Personally, I'm waiting for the first proper reviews of the S21 Ultra, to see if Samsung have been able to improve on the lackluster performance of the 108MP camera on the S20+...
> Phones include a lot of fancy math processing to disguise these faults, but that means the image is much less a record of reality than a piece of constructed art. However higher numbers attract higher prices, which is all phones are really for to the vendor.
I'll happily agree that SLR photographs are higher quality than those from phones. And obviously, SLRs can use special lenses and physical filters to achieve things which are physically impossible with a camera lens.
Fundamentally though, neither is a perfect "record of reality". To vastly over-simplify, both take an analog input, split it into separate channels and then quantify these into a set of digital values.
What you get is not what you saw. Hell, even eyeballs and traditional analog/chemical film quantifies stuff to some level.
So, the question is: is the representation you get out of said "imperfect record of reality" fit for purpose?
And at least for me, when it comes to smartphones the answer is generally "yes".
I've recently moved into a flat, and decided to amuse myself by getting some A3 prints of the many (many many many) street art photos I've taken. These are from a range of compact cameras and mobile phones, ranging from 2014 all the way up to last year.
And the quality of these A3 prints is more than good enough for framing and hanging on the wall. If anything, I've actually tended to prefer the photos from the smart phone, since they tend to be sharper and have better colour balance.
(TBF, you can equally argue that they've been artificially tweaked by the camera software on the phone. But the result either way is a photo which looks measurably better to my admittedly unprofessional eye)
And if my smartphone photos - from handsets which are now several generations old - are good enough to print at A3 size, then I really don't see the need to lug a DSLR or even a compact camera aronud with me!
"To vastly over-simplify, both take an analog input, splits the arriving image into separate channels and then quantify these into a set of digital values.
A traditional colour film actually does something quite similar. It splits the incoming image into three primary channels recorded on stacked emulsions separated by colour filters. The equivalent of the 'digital' data consists of the degree to which the silver nitrate grains in the latent image converted to metallic silver on development. That process bleaches the filters and replaces the silver in each layer with a dye of the required primary colour.
The biggest differences are the density of the grains versus pixel density ( for a 35 mm frame two to thee billion for film versus 20e6/mm2 for a 200 MP sensor), and the random and overlapping distribution of grains in the three layer emulsion versus the planar Bayer filter in the digital camera. Furthermore, the Bayer filter uses four adjacent photo sites in each axis (actually a 4x4 clump) to determine the intended colour of a pixel. Obviously this constitutes a reduction in real resolution as every resultant pixel is some sliding average of a clump.
Neither of film nor digital sensors actually reproduce exactly what was in front of the camera (or even exactly what came through the lens) but the intrinsic artefacts of a Bayer filtered digital sensor are vastly greater than those of a film. For example, the intrinsic noise of a film emulsion is much lower, as a minimum photon count is needed to 'seed' a silver halide crystal for it to be converted to silver when processed. Otherwise it is generally washed out during processing. In the case of 'fog' (an unwanted overall background density), it is typically quite uniform due to the high density of the grain. On the other hand, a digital sensor does not start from a zero point and accumulate a charge as it receives photons. Instead, it is precharged and loses charge as photons arrive. However the inherent leakage of photo sites varies, so at lower image light levels the photo site values due to leakage of the precharge increasingly dominate in the image.
Apologies to all readers - "( for a 35 mm frame two to thee billion for film versus 20e6/mm2 for a 200 MP sensor)"
should read
"(for a 35 mm frame two to thee billion grains per cm2 per layer for film versus 20e6/cm2 for a 200 MP sensor)"
> science science science
*nods wisely*
It does indeed all come down to making sure the quantum flux is correctly transliterated by the gigawatt multiplexer. Very well put.
(I don't think you got away with it - Science Dept Ed)
> Neither of film nor digital sensors actually reproduce exactly what was in front of the camera (or even exactly what came through the lens) but the intrinsic artefacts of a Bayer filtered digital sensor are vastly greater than those of a film
And that's part of the point, which I think the various elitist "smartphone cameras are worthless" are missing.
Yes, a DSLR will have far better optics and a bigger sensor. But it's still only capturing an approximation of what it saw. And it's still measurably inferior to a traditional film camera. It's just that digital is far more convenient - and cheaper, since you don't have any development costs. And the fact that you can pull RAW data directly from the CCD means that you can use software to tweak the output to more closely fit your requirements[*], rather than being stuck with just what you got at the time.
And to be fair, smartphones physically can't do what DSLRs can. But what they can do is often more than good enough for most people. And they have another key benefit, which is that they're far smaller and lighter than a DSLR, and most people carry one around with them at all times.
After all, the best photo is the one that you take which whatever you have available at the time.
Still, like I said below, trolls gotta troll, and elitists have to sit at home clutching their mega-expensive hardware, while the rest of us go out (at least in better times) to have a good time and take lots of photos, often in places where you wouldn't take a DSLR...
[*] Using software to improve visual quality? Sounds like something those nasty smartphone users would do!
To get better pictures you need talent and a bigger image sensor, that and semi half decent lenses too. You know, the kind of problems that all smartphones will always struggle with, problems that no amount of trickery will ever be able to solve.
> Smartphone pictures are commonly and utterly worthless
Indeed, you're quite right. There's never been a single photograph of any worth whatsoever which has been taken on a smartphone. Or a compact camera. Or a 35mm fixed-focus film camera. Or a Polaroid. Or any camera which cost less than a small car and doesn't come equipped with enough lenses to build a telescope which can resolve the writing on a postage stamp sat atop a mountain on Mars.
I mean, look at the photos which were entered for the 2019 Mobile Photo Awards.
https://mobilephotoawards.com/street-photography-winners-9th-mpa-winners/
Utterly useless, the lot of them.
Still, trolls gotta troll...
Photos from Samy's 108 Mpix sensor look bad at ISO 50 and it's all downhill from there. Pixel binning can only compensate for a lack of photons per pixel. It can't compensate for the circuitry and optical compromises.
A lens array (insect eye) could make use of 200 Mpix in a phone's form factor. That would come with some other bonus features, like variable DOF and depth capture, but that's a lot of image processing for current technology.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
