Friday, July 14, 2017

Light Co. Starts Shipping its Computational Camera

Light starts shipping the L16 camera to its pre-order customers, after 4 years of development. The company explains what it has accomplished in these 4 years:
  • It took us three years to design and build our own custom ASIC chips, which are needed to control all 16 camera modules at the same time.
  • We also developed our own 70mm and 150mm camera modules, complete with custom optics and electrical components. To put this in perspective, most smartphone cameras contain 30mm or 50mm lenses. The higher focal length lenses we were looking for weren’t even on the market yet, so we had to invent them ourselves.
  • We created proprietary image-fusing algorithms and processing pipelines that align each of the base camera modules.
  • We produced Android software to operate our camera and a Mac/Windows application for depth-of-field editing.
  • We implemented an e-commerce platform and initiated a complex global manufacturing and supply chain.

Light Director of Hardware Engineering, Brian Gilbert,
with the first 'lunch box' prototype
Light final product. Each lens is annotated
with its range of distances, focal length, and aperture

14 comments:

  1. We did all this and for what you ask? To enter the market for "second camera" because everyone out there is just dying to have a second camera with them and price is no concern. We are going to steal the market from all those P&S cameras, and those DSLRs and... wait. What was that? That market is dead? Pity. Such an impressive complicated thing. So much sincere effort. So much investors' money.

    ReplyDelete
  2. Curiously dismissive of one of the most innovative ideas in photography in the last few decades.
    Since the Majority of photos in tbe world are taken with smartphones, why not ensure that those photos are some of the finest out these.
    Wbile I love the superb DSLRs and DSLMs out there, none of them are seriously pocketable. And why lug huge pieces of equipment that refuse to get more portable with time. Also, if this thing even shoots nearly as good at entry level DSLRs, with tbe various focal length and sticking options, it does a lot of what was promised.

    The photo below highlights how far they have come. This photo highlights each single hair on the model, and if it is done by merging the various photos from the various cameras, then it is indeed an engineering marvel.

    https://spot.light.co/content/images/2017/07/SuperBloom_D-0790_L16_00070-Edit-2.jpg

    This will move into commercial production, and be picked up for higher end smartphones to begin with (Samsung S9 or iPhone 9?).
    Hopefully within a few years we all have smartphones that actually rival DSLR and DSLM camera photos, without the heft.

    ReplyDelete
    Replies
    1. There are many engineering marvels that took years to sort out and bring to market. Unfortunately technical excellence and innovation are not sufficient for market success. I am sure there will be near-term success as pre-order innovators and some early adopters try out the device. Crossing the chasm, as it is called,to further early adopters and more mainstream is something I am personally skeptical about. Incorporation of more than a few cameras (sub-cameras) in a smartphone is also unlikely just from ergonomic considerations, and also from cost. But, we will see. I felt exactly the same way about Lytro and was sorry to be right. I do think the value proposition of Light Co. is weak, given the current market trajectory and velocity.

      Delete
    2. The photo is interesting. The micro contrast is really good, capturing the details of the face very well. BUT the resolution is not, the hair details are missing showing as thick strands. Any entry lvl mirrorless or DSLR w a decent lens can beat that. The bokeh is also good.

      Delete
  3. To compare the Light Camera with Lytro is uninformed and misguided. Maybe you could share Details about the Lytro's Pricing and Actual Mega Pixel Resolution of each photo. Also, while you are at it, could you be generous enough to do a physical dimensions comparison between the two, as well.

    Thank you.

    ReplyDelete
    Replies
    1. Not misinformed nor misguided. New computationally based photographic system combining multiple sets of pixels to do something that could not be done before in a consumer camera, in a new form factor, in the same market space - sounds comparable to me. But don't let me rain on your parade. The market will decide, in the end. To me, the Light Co. device seems like something from a Rube Goldberg cartoon, except it has been reduced to practice and seems to function as hoped. I look forward to reading reviews from real users.

      Delete
  4. Distracting people by placing a nice photo model is one thing, but are there "hard" measured facts in comparison to a DSLR? For example, how can it compete with current DSLR's MTF? And I mean a pretty good DSLR combination, say a A9 + a Zeiss Batis???
    Today there's no room left for compromises!
    -dkf

    ReplyDelete
    Replies
    1. The notion of matching of exceeding "DSLR performance" seems rather problematic in itself. The value function is multi-dimensional - on which dimensions would we be exceeding the DSLR performance for the comparison to be meaningful, and which of the broad spectrum of "DSLR performance" products would we be using for comparison. An entry-level little DSLR, an A9, a D810? Which of their features really define "DSRL performance", given that the DSLR class is so heterogeneous? Is it interchangeable lenses? Dynamic Range? Resolution? MTF? Low light? 3A accuracy and speed? Prestige? Impressive look? Number of shots per battery? Flash sync options? Optical zoom? FPS? Extra-long exposures?

      Smartphones actually match or beat many DSLRs on selected dimensions of the merit function. Resolution is there. MTF is diffraction limited or almost there. They can shoot full resolution at video rates, at least in a burst. With a bit of computation (like HDR+) they hold their own in low light and dynamic range. They are pretty much good enough for the 99% of their users. This a tricky point. When a technology or a product becomes good enough adding "improvements" can in fact produce negative marginal utility as the cost increases do not even produce a matching value improvement. This not to say that there is no room for further improvements, but they will have to be introduced carefully. Even a simple dual camera configuration has been a hit and miss story so far, and it is not clear which way this will go.

      I think Eric is absolutely correct comparing Light to Lytro. For me the fundamental similarity, and I believe the pitfall, is the fact that the proposition here is a second camera. This thing is not replacing anything , except maybe (at best) DSLRs, which are a fairly small and declining market, and which will retain their demand floor for a number of reasons. Neither Light nor Lytro could hope to have smartphone users give up those devices, and with them come perfectly adequate casual-use cameras. So both Lytro and Light would have to carve out a space in users' lives for a second camera, and an expensive one at that. That's unlikely to happen, as we see the opposite trend killing P&S cameras - many of them truly capable and quite pocketable, but still unable to retain their presence in people's daily lives.

      As for the notions that 16 folded cameras will someday soon squeeze into smartphones, I would suggest guys who believe that go and open a modern phone or two. Or look up a teardown. Not going to happen.

      Delete
    2. I totally agree with you. And I'd like to extend my claim beyond your list: such a product should match _all_ of those aspects.
      And that can become pretty hard, even with a low-range DSLR set, and even if you accept some (hopefully imperceptible) trade-offs due to reduced size...
      Well, to go with Eric: Let's wait for the market to decide. I always feel excited about unexpected success stories. Good luck, Light!

      Delete
    3. I don't agree with the above post that "With a bit of computation (like HDR+) they hold their own in low light and dynamic range". Low light performance of smartphones is far poorer than of DSLRs. Smartphone performance is quite good in good lighting but falls off a cliff even in evening outdoor lighting.

      Delete
    4. This used to be the case, but it is changing rapidly. Is your comment based on experience with current Pixel phones? Thing are moving fast and the gap is closing in interesting ways. Smartphones are moving to RAW, burst fusion, other techniques. Some discrepancies will remain, but will they matter in practice? Will they warrant paying big bucks for and carrying around a second camera, for more than a small hard core sliver of the users?

      Look at this, admittedly a bit out-there example.

      https://research.googleblog.com/2017/04/experimental-nighttime-photography-with.html

      Delete
    5. Thanks for the link. It's nice to see that you put your new hopes in a more than a decade-old technology:

      http://image-sensors-world.blogspot.com/2005/11/electronic-still-image-stabilization.html

      Delete
  5. All too often we are guided by what we think is important or desired. Lets face it if you're reading this blog you have an enthusiasm for technology. As an exercise ask your significant other how much they need a device like this. If they're excited then Light is on to something but I have a feeling that's not likely. My wife was one of the first people to buy a digital DSLR some 15 years ago and it just happened out of nowhere (I didn't even know why). The market exploded shortly after that. The same thing happened with smartphones. This however was met with "meh, that's nice I suppose" so I think Eric will be right on this one.

    ReplyDelete
    Replies
    1. Underlying much of our attitudes is the assumption that improvements will go on forever, at a sustained rate, that there is an elastic demand and appetite for more and more and more. That is not necessarily true, life is not linear, and the marginal utility of things changes over their lifetime, development, and market penetration. Going from no running water to running water, big, huge deal. Everybody who is anybody wants that. Adding a water heater to it, still important. What do you do after that? Softeners, bringing the heater to the point of use so you have hot water sooner, electronic faucets, etc... nice, shiny, even worth it for some, but really, no comparison to that initial gain. Similar demand curves tend to exist in other areas. At some point stuff gets just good enough, and even real improvements tend to be perceived as not worth the cost or the hassle (which really is just another cost). And you end up with those anonymous tight-fisted bastards in the store, who will look at your doodad, say "nice" ("nice", you bastard? It's POSITIVELY FABULOUS and look how much effort it took to make it happen! Do you even have an idea? 16 cameras, flipping mirrors, piezoelectric actuators, 47 calibration steps, endless nights away from the TV, millions of investor's dollars), put it back on the shelf, and be gone, taking their $1000 with them. The $1000 you so deserved. Like they were never there. Problem is, people just have too many things going on to be as crazy about your particular thing as you are. Plus, their wealth doesn't depend on them not seeing that they can do just fine without it - quite the opposite, really, and they eventually learn that lesson. So their judgment is harder to cloud than yours, on this particular subject at least.

      One way to deal with the "problem" is with forced deployment - phone cameras will be getting better in subsequent generations not because customers are pulling for these improvements, but because manufacturers have their own reasons to push them, and absorb most of the cost. People will take free improvement, why not, especially when there is no easy choice not to. But ask them to buy another separate camera. Go ahead. I want to see this unfold. I'd be perfectly happy to be proven wrong, there would be some learning in such an outcome. But it is unfortunately quite unlikely.

      Delete

All comments are moderated to avoid spam.