13 ms·

Inside the Development of Light, the Tiny Digital Camera That Outperforms DSLRs

127 points by teklaperry 10y ago

sytelus 10y ago
TLDR; Phone cameras are not as good as SLRs because of tiny sensor, tiny lens and fixed focal length. How do we improve them then? Well, put as many as 16 different sensor+lens combinations in a small body, capture multiple views simultaneously and then algorithmically combine them to generate picture that looks as good as SLR. I think its cool idea and this is where smartphone industry is heading anyway, iPhone 7 being the first example. The key is algorithms to combine multiple views which can invariably introduce errors and artifacts for the trained eye. In extreme case, the entire back side of phone can be just rectangular matrix of 100s of lenses and sensor capturing the scene simultaneously and generate gigapixel image that can be viewed at variable focal length after the fact. That would be cool!
- sohkamyung 10y ago
  It sounds like a good idea. But it may not be suitable for all conditions where people use a DSLR. One example I can think of is macrophotography. In this case, the subject is simply too close to the camera to be covered by the numerous lenses of the Light camera. I'm also not sure if the mechanical mirrors can be moved fast enough for high-speed action photography; but it might be. Of course, it is a very specific example. For general usage, the Light may be a compelling replacement for a DSLR.
  - 6nf 10y ago
    Yea the closer you get to the object, the worse the picture. But on the upside you start to get pretty good 3d scanning. With multiple views from different angles you'd be able to create detailed 3d scans of small objects. You may want to use this in video chat - each person's face will be rendered in high accuracy in 3D! Unfortunately the screen is on the other side of the cameras...
    - fwn 10y ago
      ..but at that price and with all the optical problems they seem to solve, a flipscreen is probably not that big of a deal to add.
  - Tepix 10y ago
    For "normal" macro photographs, smartphone cameras are already doing an OK job. If you put a macro lens in front of your cellphone camera, it can be better. Phone cameras have issues elsewhere: * Low light exposures with low noise * low depth of field * telephotography
    - dharma1 10y ago
      I would add lower dynamic range and noise in general compared to dSLRs, because of small sensor/photosite size
  - Obi_Juan_Kenobi 10y ago
    While this approach won't get you good macro performance, with some tweaks the DoF from small sensors could be a real boon. If you could pack a few of these close together, especially in a small package, there's a wonderful opportunity for wide-angle macro photography.
- Yetanfou 10y ago
  ...iPhone 7 being the first example. No, iPhone 7 is certainly not the first example of a multi-camera phone. There have been several multi-sensor phones, this year both Huawei (P9) and LG (G5) were launched with such. In an article on these devices it is noted that 'Apple's iPhone 7 Plus is rumored to be following their lead later this year'. Correct, Apple did follow their lead, especially Huawei's as the P9 does just what is described in the article: it uses multiple sensors to increase image quality. http://www.theverge.com/2016/4/11/11406098/lg-g5-huawei-p9-two-camera-smartphone-trend-app http://www.theverge.com/2016/4/11/11406098/lg-g5-huawei-p9-t... Please don't perpetuate this myth of Apple being first with such and so. As a rule they are not first with anything, instead they see what others are doing and produce their own version of the idea. A refined version often, but that does not make it the first version.
  - jahnu 10y ago
    They are often the first to remove things. #courage
  - reitoei 10y ago
    There were actually a couple of cameras before the P9 and G5 using dual lenses (and I assume sensors) for doing weird 3D/DOF field stuff: http://www.pocket-lint.com/news/138719-dual-lens-smartphone-cameras-the-history-running-up-to-iphone-7-plus http://www.pocket-lint.com/news/138719-dual-lens-smartphone-... Your (and the most important) point about the iPhone 7 not being the first is bang on.
  - glenndebacker 10y ago
    "Please don't perpetuate this myth of Apple being first with such and so. As a rule they are not first with anything, instead they see what others are doing and produce their own version of the idea. A refined version often, but that does not make it the first version." I really don't understand why people are getting so worked up of company X of Y being first or debunking Apple myths.... I personally couldn't care less (and I think most consumers) because I value a good working implementation before the fact that a brand is the first in anything. Why does that matter?
- agumonkey 10y ago
  I've been told about superresolution before, since, I'm waiting for smartphones to use it for pictures. The megapixel race produced such horrendous sampling and compression artifacts.
- Mithaldu 10y ago
  It's a technological Beholder, only misses the tentacles. :D
helloworld 10y ago
Our first-generation L16 camera will start reaching consumers early next year, for an initial retail price of $1,699. It's a cool concept, but I'm hoping that the second generation is a lot cheaper.
- misev 10y ago
  If it's equivalent in quality to a DSLR, especially in low-light, then it's definitely worth it.
  - hyperbovine 10y ago
    Not so sure, for a lot less than $1699 you can buy some very good micro 4/3 cameras which would probably give this a run for its money in low light settings. micro 4/3 is not quite "tiny", but judging from the picture of the founder holding one up, neither us this (despite what the headline claims).
    - misev 10y ago
      The largest part of a micro 4/3 or a DSLR is the lens usually, and this thing goes around that, so I'd say it's seriously smaller. We can just speculate at this point about the low light performance though, I can't wait to see some proper review.
      - slantyyz 10y ago
        You can consider a fixed lens 1" sensor compact like a Sony RX100. It's got half-decent low light performance. It may not beat a 2016 consumer level DSLR for low light, but it's probably comparable to a 2008-2010 Rebel.
    - bobsil1 10y ago
      There's a mobility premium since this fits in your pocket. Much smaller than an A6300 with pancake lens. You could potentially jog with this in a wildlife refuge, then take bird shots.
    - 10y ago
nradov 10y ago
It's an impressive feat of engineering but I'm having trouble seeing the point. Does it actually take better photos than the latest generation of compact cameras that have legitimately good optics and sensors? I don't doubt that it's better than a cell phone camera, but if you're going to carry another device then why not bring one with a real zoom lens?
- misev 10y ago
  It claims to be as good as a DSLR.. which is better than a compact. We'll just have to wait for some actual tests and reviews that will substantiate the claims. I'm optimistic, even if the first version might not be there, this development seems to be going in the right direction with probably a lot of room for improvement. If it's only as good as a compact, then I'd think it's still a better option due to the selective DOF that it is supposedly able to do. But then it's definitely on the pricey side, so it's questionable if it's still worth it.
- alkonaut 10y ago
  This one is arugably a camera in the "latest generation of compact cameras". It competes with them since it's a separate device, and it competes with the highest end ones, such as the Sony Rx-series and the Fuji x100t.
ianpurton 10y ago
You buy a DSLR because it has a large sensor that can capture lots of light and detail. Lot's of smaller CMOS sensors when combined won't be able to capture the same level of detail. So I fail to see how this camera what outperform a DSLR.
- sohkamyung 10y ago
  It might, depending on the usage. The Light camera captures the same image using multiple sensors, which may work out to the same, or higher resolution to a single DSLR sensor. One interesting thing the multiple sensors on the Light can do, which a single DSLR sensor can't, is to capture the same scene with multiple exposure settings, as mentioned in the article. Combining these different exposures together may allow for higher contrast (HDR) images to be captured.
  - dingaling 10y ago
    It's not resolution which is the limiting factor for small sensors; in fact many of them have higher per-sensor resolution than a DSLR. It's the fact that as you push more and smaller detector-sites into a such a small array you experience more and more noise, which means the image quality degrades rapidly with falling light. There's a balance between size of detector-sites and array density and any technology that reduces small sensor noise can equally be applied to large sensors with larger detector-sites in order to maintain their advantage. The best analogy I encountered was to consider each detector-site as a 'light-gathering bucket'. A big bucket can catch more photons and hence has a higher SNR than a small bucket once you deduct cosmic rays and read-out errors etc. So a medium-format 21MP is better than an APS-C 21MP array which is better than a phone-sized 21MP array. It doesn't really matter how many small-arrays you point at one target because it's the SNR at the individual detector-sites that matters. ~~ TLDR: this camera can never compete directly with a DSLR on raw photo quality but will offer other interesting capabilities that DSLRs can't.
    - mnw21cam 10y ago
      Noise is much more noticeable in dark areas of an image, so having a method to produce a HDR image will naturally improve the noise. So, I think this camera could quite easily compete directly with a DSLR on photo quality - but a good DSLR will easily be better in low light conditions.
misev 10y ago
This is exciting development, going in just the right direction. Divide and conquer! If the first prototype proves to be at least close to DSLR image quality, it will be a revolution.
pmlnr 10y ago
people,it's not the sensor why a dslr is the etalon, it's the lens you can put on them.
- mnw21cam 10y ago
  Not so. It's easy to make a really good small lens - the article mentions that the modern smartphone lenses are dirt cheap and diffraction limited. It's really hard to make a large lens diffraction limited. The advantage of the DSLR is the sensor.
  - pmlnr 10y ago
    I wasn't suggesting a debate, rather stating a fact. Any sensor these days beats the sensor, for example, in a Nikon D100, yet the photos made with those machines with professional lens are still better than most the "really good" mini lens. Yes, you can mimic bits and pieces with software, sometimes, but in most cases there are decades of engineering behind the more expensive lens which will take a long while to be beaten. Diffraction is not the only issue you need to address in a lens, and not everything is easier to deal with when it's miniaturized. Besides, most "legendary" lens are actually loved for their mistakes and flaws in their optics.
    - StavrosK 10y ago
      How does any sensor beat the Nikon sensor? They clearly don't, since they are so small.
      - lorenzhs 10y ago
        Not GP, but the D100 was introduced in 2002. Sensor technology has progressed a lot since that day - noise reduction has become much better, so has autofocus, etc etc.
  - TeMPOraL 10y ago
    I think GP is referring to the ton of lenses you can buy and attach to your DSLR yourself. As for the sensor, if it's really that good, why nobody put one on a smartphone?
dharma1 10y ago
Would like to see raw samples at full resolution. Also wonder how fast the sensor will be - fast enough for video? At what resolution/fps?
- Spare_account 10y ago
  RAW would be an interesting challenge for this camera. As soon as the camera carries out the overlaying and merging of the sensor data you're no longer in RAW territory. Presumably the OS could store each sensor's RAW data along with sufficient metadata to allow post-processing software to do the stitching after the fact.
  - alkonaut 10y ago
    Since the article says the post processing can be done outside the camera on a separate computer or in camera, that's likely the case already.
cconcepts 10y ago
I found this article really engaging. His ability to explain the product and its guiding principles in laymans terms immediately destroys the many doubts that cropped up in my mind as I read. The lack of unneccessary jargon gave me an affinity to the product that I couldnt have got had he just appealed to the photophiles. I still have questions but I thought his clear communication style was quite compelling to someone like me who only has a cursory understanding of photography.
tambourine_man 10y ago
I think we can all agree that multiple camera/lenses and smart algorithms, essentially computational photography, is the way to beat the physics of the small sizes of our phones. However, the article has many words and not a single photo from the said camera, which I found disappointing
- joshvm 10y ago
  The pictures in the article are from the camera, perhaps you adblocked them? Here are some more: https://www.light.co/gallery https://www.light.co/gallery
  - tambourine_man 10y ago
    Thanks!
jfindley 10y ago
It's a really really interesting idea - I'm looking forward to see how it turns out. If it really is as good as claimed (and actually makes it to market), I'll almost certainly buy one. However, I'll be buying one in addition to my existing DSLR equipment, not as a replacement. The tech as presented in this article isn't really competing with DSLRs so much, as with m4/3 systems and things like the sony A7R series. The real reason for paying the size, weight and cost penalty of a DSLR these days is focus performance, near-zero lag and extreme flexibility - none of which this is going to be able to duplicate. For example, with my DSLR, 500mm lens and large tripod I can capture the exact moment a bird dives into a river to spear a fish, perfectly frozen in time. That's simply not possible with this system, and extremely difficult with m4/3 and other small-sized equivalents. So no, it's not going to outperform DSLRs. It might be a useful addition to your camera equipment, though.
- iwintermute 10y ago
  why it would be "extremely difficult with m4/3 and other small-sized equivalents" ?
  - lb1lf 10y ago
    Short answer - focus performance tends to leave a bit to be desired as cameras go smaller. (Though this can -for some subjects - be compensated for by focusing on the place the action is going to be in advance.) Also, long - as in really long - lenses are hard to come by in MFT. Most useful for wildlife.
    - iwintermute 10y ago
      There's also APS-C sensors, and even mirrorless full-frame
    - slantyyz 10y ago
      The often mocked Nikon 1 series is generally considered to have the best all-around PDAF of mirrorless cameras. It can track about the same as a DSLR (definitely faster than a 5D1, possibly faster than a 5D2). It has a weaker native lens selection than M43 but it has a 70-300mm lens that translates to 190-810mm (factoring 2.7 crop factor) that birders love. It can also use adapted F mount lenses with decently fast AF (center point only). On the other hand, the N1 series has so many other weaknesses (like being massively overpriced for what you're getting) that it's not a great system for most people.
roye 10y ago
strange that Lytro isn't mentioned anywhere, given they developed a product that works similarly.
- astrostl 10y ago
  Lytro made single-lens light-field cameras which let you refocus an image using their software. Very, very different. I say "made" because the company has since pivoted to VR cameras.
sathackr 10y ago
Just curious...what's the big deal with a DSLR? I really don't even see a need for the complexity of one anymore. The SLR camera was designed so that you could see the actual scene through the same lens that would be taking the picture. Mirror in place, light goes to viewfinder. Take a picture and the mirror moves so the light can go to the film instead. But on a digital camera, there is a live view of the scene on the screen on the camera. Who wants to put there eye to the tiny viewfinder when they have a 3" display of the same thing. Without the need for a viewfinder, there is no need for a moving mirror inside the camera. Just project the image directly onto the sensor. Do people with DSLRs still use the viewfinder? If so, why?
- lorenzhs 10y ago
  For some, it's a matter of preference. But there are good arguments for it - the viewfinder isn't limited by display resolution, for example. On the other hand, modern mirrorless cameras are marvellous, too. For example, the camera can bring your attention to areas that exceed the camera's dynamic range (and would thus be too bright / too dark), or highlight exactly which bits are in focus. So the digital versions are getting better, but there are still reasons to use an analogue viewfinder. It's not as clear-cut as you think. I encourage you to try it out!
- ccozan 10y ago
  It's the delay between the live image and the image on the screen. For certain situations ( like wild / nature photography / sports ) , this _is_ a limiting factor. For selfies, not so much.
- detaro 10y ago
  big deal with the DSLR: until very recently, if you wanted a large high-quality sensor you needed a DSLR (or medium format, but that's a totally different can of worms and level of pricing). They still are ahead in high-performance auto-focus systems and selection of really good lenses, but mirror-less cameras are getting closer and closer there as well. The vast majority of enthusiast and pro users uses the viewfinder regularly, if not exclusively. Professional mirror-less cameras basically have to have a (digital) viewfinder. * The 3" screen is a lot smaller compared to the image you can see through a viewfinder, and doesn't fill your field-of-view as much. Fixed with digital viewfinders. * thus, many people find it quite hard to get a good impression of the image with a screen. * The screens have limited resolution and delay. Resolution is especially a problem with digital view finders. * In many DSLRs, using the screen limits the auto-focus performance. * In bright conditions, the back-screens are hard to see. (Esp. if it is a touch screen with smudges on it ;)) Having both options is great, digital displays add a lot of features as well: focus-peaking, brighter image in low-light, easier to use the camera held high or close to the ground. Mirrorless cameras with digital viewfinders are becoming more and more popular, but the viewfinders won't be going away.
radiowave 10y ago
I wonder whether this technique could be adapted to produce a something like a digital equivalent of the Fuji/Hasselblad XPan camera, or a 617 camera (with an aspect ratio of about 3:1). The stumbling block with conventional sensors is that no one wants to commit to a production run of an odd shape of sensor.
mgleason_3 10y ago
Great idea, but did I miss the Results Section (data which shows it actually outperforms DSLRs)? "Are SLRs actually better?". The author talks about some of the ways. Noise, color range, accuracy, autofocus, lens selection and focal range are just a few of the ways. Take a picture of a sunset with both types of cameras and you'll immediately see the difference. Even on my iphone 6s the colors end up all wrong because the sensor gets overloaded. Reds come out orange. Bright orange areas come out yellow. You can see banding between colors. The photo looks very different from the sunset you see with your eyes. With our few years old canon DSLR it comes out much closer to what we actually see.

127 points by teklaperry 10y ago

TLDR; Phone cameras are not as good as SLRs because of tiny sensor, tiny lens and fixed focal length. How do we improve them then? Well, put as many as 16 different sensor+lens combinations in a small body, capture multiple views simultaneously and then algorithmically combine them to generate picture that looks as good as SLR. I think its cool idea and this is where smartphone industry is heading anyway, iPhone 7 being the first example. The key is algorithms to combine multiple views which can invariably introduce errors and artifacts for the trained eye. In extreme case, the entire back side of phone can be just rectangular matrix of 100s of lenses and sensor capturing the scene simultaneously and generate gigapixel image that can be viewed at variable focal length after the fact. That would be cool!

sohkamyung 10y ago
It sounds like a good idea. But it may not be suitable for all conditions where people use a DSLR. One example I can think of is macrophotography. In this case, the subject is simply too close to the camera to be covered by the numerous lenses of the Light camera. I'm also not sure if the mechanical mirrors can be moved fast enough for high-speed action photography; but it might be. Of course, it is a very specific example. For general usage, the Light may be a compelling replacement for a DSLR.
- 6nf 10y ago
  Yea the closer you get to the object, the worse the picture. But on the upside you start to get pretty good 3d scanning. With multiple views from different angles you'd be able to create detailed 3d scans of small objects. You may want to use this in video chat - each person's face will be rendered in high accuracy in 3D! Unfortunately the screen is on the other side of the cameras...
  - fwn 10y ago
    ..but at that price and with all the optical problems they seem to solve, a flipscreen is probably not that big of a deal to add.
- Tepix 10y ago
  For "normal" macro photographs, smartphone cameras are already doing an OK job. If you put a macro lens in front of your cellphone camera, it can be better. Phone cameras have issues elsewhere: * Low light exposures with low noise * low depth of field * telephotography
  - dharma1 10y ago
    I would add lower dynamic range and noise in general compared to dSLRs, because of small sensor/photosite size
- Obi_Juan_Kenobi 10y ago
  While this approach won't get you good macro performance, with some tweaks the DoF from small sensors could be a real boon. If you could pack a few of these close together, especially in a small package, there's a wonderful opportunity for wide-angle macro photography.
Yetanfou 10y ago
...iPhone 7 being the first example. No, iPhone 7 is certainly not the first example of a multi-camera phone. There have been several multi-sensor phones, this year both Huawei (P9) and LG (G5) were launched with such. In an article on these devices it is noted that 'Apple's iPhone 7 Plus is rumored to be following their lead later this year'. Correct, Apple did follow their lead, especially Huawei's as the P9 does just what is described in the article: it uses multiple sensors to increase image quality. http://www.theverge.com/2016/4/11/11406098/lg-g5-huawei-p9-two-camera-smartphone-trend-app http://www.theverge.com/2016/4/11/11406098/lg-g5-huawei-p9-t... Please don't perpetuate this myth of Apple being first with such and so. As a rule they are not first with anything, instead they see what others are doing and produce their own version of the idea. A refined version often, but that does not make it the first version.
- jahnu 10y ago
  They are often the first to remove things. #courage
- reitoei 10y ago
  There were actually a couple of cameras before the P9 and G5 using dual lenses (and I assume sensors) for doing weird 3D/DOF field stuff: http://www.pocket-lint.com/news/138719-dual-lens-smartphone-cameras-the-history-running-up-to-iphone-7-plus http://www.pocket-lint.com/news/138719-dual-lens-smartphone-... Your (and the most important) point about the iPhone 7 not being the first is bang on.
- glenndebacker 10y ago
  "Please don't perpetuate this myth of Apple being first with such and so. As a rule they are not first with anything, instead they see what others are doing and produce their own version of the idea. A refined version often, but that does not make it the first version." I really don't understand why people are getting so worked up of company X of Y being first or debunking Apple myths.... I personally couldn't care less (and I think most consumers) because I value a good working implementation before the fact that a brand is the first in anything. Why does that matter?
agumonkey 10y ago
I've been told about superresolution before, since, I'm waiting for smartphones to use it for pictures. The megapixel race produced such horrendous sampling and compression artifacts.
Mithaldu 10y ago
It's a technological Beholder, only misses the tentacles. :D

helloworld 10y ago

Our first-generation L16 camera will start reaching consumers early next year, for an initial retail price of $1,699. It's a cool concept, but I'm hoping that the second generation is a lot cheaper.

misev 10y ago
If it's equivalent in quality to a DSLR, especially in low-light, then it's definitely worth it.
- hyperbovine 10y ago
  Not so sure, for a lot less than $1699 you can buy some very good micro 4/3 cameras which would probably give this a run for its money in low light settings. micro 4/3 is not quite "tiny", but judging from the picture of the founder holding one up, neither us this (despite what the headline claims).
  - misev 10y ago
    The largest part of a micro 4/3 or a DSLR is the lens usually, and this thing goes around that, so I'd say it's seriously smaller. We can just speculate at this point about the low light performance though, I can't wait to see some proper review.
    - slantyyz 10y ago
      You can consider a fixed lens 1" sensor compact like a Sony RX100. It's got half-decent low light performance. It may not beat a 2016 consumer level DSLR for low light, but it's probably comparable to a 2008-2010 Rebel.
  - bobsil1 10y ago
    There's a mobility premium since this fits in your pocket. Much smaller than an A6300 with pancake lens. You could potentially jog with this in a wildlife refuge, then take bird shots.
  - 10y ago

nradov 10y ago

It's an impressive feat of engineering but I'm having trouble seeing the point. Does it actually take better photos than the latest generation of compact cameras that have legitimately good optics and sensors? I don't doubt that it's better than a cell phone camera, but if you're going to carry another device then why not bring one with a real zoom lens?

misev 10y ago
It claims to be as good as a DSLR.. which is better than a compact. We'll just have to wait for some actual tests and reviews that will substantiate the claims. I'm optimistic, even if the first version might not be there, this development seems to be going in the right direction with probably a lot of room for improvement. If it's only as good as a compact, then I'd think it's still a better option due to the selective DOF that it is supposedly able to do. But then it's definitely on the pricey side, so it's questionable if it's still worth it.
alkonaut 10y ago
This one is arugably a camera in the "latest generation of compact cameras". It competes with them since it's a separate device, and it competes with the highest end ones, such as the Sony Rx-series and the Fuji x100t.

ianpurton 10y ago

You buy a DSLR because it has a large sensor that can capture lots of light and detail. Lot's of smaller CMOS sensors when combined won't be able to capture the same level of detail. So I fail to see how this camera what outperform a DSLR.

sohkamyung 10y ago
It might, depending on the usage. The Light camera captures the same image using multiple sensors, which may work out to the same, or higher resolution to a single DSLR sensor. One interesting thing the multiple sensors on the Light can do, which a single DSLR sensor can't, is to capture the same scene with multiple exposure settings, as mentioned in the article. Combining these different exposures together may allow for higher contrast (HDR) images to be captured.
- dingaling 10y ago
  It's not resolution which is the limiting factor for small sensors; in fact many of them have higher per-sensor resolution than a DSLR. It's the fact that as you push more and smaller detector-sites into a such a small array you experience more and more noise, which means the image quality degrades rapidly with falling light. There's a balance between size of detector-sites and array density and any technology that reduces small sensor noise can equally be applied to large sensors with larger detector-sites in order to maintain their advantage. The best analogy I encountered was to consider each detector-site as a 'light-gathering bucket'. A big bucket can catch more photons and hence has a higher SNR than a small bucket once you deduct cosmic rays and read-out errors etc. So a medium-format 21MP is better than an APS-C 21MP array which is better than a phone-sized 21MP array. It doesn't really matter how many small-arrays you point at one target because it's the SNR at the individual detector-sites that matters. ~~ TLDR: this camera can never compete directly with a DSLR on raw photo quality but will offer other interesting capabilities that DSLRs can't.
  - mnw21cam 10y ago
    Noise is much more noticeable in dark areas of an image, so having a method to produce a HDR image will naturally improve the noise. So, I think this camera could quite easily compete directly with a DSLR on photo quality - but a good DSLR will easily be better in low light conditions.

misev 10y ago

This is exciting development, going in just the right direction. Divide and conquer! If the first prototype proves to be at least close to DSLR image quality, it will be a revolution.

pmlnr 10y ago

people,it's not the sensor why a dslr is the etalon, it's the lens you can put on them.

mnw21cam 10y ago
Not so. It's easy to make a really good small lens - the article mentions that the modern smartphone lenses are dirt cheap and diffraction limited. It's really hard to make a large lens diffraction limited. The advantage of the DSLR is the sensor.
- pmlnr 10y ago
  I wasn't suggesting a debate, rather stating a fact. Any sensor these days beats the sensor, for example, in a Nikon D100, yet the photos made with those machines with professional lens are still better than most the "really good" mini lens. Yes, you can mimic bits and pieces with software, sometimes, but in most cases there are decades of engineering behind the more expensive lens which will take a long while to be beaten. Diffraction is not the only issue you need to address in a lens, and not everything is easier to deal with when it's miniaturized. Besides, most "legendary" lens are actually loved for their mistakes and flaws in their optics.
  - StavrosK 10y ago
    How does any sensor beat the Nikon sensor? They clearly don't, since they are so small.
    - lorenzhs 10y ago
      Not GP, but the D100 was introduced in 2002. Sensor technology has progressed a lot since that day - noise reduction has become much better, so has autofocus, etc etc.
- TeMPOraL 10y ago
  I think GP is referring to the ton of lenses you can buy and attach to your DSLR yourself. As for the sensor, if it's really that good, why nobody put one on a smartphone?

dharma1 10y ago

Would like to see raw samples at full resolution. Also wonder how fast the sensor will be - fast enough for video? At what resolution/fps?

Spare_account 10y ago
RAW would be an interesting challenge for this camera. As soon as the camera carries out the overlaying and merging of the sensor data you're no longer in RAW territory. Presumably the OS could store each sensor's RAW data along with sufficient metadata to allow post-processing software to do the stitching after the fact.
- alkonaut 10y ago
  Since the article says the post processing can be done outside the camera on a separate computer or in camera, that's likely the case already.

cconcepts 10y ago

I found this article really engaging. His ability to explain the product and its guiding principles in laymans terms immediately destroys the many doubts that cropped up in my mind as I read. The lack of unneccessary jargon gave me an affinity to the product that I couldnt have got had he just appealed to the photophiles. I still have questions but I thought his clear communication style was quite compelling to someone like me who only has a cursory understanding of photography.

tambourine_man 10y ago

I think we can all agree that multiple camera/lenses and smart algorithms, essentially computational photography, is the way to beat the physics of the small sizes of our phones. However, the article has many words and not a single photo from the said camera, which I found disappointing

joshvm 10y ago
The pictures in the article are from the camera, perhaps you adblocked them? Here are some more: https://www.light.co/gallery https://www.light.co/gallery
- tambourine_man 10y ago
  Thanks!

jfindley 10y ago

It's a really really interesting idea - I'm looking forward to see how it turns out. If it really is as good as claimed (and actually makes it to market), I'll almost certainly buy one. However, I'll be buying one in addition to my existing DSLR equipment, not as a replacement. The tech as presented in this article isn't really competing with DSLRs so much, as with m4/3 systems and things like the sony A7R series. The real reason for paying the size, weight and cost penalty of a DSLR these days is focus performance, near-zero lag and extreme flexibility - none of which this is going to be able to duplicate. For example, with my DSLR, 500mm lens and large tripod I can capture the exact moment a bird dives into a river to spear a fish, perfectly frozen in time. That's simply not possible with this system, and extremely difficult with m4/3 and other small-sized equivalents. So no, it's not going to outperform DSLRs. It might be a useful addition to your camera equipment, though.

iwintermute 10y ago
why it would be "extremely difficult with m4/3 and other small-sized equivalents" ?
- lb1lf 10y ago
  Short answer - focus performance tends to leave a bit to be desired as cameras go smaller. (Though this can -for some subjects - be compensated for by focusing on the place the action is going to be in advance.) Also, long - as in really long - lenses are hard to come by in MFT. Most useful for wildlife.
  - iwintermute 10y ago
    There's also APS-C sensors, and even mirrorless full-frame
  - slantyyz 10y ago
    The often mocked Nikon 1 series is generally considered to have the best all-around PDAF of mirrorless cameras. It can track about the same as a DSLR (definitely faster than a 5D1, possibly faster than a 5D2). It has a weaker native lens selection than M43 but it has a 70-300mm lens that translates to 190-810mm (factoring 2.7 crop factor) that birders love. It can also use adapted F mount lenses with decently fast AF (center point only). On the other hand, the N1 series has so many other weaknesses (like being massively overpriced for what you're getting) that it's not a great system for most people.

roye 10y ago

strange that Lytro isn't mentioned anywhere, given they developed a product that works similarly.

astrostl 10y ago
Lytro made single-lens light-field cameras which let you refocus an image using their software. Very, very different. I say "made" because the company has since pivoted to VR cameras.

sathackr 10y ago

Just curious...what's the big deal with a DSLR? I really don't even see a need for the complexity of one anymore. The SLR camera was designed so that you could see the actual scene through the same lens that would be taking the picture. Mirror in place, light goes to viewfinder. Take a picture and the mirror moves so the light can go to the film instead. But on a digital camera, there is a live view of the scene on the screen on the camera. Who wants to put there eye to the tiny viewfinder when they have a 3" display of the same thing. Without the need for a viewfinder, there is no need for a moving mirror inside the camera. Just project the image directly onto the sensor. Do people with DSLRs still use the viewfinder? If so, why?

lorenzhs 10y ago
For some, it's a matter of preference. But there are good arguments for it - the viewfinder isn't limited by display resolution, for example. On the other hand, modern mirrorless cameras are marvellous, too. For example, the camera can bring your attention to areas that exceed the camera's dynamic range (and would thus be too bright / too dark), or highlight exactly which bits are in focus. So the digital versions are getting better, but there are still reasons to use an analogue viewfinder. It's not as clear-cut as you think. I encourage you to try it out!
ccozan 10y ago
It's the delay between the live image and the image on the screen. For certain situations ( like wild / nature photography / sports ) , this _is_ a limiting factor. For selfies, not so much.
detaro 10y ago
big deal with the DSLR: until very recently, if you wanted a large high-quality sensor you needed a DSLR (or medium format, but that's a totally different can of worms and level of pricing). They still are ahead in high-performance auto-focus systems and selection of really good lenses, but mirror-less cameras are getting closer and closer there as well. The vast majority of enthusiast and pro users uses the viewfinder regularly, if not exclusively. Professional mirror-less cameras basically have to have a (digital) viewfinder. * The 3" screen is a lot smaller compared to the image you can see through a viewfinder, and doesn't fill your field-of-view as much. Fixed with digital viewfinders. * thus, many people find it quite hard to get a good impression of the image with a screen. * The screens have limited resolution and delay. Resolution is especially a problem with digital view finders. * In many DSLRs, using the screen limits the auto-focus performance. * In bright conditions, the back-screens are hard to see. (Esp. if it is a touch screen with smudges on it ;)) Having both options is great, digital displays add a lot of features as well: focus-peaking, brighter image in low-light, easier to use the camera held high or close to the ground. Mirrorless cameras with digital viewfinders are becoming more and more popular, but the viewfinders won't be going away.

radiowave 10y ago

I wonder whether this technique could be adapted to produce a something like a digital equivalent of the Fuji/Hasselblad XPan camera, or a 617 camera (with an aspect ratio of about 3:1). The stumbling block with conventional sensors is that no one wants to commit to a production run of an odd shape of sensor.

mgleason_3 10y ago

Great idea, but did I miss the Results Section (data which shows it actually outperforms DSLRs)? "Are SLRs actually better?". The author talks about some of the ways. Noise, color range, accuracy, autofocus, lens selection and focal range are just a few of the ways. Take a picture of a sunset with both types of cameras and you'll immediately see the difference. Even on my iphone 6s the colors end up all wrong because the sensor gets overloaded. Reds come out orange. Bright orange areas come out yellow. You can see banding between colors. The photo looks very different from the sunset you see with your eyes. With our few years old canon DSLR it comes out much closer to what we actually see.