The Light L16 – a multi-aperture computational camera | Hacker News Reader

The Light L16 – a multi-aperture computational camera

light.co

161 pointsderwiki111 comments

Constructive feedback about the video and website (if the founders are reading)... I'm probably not the target demographic since I don't know much about cameras, but the video did not quickly explain to me why this camera was so special aside from me seeing multiple lenses; I would have just turned it off after the 15 second mark (even if I saw this on Product Hunt) if this wasn't posted on HN. It felt like the video didn't really convey anything and I couldn't help but feel that it was a waste of my time watching it to the end. Of course it could just be me.

If only this was front and center on their webpage (it took me a really long time to find this):

"With 16 individual cameras, 10 of them firing simultaneously, the L16 captures the detail of your shot at multiple fixed focal lengths. Then the images are computationally fused to create an incredible high-quality final image with up to 52 megapixel resolution."

So you can fit a camera with the power of a DSLR in your pocket now, while having the same ease of use as a smart phone?

As a quick fix: drop the video and fix your site so that you can explain why your camera is so special at the top of the page. Use the image gallery as quick proof right after the explanation.

Or to be blunt: It's about 1:30 long, with 1:29 of that people taking photos. There's about a second of a guy tweaking parameters on a computer to change the aperture, etc, in post processing.

Also agree. needs more "meat". maybe add 30 seconds in the middle actually explaining the USP of this thing.

It's not just you. The video, while beautiful, is completely superfluous and distracting from the target market. In the video, nice people use a cool smartphone to take pictures. One edits them on their computer. People are enjoying what they are doing. I had to guess, based on the fact that the camera appears to have multiple (independent?) lenses, that some kind of computational method was combining the image data, and then returning to the site revealed that this is basically the case.

Many questions remain unanswered. Will I be able to fully edit the photos on my phone, or do I need to get a Mac and some custom software to really take advantage of this system? Will the software on the device be open? I'd like insurance against the unfortunate case that you make a nice device but then go out of business. Will the battery be replaceable? A non-replaceable battery in a $1600 handheld device is a really nasty design decision.

In short, really nice, and I would love to drop $200 right now to reserve one, but I'm afraid that I won't be able to undo the reservation if it turns out that the feature set of the phone doesn't match up with my practical needs. Marketing: you are losing sales right now. Go back to the beginning and explain to us why this is a good device.

In addition, neither the video nor the FAQs clearly explain the basic technical details that users look for in a device

The other video, at the bottom of /camera is much better: https://vimeo.com/141273968

Wow, that is an order of magnitude more interesting.

This video is much, much more engaging, explanatory and human.

This should come first.

And will I be able to replace the battery? ;)

I had the same reaction. My assumption was that this camera allows for variable depth of field in post production, but virtually nothing about the video says this. They showed someone tweaking a photo but the shot was too short and subtle to notice.

Personally I'm completely turned off by the cellphone form factor, its possibly the worst possible way to hold something steady to take a photograph.

Some raw (is there an equivalent here?) or high res samples would be greatly appreciated in exchange for my deposit.

I had an interesting reaction reading this: I'm totally sold on the idea and science behind this. We have large scale radio telescopes, why not cameras?

But, I'm super turned off by the proposed plan: modified android OS combined with the promo photos makes me think viewfinder lag is going to be a bitch. v1 seems likely to feel like a janky short-run phone from a new manufacturer. Which is too bad, because the concept and possibilities are amazing.

Light; the articles mentioned a partnership with FoxConn; why not work with them or Samsung to put one out that you just license tech for?

And, it would be cool to see some high res samples.

Probably will work similar to Lytro: https://pictures.lytro.com/

Lytro is a light field camera, I'm pretty sure the Light will NOT work similar to a Lytro.

Light is fusing (at capture time I think) a few normal-sized sensors each using different exposure duration, focal length, etc.

On the other hand Lytro collects a lot of tiny images each at the same exposure duration and focal length, but very close to each other. It's like having very big pixels but for each pixel you retain the color response from a large number of directions, instead of just the average color for that particular pixel, as in a regular camera.

Continuous refocusing can be performed after the fact in a Lytro, not possible (or very limited) with the Light. But the Light should be able to do native HDR in one take (and real time HDR video maybe), not possible with a Lytro.

If you break it out of the gallery, you can get a little better look. http://light.co/content/2-gallery/gallery_modal_13.jpg http://light.co/content/2-gallery/gallery_modal_14.jpg

Pretty impressive.

I liked that first picture, too. And on closer inspection it's interesting -- some different color temperatures in different spots. I wonder if you get a natural HDR type capability using different sensors like this. The second one makes me think low light should be reasonably good, which is what I would hope for considering the technology description.

Yeah, there does seem to be a little uncanniness to the lighting a la hdr. They feel painted. I'm curious how much post-processing went into each of these.

> Yeah, there does seem to be a little uncanniness to the lighting a la hdr.

It is HDR. Multiple exposures means you can capture the full dynamic range in one "shot" and synthesize together the nicely exposed parts of each image.

> I'm curious how much post-processing went into each of these.

I mean, a bunch, but the camera's firmware is doing it. I'm sure typical punch up postprocessing happened for publication, still the interesting dynamic range stuff is happening in the camera.

Is there any reason you couldn't do a single-shot HDR with a single sensor, by modulating the sense amplifier gain on a per-pixel (or per-row, or per-column) basis? The ISO sensitivity of a digital camera is just a gain coefficient in some circuit, so it seems like it could be controlled during the readout.

Composite HDR shots are supposed to be done such that the shutter at the fastest speed properly exposes the highlights and the slowest speed completely blows out on the highlights, but the extremely dark shadows are properly exposed. In other words, it's about filling in the details when overexposure causes amplitude clipping.

How most people end up shooting HDR, however, the component images mostly all have full detail in the shadows and highlights, meaning the final image really not much more than you've described. You can do that with a single shot and a contrast slider.

About the low light... You might want to take a look at this image as well: https://light.co/content/2-gallery/rtalt6.jpg

That's insane how well the faces are exposed given the dynamic range of the light in that shot. With an SLR the correct exposure is like walking a tightrope. Definite postprocessing required to get a result near the L16 image.

with lightroom, you'll get this by moving few sliders left or right in 10 seconds, given pic is coming from half-decent dslr, full frame ideally.

all promotional photos anywhere are carefully photoshopped anyway.

That doesn't seem insane at all. There's not much light from the sun that has already set, there's light from the bulbs above and there looks to be light coming from behind the photographer.

Also I see extreme amounts of noise even though it's downsampled to 1920 on the long side, especially in the black guys shirt. There seems to be lots of lost detail and and also loss of information in the shadows. Of course it would help to have full res images to say something substantial, but it doesn't look all that promising to me.

I don't see the problem with exposing that on an SLR. Just meter for the faces? It looks just like any noisy high-ISO SLR image to me?

Any modern full-frame Canon with the ol' "nifty fifty" (cheap fixed 50mm, 1.8f lens) could get a shot like this. I've taken portraits in-doors in candlelight with mine.

It took me a few looks at this picture to pick up on the fact that it is all in focus, whereas the depth of field of the f/1.8 lens is going to give you one of the subjects' eyes in focus.

I was a little disappointed in the amount of noise and the noise reduction in this pic until I started thinking about how it would look with other cameras. Probably a blurry, noisy mess with cellphone or compacts. To get the same effect as this pic with a DSLR, you'd have to massively stop down that 50mm lens and would end up boosting the ISO so high that even a modern full-frame DSLR wouldn't be that much better.

You won't get both the front and the back of the table in focus, but given how far back you'd have to sit to get all three people in frame, you'd probably be able to get all three people in focus, especially if the guy in the back leans in a little. Also, I mentioned Canon because their high ISO modes are particularly low noise these days, even before noise reduction. And Nikon might have caught up as well, I haven't reviewed lately.

But that's actually not the most important part.

The most important part is that there are only two cameras in my life: my "good" camera and my smartphone, aka the camera I use when I know I'm going to be taking photos, and the camera I have on me all the time.

Since this image is probably boosted and noise reduced and photoshopped to make it the most presentable representation of the product, I really think the same amount of effort would get you better results, for less money, out of a DSLR. If you're into spending money, the right DSLR and lens combo would get you far, far better results.

On the other hand, yes, DSLRs are bulky and you don't want to carry them everywhere. But the Light is not a smartphone, so it's yet another device to have to remember and carry around, so while it's smaller, I'm still likely to leave it home because I just don't think to bring cameras with me everywhere. I've got an older smartphone (Galaxy Note 3) that can take photos pretty near this quality in its fakey HDR mode. I've seen some shots out of the newest iPhone that very much rival this sort of photo. And that iPhone is cheaper than the Light, plus it's not an extra device.

That's why it's most interesting as a concept and not a product. I'd really like my next iPhone (the iPhone 8S if if past shopping behaviour is any indication) to have 3-4 lenses and sensors, if it meant it could be more sensitive in low light or use different focal lengths in good light, and still not be bigger than my current iPhone is. So this company should just make a showcase product, and then wait to be bought by sony/apple/samsung.

This looks like something I could do at iso1600 with my old APS-C camera, but then I'd need to do it wide open which would severly limit the DOF. With a modern full frame one could probably pull this off with iso 12800 and just stop down accordingly, without seeing much more noise. I agree it's still impressive to even push the modern FF to it's limits to match though (but comparison needs to be done at full resolution, not at screen res)

it looks a lot like HDR on smart-phone grade sensors

the advantage here is that the HDR frames can (presumably) be captured simultaneously, reducing motion blur

plus they seem to suggest some Lytro style ability to change the focus/depth of field after taking the photo

Agreed. They've got a fascinatingly lifelike, "through your own eyes" look. Which to me is more interesting since our perception of sight is actually a mental re-representation, not a raw view of the data streaming in from your retinas at a given moment. So, perhaps somewhat analogous to multi-camera/aperture shots recombined in software.

I'm looking at http://light.co/content/2-gallery/gallery_modal_13.jpg. Can you tell me exactly what looks impressive about that? The leafs outside seems to be more in focus that the little guys eyes. They even had the chance to set focus after capture, so I don't see why this should be happening.

I see noise even at this resolution. I also see really bad highlight clipping (the "hitting a wall" type of clipping) and purple fringing like color shifts in the transparent plastic right above the vinyl.

To see something interesting, take a look at the gray box in the lower left corner. In the absolute corner there's "increased resolution", and then to the right of this there's a sharp line where a more smudged area takes over. And no, that is not beacause of shallow DOF.

> Some raw (is there an equivalent here?) or high res samples would be greatly appreciated in exchange for my deposit.

Yes, supposedly it outputs (or will output) raw/mosaiced DNGs[1]. I think it will be some time before we'll have access to these though. There's about a year until release and even the files available (downsampled to 1920 on the long side) looks noisy and a bit smeared (probably from noise reduction).

[1] http://support.light.co/hc/en-us/articles/205746673-What-ima...

There was a pretty good interview with their CTO a few months ago on Imaging Resource. Much more technical than what you see on the site, also gives a glimpse on what they were thinking when they were developing this.

http://www.imaging-resource.com/news/2015/04/27/light-interv...

Personally I think the money shot is the synthetic DoF simulation on the skater:

https://light.co/content/2-gallery/gallery_modal_11.jpg https://light.co/content/2-gallery/gallery_modal_06.jpg

The subject isolation properties is a pretty big reason why people still bother with large-sensor cameras. As an amaterur, I don't have big demands for image quality, so I only bring my D60 or F4 when I want subject isolation. Otherwise I'm pretty happy with my smartphone as a camera (Lumia 930).

I too am interested in this for shallow dof, and was excited at the idea that these lenses are effectively f1.2. But there's no way that skater shot is f1.2-equivalent. In fact, there are no sample shots anywhere I can see that back up that claim.

In some of the photos the bokeh looks right weird. But for a version 1 of a product I think this is really cool.

Care to point out specific examples? Curious what you see, exactly

https://light.co/content/2-gallery/gallery_image_2.jpg

checkout the white outline around the knight, it has a halo. Then the fuzzy bokeh looks stepped somehow.

100% agreed

No, that's not like f/1.2 on a full-frame camera. But you have to consider the sensor size, and thus multiply by the crop factor. Most photographers know this about focal length (50mm on full-frame becomes a 100mm lens on micro four thirds), but it seems people forget that the same goes for aperture. f/1.2 on a full-frame becomes a f/2.4 on m43, so you'll need an even bigger (smaller number) aperture to have the same effect on a smaller sensor.

I'm guessing these sensors are way smaller than m43 even, so that f/1.2 won't produce anything near the stunning bokeh it would on a full-frame camera.

Yes, though if that's their angle the marketing line is very misleading: they're discussing focal length as 35mm-equivalent, I had assumed they were doing the same with aperture.

Actually once you have information on depth (which isbwhat they get from multiple camera perspectives), and your bas image at near-infinite DoF you can pretty much dial your depth of field however you want. Actually your next limit is how accurately you can measure depth, and that is limited by the distances between the individual sensors.. With a longer baseline you could do pretty thin focused slices, and blur the rest.

The multi-aperture thing isn't what makes this interesting. What is interesting is that it takes a bunch of little sensors and combines them into a single better image with software.

Because a lot of work has been done into making very good sensors and optics for phones, this approach does make some sense. It also lets you have more sensor surface area without the bulky optics. I'm pretty sure that there are telescope arrays that work in a similar way.

I'm not about to replace my cameras with these, but it's got potential.

There's more information in http://www.theverge.com/2015/10/7/9473793/light-l16-16-lens-... and http://www.technologyreview.com/news/542121/a-high-end-camer... .

The multi aperture thing doesn't make much sense, it appears to have multiple lenses which will have debilitating parallax error due to the positioning. I don't see how that's possible. I'm also struggling to see why they laid out the sensors in such a non geometric pattern too.

Parallax could be corrected by software as long as the subject isn't too close. You might lose a little detail around the edges but with enough redundancy it's probably not a big deal

In their examples there's a macro photo of a chess board.

Yeah, it does look a little funky around the edges of the pieces in the background, but surprisingly not the foreground: https://www.dropbox.com/s/o4rx62fjo63nvk4/Screenshot%202015-...

https://light.co/content/2-gallery/gallery_modal_05.jpg

You're right, that's like no lens blur I've ever seen.

It's a lot like an unsharp-mask halo though

Right. Also surprising how much fine noise there is in the background — it ‘should be’ out of focus enough there there would be nearly no noise at all…

If you look closely at that photo, the result is really weird. The black pieces behind the plane of focus have very uncertain positions that the software didn't bother fixing. It's bokeh, in a way. A horrible way.

Yup, it's not nice. Headache inducing.

I think that's just a zoomed up picture with low depth of field. I'm not a photographer so correct me if I'm wrong. The chess pieces in the background look still quite big compared to the pieces in the front. It suggests that the photo was taken from around 30-40 cm.

I wonder what the pixel density on these sensors is like.

Because even if you have a bunch of small sensors combined with software, they will still be noisy, and low light performance will be bad if they are standard small sensors

For instance, I wouldn't ditch my dSLR for this - http://light.co/content/2-gallery/gallery_modal_02.jpg

I don't see them competing with SLRs. You'll need a lot of small sensors to capture as much light as a one big SLR sensor.

However, it's impossible to put larger sensors in thin phones because of the size of the optics, so putting a few small sensors and doing magic in software is a way to overcome this. I can believe that we'll be seeing this technology coming to phones in the not to distant future.

I don't know if it was intentional, or necessitated by the internal mechanics, but I find the layout of lenses to be visually unpleasant. It looks like someone took a camera and gave it chicken pox...

Aside from that, this is a really cool idea! But I don't know how it's being positioned. Is it a point and shoot replacement? Or a mirrorless camera replacement?

Or maybe it's just a technology demonstrator - this sort of system, suitably minimized, would make an excellent smartphone camera.

At the asking price ($1,700) it's better be a DSLR replacement. Perhaps it can find its way into a smartphone one day, who knows.

At 53 MP and three sets of useful primes, it seems like a dSLR replacement to me. I'm pretty excited, but I don't want to pay even the preorder price for first-generation technology like this without a long breaking-in period.

The Verge's review does say that it compares well with the 5DIII in detail. It also says that the Light's photos "weren't as punchy" -- not sure what that means. Saturation or contrast? I'll wait on a good comparison review.

> I don't want to pay even the preorder price for first-generation technology like this without a long breaking-in period.

It's not even a pre-order price. It's a crowdfunding contribution, for which the company is not obliged to give you anything in return.

Hmm. I was going from the re/code article (http://recode.net/2015/10/07/photography-startup-light-takes...).

> The company will take preorders, with a limited number of units available for $1,299, until Nov. 6.

The reservation form on the actual site is the same way. Why do you call it crowdfunding?

Because they're collecting the funds using Tilt.com, a crowdfunding service. You can also read their terms and conditions when you click on the Reserve button, where you'll find that you're "contributing" funds to an "organiser".

From Tilt's Terms:

"If you’re a Contributor, you make payments to an Organizer at your own risk. Unforeseen or other events may prevent an Organizer from providing goods, services or rewards exactly as promised. If things go awry, please be understanding and consider cutting the Organizer some slack. At the end of the day, though, it’s the Organizer’s responsibility, not Tilt’s."

see: https://news.ycombinator.com/item?id=10351199 for definition of punchy

53 MP that barely makes ok pictures at 2 MP after all the computation

I'm curious if the asymmetry has anything to do with the "computational" component, e.g. in solving a system of equations (like estimating distance across multiple sensors), it's best to have an independent basis. If the sensors were arranged in a perfect grid pattern, then comparing some sets of sensors (i.e. colinear ones) wouldn't give you as much data as others?

...or perhaps I'm letting my understanding of computer vision/geometry/linear algebra run away from me...

I agree with the unpleasantness, as the multiple irregular holes does stimulate my trypophobia slightly. (Note: do not Google "trypophobia" if things with multiple irregular holes makes you feel uneasy, as you will become very disturbed.)

I wasn't aware I have some trypophobia, but now I do heh.

The only people without trypophobia are those who do not know about it.

The search results seems to be about skin disorders/infestations/infections or tattoos that simulate them... I wonder if that's more a healthy response to avoiding things that make your body fall apart?

In any case, yeah don't google it.

For people with trypophobia like me surely will terrified by it.

I thought it was meant to give it a "natural" feeling, i.e. it feels like raindrops to me.

This camera, like many that have come before it are likely to be subject to the uncanny valley of market. Consumers won't want to stomach the price tag and professionals are already accustomed to their DSLR's, and in most cases have significant investment in specific brand and/or lens type with a lot of expensive glass.

Professionals don't care about the bulkiness of their cameras, they care about the performance and flexibility. DSLRs have 20+ years of development in terms of speed, durability, flexibility and the raw image processing power that the giants have spent decades perfecting (lots of work on DSPs).

How many pictures can it take in burst mode? How long does the battery last? Can I use my suite of existing lenses with this?

Sorry to say this camera, while unique and technical marvel, has an uphill battle from the business perspective. It's interesting from a niche perspective but I don't see it becoming a huge success financially for them, which is unfortunate because many talented people probably worked on this.

Moreover, camera concepts like this and Lytro are really good for people who are terrible with a camera, because it lets you "find the shot" in post. If you are good at using a camera and can frame the shot, you have much less of a need for these things.

I agree with everything that you say. However, it's also eerily reminiscent of the explanations I was given 25 years ago about why digital cameras would never catch on. Digitals were far more expensive than a simple Canon or Polaroid, so it would never catch on with consumers. As for professionals, they had far to much experience with developing negatives to throw it all away for some flash in the pan gimmick. Heck, the digitals had terrible lenses compared to what you'd get on Service Merchandise point and shoot, not to mention anything professional grade. Besides, even if CCDs kept improving, they'd be competing with a moving target. Even if digitals were better ten years down the line, it's not like Kodak wasn't going to be improving their processes, too.

A very attractive straw man, but I personally can't remember anyone saying that digital would "never catch on". 25 years ago the state-of-the-art digital camera was a 1.3MP digital back for the Nikon F3, that included a gigantic external box that the user wore over their shoulder. Only a weirdo would use a digital camera in 1990, but it was pretty obvious that digital was the future. 15 years ago the digital camera (Nikon D1) was just barely a good idea for dedicated enthusiasts and still basically terrible for everybody else.

My first thoughts as a working professional for 11 years (retouch, photographer and later videographer).

Forget a raw file format (that you will be able to use in LR/PS) and the flexibility and editing headroom it gives you. Perhaps there will be some specialist proprietary software like for the Lytro (that will be unsupported in 10 years, leaving you with unusable files). Considering this I really hope, as The Verge article says, "the output isn't as punchy" (which normally means contrasty and saturated).

The lack of a raw file format will also mean that I can forget DCP camera profiles to normalize the output with my other cameras.

What we know nothing about: what will AF be like? What will the dynamic range be like (considering they're stitching, not stacking, smartphone sensors)? Noise will probably be OK as you can downsample. But Sony is and Canon has 42/49 MP FF cameras out that doesn't need downsampling to look good. A year from now (when Light comes out) more cameras will have followed.

Focal length range is a bit strange. 35mm equiv is a bit on the long side to be an all-round solution. 24-70 would have been OK for me, but I realize that's personal.

Perhaps we shouldn't expect the death of the MILC/DSLR because of this camera just yet. Many hobbyist though, has left both the compact and the MILC/DSLR for a more enjoyable smartphone experience. Perhaps it could lure them back to buying a camera (without a phone in it)

If you take a look through the FAQ, they do say that they are designing it to work within common photography workflows, giving example formats of JPEG, TIFF, and raw DNG (though I'm assuming the DNG is after their proprietary postprocessing, not a set of the true raw sensor outputs). They're asking for input going forward, so take a few minutes to express your concerns about workflow and maybe it will be a better product for it.

I believe they are doing quite a bit of stacking, as they say they are processing up to 130 megapixels from 10 sensors to get the final image. There should definitely be a dynamic range improvement. I hope they can somehow apply their processing to video streams and provide a wide dynamic range (maybe even HDR) video, but I'd be surprised if they can do more than provide a cellphone-like 4K stream from a single sensor.

Focus speed shouldn't be an issue. I'm guessing it's using tiny cellphone-sized camera sensors, so even at the f/2.4 printed around the lens, they'll have massive depth of field. It looks like they even have the DOF printed on each lens: 10cm-infinity for the 35mm lenses, 40cm-inf for the 70mm, and 100cm-inf for the 150mm. Any focusing will be done computationally at post-processing time.

I'm interested in the concept; I just hope they're a little more developer-friendly then Lytro turned out to be. I know there's a lot of proprietary secret sauce going on with the computational post-processing, but if they can open up their system just a bit to let the crowd in to poke around, there's probably a lot of unique creative opportunities.

Thanks for an excellent reply! Well, I had totally overlooked the FAQ and I see now that they are hoping to deliver DNG, and it doesn't seem to be the demosaiced linear DNG variant (which is very good). As you say it most certainly needs to be processed/exported (either in camera or in post) to DNG. If the software lives as long as the camera, that's fine by me. Many camera makers already "cook" their raws so perhaps this doesn't need to be bad.

> Any focusing will be done computationally at post-processing time.

I catch myself being stuck in old ways of thinking. Again you are right and you just threw some fuel on my excitement fire.

I must say I love the innovation that new actors are bringing to the market. However, I will remain a bit skeptical until release. A lot can happen in a year.

Correct me if I'm wrong, but wouldn't it be entirely possible for them to also save the raw sensor files along with the processed secret sauce image?

Tes, the "raw" data in this case would just be a bunch of sensor readouts from the different sensors. In order to make a useful image you would simply need a computer version of the same software that sits in the camera. The same is true for all raw converters: they do the same job as the camera would when snapping a jpeg. In this case, a bit more involved than just applying a mapping and sensor pattern filter like a normal digital camera, but still essentially the same thing.

The most exciting thing about this (and things like Lytro) is that people are still discovering fundamentally different ways to do 'camera'. Perhaps one day our phones will use a dozen cameras, some optimised for low light, others for high dynamic range, and others still for action, with software figuring out the best combination to use for the each shot.

It's fascinating how difficult a task it seems to be for cameras to shoot at even moderately "non-ideal" conditions. Over- and underexposure, motion blur, focus, chromatic aberration... red eyes...

It seems like a straight-forward problem to solve. I doubt putting 16 different cameras next to each other is something we'll see in smartphones anytime soon but looking at the results (especially low-light shots) is impressive.

Shame about the price, though. Honestly, I'd rather get a great Nikon for half. I guess we're a few years away from this having a chance in the mainstream.

Despite all the interesting cameras that have been made recently, they all rely on one piece of technology that is pretty archaic, the lens. Today's lenses are definitely limiting the image quality available on small sensors, and they seems to improve at a snails pace. I'm cautiously optimistic that this camera may be able to correct for many of the issues with the lenses by using the multiple lenses to create the single image.

Don't really know much about optics to really know the viability of this specially after Litro. But here is what I like

1. Its a different take on the problem, stitching multiple lens and sensor data to make a whole that's better than the parts. Even if this don't work I think that idea would probably live to see another day.

2. There are comments about DSLR being fast, viewfinder easier to work on all situations and I agree to all of that. But I think the biggest point is what one of the founder said in the video, it doesn't matter if you have the best camera @home. This is something you can carry just like your smartphone.

3. Yet even this is a separate device, what I would like eventually is for this idea to be improved and implemented for smartphones. Maybe a 3 lens system that's not as good but much better than current cameras?

4. And I really loved the way the lenses were placed horizontally. That looks to be a simple yet effective innovation (not that I know if it has bad optical consequence).

> Its a different take on the problem, stitching multiple lens and sensor data to make a whole that's better than the parts.

I've seen the idea discussed a few times in the past, essentially doing for cameras what is done with multiple radio telescopes to create something that would be impractical to achieve with a single unit. This is the first time I've heard about someone having implemented the idea outside of a proof-of-concept. I'd be interested to see how well it works in practice. I'm not that much much of a photo taker myself, but a few of my friends are camera bugs so I'll be passing the link on for them to dissect.

I tend to use the HN crowd like that :). You can expect to find from an optics professor to professional photographer and everything in between to comment on this.

The reason you have your camera at home and your smartphone with you is because it is a smartphone. I'm not totally sure it has anything to do with the form factor.

According to the FAQ they support raw DNG. Does that mean that you can get the raw sensor data for each individual sensor and you can hack around your own algorithm for merging them?

Well at least this looks like something new! But notify me again when I can read the review on dpreview.com, the site was kind of light (pun!) on the details.

I agree, all this is nothing without thorough review. Looking forward to that, this really looks interesting though.

This is a straight up weird product at the price point they're targeting. And given that they're asking for money up front and not providing sample raw images and/or full resolution jpegs makes me very hesitant.

This is quite interesting/cool but the two primary reasons I prefer DSLRs to any point and shoot are: 1. The viewfinder. Screens just don't cut it in sunlight or at night (too much or too little light). 2. Time between getting the camera out and getting that shot is very very quick with a DSLR.

It's hard to imagine this replacing my DSLR. If the cost is low enough it could replace my point and shoot though... I'm still more likely to have my phone on me than a camera...

I have an interest in photography and little to no interest in cameras like Lytro, and probably this one although there's not enough information yet to tell. As you point out handling and speed are important. Nobody cares about a "light field" camera if it's impossible to hold and takes three seconds to turn on. $1300 is a huge chunk of money for unknown performance.

This is interesting, but silly. It's hard to imagine a practical use for this other than perhaps wanting to flip between two versions of a photo with one wide open and one sharp across the frame. But you'll have different shutters anyway to keep the same exposure.

Seems like this was spawned from a moment of inquiry where someone was so preoccupied with whether they could that no one stopped to think about whether they should.

I don't think the point is to let you switch between different versions of an image (though the after-the-fact adjustments of depth of field are really slick). I think it's more that with a bunch of sensors and lenses you can simulate a much larger sensor and lens, so you get the low noise and choice of lens profiles of a DSLR without having to actually lug one around, or a bag of lenses to go with it. Some of those pictures in their gallery are pretty awesome, particularly in terms of noise. I'd never have guessed that they were taken with something so portable.

I posted this link in the other duplicate thread: https://pictures.lytro.com/. Not worrying about focus until post processing is pretty awesome.

It doesn't just let you change focus on the fly; it also gives you a honking giant image -- 52 MP according to the Verge. The lenses themselves range from 35 to 150 mm, which gives you a fair amount of play in terms of zoom, too.

So, good low-light capability in a box about the size of a smartphone plus high-quality high-resolution images.

I think it's interesting as a concept, not as a product. If the iPhone 8 comes with four lenses on its back, each with a small sensor, and that means it can use all four to take high-iso pics in near darkness, or fire an individual sensor+lens to take tele or closeups in good lighting etc., then that would certainly be interesting as a product.

I try not to engage in classic HN-style middlebrow dismissal, but can I take a moment to complain about their "Watch the Film" promotional video? It tells me nothing more than "this is a device you can take photographs with". Absolutely useless .

why do they arrange the lens irregularly, what's the reason behind this arrangement?

Unique geometrical offsets for each lens means more reliable deconvolution.

My guess is to create better sub pixel information when stitching the image together. https://en.m.wikipedia.org/wiki/Sub-pixel_resolution

If all the sensor pixel data was aligned it could create bias between the edge of the pixel and the center.

Edit: another thought is triangulation for determining depth

They're not irregularly arranged. If you look at the image of the front of the camera at the bottom of the http://light.co/camera page, they're arranged in concentric circles around a point that's just off to the right hand side of the case. As there are 15 lens and a laser rangefinder, presumably they're directed at various points in space from that center along a straight line in front of the camera. Which 10 lens used for a picture, and consequently how far along that line your subject is, will depend on the rangefinder for an optimal image.

Each camera has a different lens folded under the hood so they just tried to make them all fit as best they could. That's my bet. I don't think the exact location matters that much.

I think, rather than arranging it in specific pattern they tried to fit 16 camera modules as close as possible so as to reduce overall size. As the modules are of different sizes, resulting placement looks irregular. Please refer video 'The Story of light' on page https://light.co/camera @1:25 , wireframe of all modules is displayed.

It is about time someone makes this.I know it looks gimmicky and it kind of is, but this trend should allow more interesting photography. I personally am looking forward to it, whether this one will be good or someone else, I am not invested much, I am just looking forward to the trend.

Also, better processors should allow more sampling from sensors and aggregating that info for better photos.

I was hoping this was an application of the single-pixel camera concept I've read about few years ago[1]. Apparently not.

1. http://web.stanford.edu/~jbarral/PSY221/SinglePixelImaging.h...

You video is pretty useless. People taking pictures. Tell us why it's different, what's new.

This multi-aperture camera was delivered rather before the Light L16:

https://public.nrao.edu/telescopes/vla

It's about time someone did this for consumer products.

Everyone in the video seems to be holding the camera in an awkward way to avoid getting their fingers in front of any of the lenses.

Man oh man, do I want to believe. But given what a huge disappointment Lytro turned out to be, I'll wait for v2 or v3.

Can it make calls?

Reply on news.ycombinator.com