200 Megapixel sensor Smarthpone camera in the works, Motorola
It is still a rumor, but Motorola would be releasing a 200Mp camera sensor next year, which will be the first of its kind. Samsung and Xiaomi as well would begin employing the sensor in 2023.
Motorola's 200-megapixel camera phone, according to a report from a Twitter user going by the handle "Ice Universe," will be the first to hit store shelves. After that, Xiaomi, which is slated to introduce its new smartphone with a 200-megapixel camera in the second half of next year, will take the lead. Additional information provided by the source stated that Samsung wants to release its 200-megapixel camera phone in 2023 — following the launches of Motorola and Xiaomi devices.
In addition to the 200-megapixel camera phone, Motorola is reportedly working on the Moto Edge X which may feature a 60-megapixel OmniVision OV60A 0.61μm selfie camera sensor. The Moto Edge X would also have a 50-megapixel 1 / 1.55-inch OmniVision OV50A main camera on the rear as well.
Senior Member
Posts: 4767
Joined: 2008-09-07
I don't think more resolution will help facial recognition etc, we already have more than enough detail for a smartphone viewing close.
Image processing is what sucks.
Hmm...y'know I keep thinking about what you wrote and I really do not know why, but I think resolution is more important than sensor size and perhaps you are thinking more about the ability to split the RGBA and get higher contrast to determine image quality...but I'm thinking more about the increase in the ability of software to run facial recognition is more about scanning pixels and determining someone's identity?
No argument on a bigger and better sensor for image quality, but resolution must be the most important factor for security facial recognition software and algorithms. That is done on a CPU, and the better the CPU, the better the result of the software in "Real-Time". If this is fed a higher resolution, it would benefit security cameras, as these are fed directly into the software. As the work is done on CPU, the calculation is based on pixel data, which is a series of number representations of the RGBA channels from the sensor. In other words, with more numbers generates as a better percentile 'match' against a RGBA image database, and, can track multiple faces in a crowd. Surely?
Just my thinking, I guess. More k's of resolution is more important that the ability to determine kelvins, for security cameras.
Senior Member
Posts: 2367
Joined: 2018-04-10
My 10 mega pixel DSLR takes better pictures. Throwing mega pixels at a sensor is just a marketing move.
Senior Member
Posts: 13567
Joined: 2010-05-22
Hmm...y'know I keep thinking about what you wrote and I really do not know why, but I think resolution is more important than sensor size and perhaps you are thinking more about the ability to split the RGBA and get higher contrast to determine image quality...but I'm thinking more about the increase in the ability of software to run facial recognition is more about scanning pixels and determining someone's identity?
No argument on a bigger and better sensor for image quality, but resolution must be the most important factor for security facial recognition software and algorithms. That is done on a CPU, and the better the CPU, the better the result of the software in "Real-Time". If this is fed a higher resolution, it would benefit security cameras, as these are fed directly into the software. As the work is done on CPU, the calculation is based on pixel data, which is a series of number representations of the RGBA channels from the sensor. In other words, with more numbers generates as a better percentile 'match' against a RGBA image database, and, can track multiple faces in a crowd. Surely?
Just my thinking, I guess. More k's of resolution is more important that the ability to determine kelvins, for security cameras.
Looking too deep isnt key to checking identity unless you get down to DNA level, otherwise you will be looking at sweat formation, dirt and womens makeup (for example) which are useless for this purpose as the possible variations are huge.
For example, maps for navigation dont show the composition of soil and its formation because this has no value, they show elevation and larger patterns/shapes.
Similar applies to facial recognition and fingerprint maps.
Also, if the lense isnt vastly improved, increased resolution will do little more than add noise.
It will not increase detail capture.
ps
Higher res images require vastly more CPU power and memory (+memory speed) to process.
The problem with current systems isnt image quality, its programmed recognition algorithms that arent good enough.
There will inevitably be some constraints due to available processing power but as time goes on that will improve.
Raising the resolution would be a backward step though.
pps
Its possible processing of many captured frames of a noisy still subject can pull a bit more resolution from it but the extra processing would be put to better use running an improved algorithm for recognition.
Senior Member
Posts: 1845
Joined: 2012-04-30
might work using less res in actual settings/pics and use 48MP.
i know sony (and others) have large sensors with high pc, but sometimes run/use them at less res to gain low light perf.
Senior Member
Posts: 1845
Joined: 2012-04-30
good news for folks buying +500$ phones, so they can record 1080p video vertically