Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Lytro camera changes everything

Sol

Sol

macrumors 68000
Original poster
Jan 14, 2003
1,564
6
Australia
With a whole section of the Forums dedicated to digital photography, it is surprising that no-one is talking about the Lytro cameras.
lytro-camera-front-right.jpg

http://www.lytro.com/

These will take photos that can have their focus adjusted by the viewer. Quite a step forward in photography, making it possible to have 3D films that are displayed on monitors capable of reading our eyes' focus and adjusting the video's focus in real-time.

It is an exciting new technology and I look forward to its ubiquity.
 
Definitely an amazing technology. Looks very fun. I played around on their website for a while and it was really cool. The price is pretty reasonable.

But this is going to be a tough sell in many ways. Most people don't really enjoy editing their photos, which this kind of camera fundamentally requires. They don't give specs on the resolution of the pictures. Are they high enough resolution to print? Are they high enough resolution to fill a monitor? How does this do in low light? Fast action?

But I agree about the long term. Those sort of criticisms of this particular product aren't as important as the potential application of the technology. Really amazing.
 
Agreed. It is impressive technology but as a product it seems deemed to fail. People who take snapshots aren't really going to care while enthusiasts and professionals want a DSLR where they can have full control and high quality. If they want to make an impact in the market they need to make it a great camera in a traditional sense as well as an innovation.
 
Sol said:
making it possible to have 3D films that are displayed on monitors capable of reading our eyes' focus and adjusting the video's focus in real-time.
Click to expand...

I hadn't considered this aspect of it before. THAT would be something to experience. The drawback with current 3D films is that despite the apparent depth in the image, your focus is still limited to a set area. Not sure how/if this would work in a cinema, but certainly for personal viewing this could be quite interesting.
 
It will have limited use in the consumer market. Pros won't have a use for it, at least not in traditional photography.
 
I think it opens up a whole new area of photography, but I don't think it will replace anything. It will be cool, as a viewer, to be able to play with the focus in a picture. But not for everything. It also doesn't work, as a viewer, for anything printed. But for some speciality uses I can see this being very useful.

Most consumers don't focus their cameras as it is, they let the autofocus do it. Nor, as pointed out already, do they tend to edit their photos. Unfortunately - because even just editing out the really bad photos would make the world a better place... but I digress.

Professionals deliberately pick their focus during the act of capturing the photo. I suspect most pros will find it easier to continue doing so rather than sitting down after the fact and picking the focus as part of post production. That said, I can see a few areas of commercial photography where having a Lytro Camera would be a good thing. Photojournalism in crowd situations, perhaps? I'm sure there will be a few fields where pros will think Lytro is the best things since sharp glass. I think some pro cameras will, at some point when the licensing has been sorted out, have a "Lytro" on/off setting similar to AF and MF, etc I am assuming that a camera can use one sensor module and it's just a matter of whether an image is processed with one algorithm or another.

I suspect we will see an explosion of Lytro images in camera clubs, photo schools, and among those who aren't fully professional, but do try to charge for their photographic services.

One interesting observation. At least so far, I haven't seen that a Lytro Camera can change it's depth of field. Everything I've seen so far seems to be showing a shallow field being moved around. Though I suppose Photoshop will introduce a tool that will allow one to pick the near and far points of focus, and to then save a PSD file with that DoF. Hmmm.... this could be interesting after all.
 
snberk103 said:
One interesting observation. At least so far, I haven't seen that a Lytro Camera can change it's depth of field. Everything I've seen so far seems to be showing a shallow field being moved around. Though I suppose Photoshop will introduce a tool that will allow one to pick the near and far points of focus, and to then save a PSD file with that DoF. Hmmm.... this could be interesting after all.
Click to expand...

Interesting idea. I suppose you could make a picture with the foreground and background in focus, but the mid-ground out of focus. Lots of creative options.
 
The question that I have is what do I do with it after I take the shot? It seems like it is a proprietary file format of some sort that requires a viewer from Lytro. If it can't be managed by PhotoShop or Lightroom/Aperture to produce a printable image, then it has limited use for the people who make professional use of photography, or even the serious amateurs like myself. We are interested in producing a final product that is an artistic statement of our creative vision.

Dale
 
As a consumer camera, the most practical default for the processing software would be to produce an image with "infinite" depth of field. Isn't that really what the average consumer would want. But that would be rather boring in press releases, wouldn't it?

Not to mention, not much different than any cheap consumer digital camera in the final result. Given the small image sensors and small apertures, consumer cameras already deliver large depth-of-field with little real need for focusing.

I don't think the average person wants creative control of focus. And those who do probably won't like the limitations it imposes.

Certainly an interesting niche product. Certainly some scientific applications, but they need to make it with a microscope adapter. (Assume that will be forthcoming, at least from third parties.)
 
Interesting tech, but I think it's going to fail.

Lytro say that this camera has a resolution of '12 megarays'. I understand that it needs to have a number of sensors for each conventional pixel of resultant image, so those 12 megarays turn into something like a 1.2 megapixel image. This is due to it's need to capture light direction as well as intensity.

The problem with this is that for any sensor technology, you're effectively throwing away 90% of your resolution just to get the re-focussing capability.

Conventional camera lens focussing is an old, established and pretty effective technology. I can't see the appeal of degrading your image quality by 10x, just to get the ability to refocus in post - when it's easy enough to get the focus right in the camera. Would anyone really want to spend time refocussing when they get back to their computer?

So, the cost in terms of image quality is just too great (and will always be too great) for the gimmicky appeal of being able to re-focus.
 
firestarter said:
when it's easy enough to get the focus right in the camera
Click to expand...

Maybe for you, as a photographer, but not everyone agrees with that. I can't tell you how many times I've seen people with a shot of a family event or a once-in-a-lifetime opportunity upset because it's out of focus. Point and shoots just can't always focus accurately or fast enough.
 
mackmgg said:
I can't tell you how many times I've seen people with a shot of a family event or a once-in-a-lifetime opportunity upset because it's out of focus. Point and shoots just can't always focus accurately or fast enough.
Click to expand...

And not just point and shoots, SLR's can get the focus wrong just as easily.
 
mackmgg said:
Maybe for you, as a photographer, but not everyone agrees with that. I can't tell you how many times I've seen people with a shot of a family event or a once-in-a-lifetime opportunity upset because it's out of focus. Point and shoots just can't always focus accurately or fast enough.
Click to expand...

They're getting much better. Facial recognition and software to sense whether focus has been achieved is now pretty standard on point and shoots. There are even cameras that only take a photo if you're in focus and smiling.

The bottom line is whether people are willing to accept a 10x degradation in image quality for this capability - and how willing they are to fire the computer up and futz with raw files and special software if all they want is a snapshot.

Added to that... P&S cameras are loosing the competition against smart phones.

admwright said:
And not just point and shoots, SLR's can get the focus wrong just as easily.
Click to expand...

That's called operator error.
 
firestarter said:
They're getting much better. Facial recognition and software to sense whether focus has been achieved is now pretty standard on point and shoots. There are even cameras that only take a photo if you're in focus and smiling.

The bottom line is whether people are willing to accept a 10x degradation in image quality for this capability - and how willing they are to fire the computer up and futz with raw files and special software if all they want is a snapshot.
Click to expand...

It's early days. Because it is a radical new lens system, it still needs to become established with some products. The way technology advances, the CCD in future cameras may have 10X degradation and still contain more pixels than your best camera today.

snberk103 mentioned the suitability of this kind of photograph in photojournalism and I would add forensics to the list. At least Deckard finds a use for it in Bladerunner.

It just seems like a superior way to record a moment in time in terms of information, despite the resolution penalty.
 
I think there's more to it than just being able to focus after the fact. They came up with this fun toy to bring in some capital. I think they have so bigger ideas.

From my understanding, the size of the DoF in the original images has nothing to do with the final processed image, the camera uses a constant f/2. They are saying that they can make images that are 100% in focus with their lab software. This could mean taking the equivalent DoF of a lens at f/11 when you're at say f/1.8. This would be huge in terms of the amount of light needed to create the same picture. It would also mean that you wouldn't have to care about your aperture when taking the picture.

As some have talked about 3D, it's actually possible to create a 3D image with one lens using a light field sensor, that's another big plus.

I'm really excited to see what they come up with as I think it's kind of naive to think that the only use of this technology is to focus after you've taken a shot.
 
I'd love to see one of these sensors coupled with a macro lens. I'd be willing to sacrifice 10x quality in order to have my 1:1 image fully in focus.
 
paolo- said:
the camera uses a constant f/2. They are saying that they can make images that are 100% in focus with their lab software. This could mean taking the equivalent DoF of a lens at f/11 when you're at say f/1.8. This would be huge in terms of the amount of light needed to create the same picture.
Click to expand...

But, is that true?

You've got extra pixels to capture angle information. That means smaller pixels. Smaller pixels = less surface area = less sensitivity.

Now, sure, with a conventional sensor, you can combine pixels and average to reduce noise. That is, it doesn't really matter if you have, say, 1000 big pixels in a row or 5,000 small ones. You can combine cells, add-up the values, and average, and you'll get close to the same sensitivity (minus extra dead space between the smaller pixels!) with of course high noise and clipping in the extreme highlights and shadows.

Can the processing algorithm both extract the angle information AND meaningfully average values to increase sensitivity?
 
If you wanted the ability to have every possible part of the image in sharp focus and yet retain accurate depth of field with the lens wide open, it seems to me that the simple way to do it without losing so much as one pixel of resolution would be to have the focus ring outfitted with a super-fast, super-accurate stepper motor. When you get the exposure levels and framing right you take the shot. The camera instantly takes a rapid series of perhaps 100 separate exposures with the focus ring spinning from macro to infinite focus. Then some sort of software algorithm could extrapolate some sort of image synthesis for the spots between the 100 samples, so that, in post, you can turn a dial or drag a mouse to change the part of the frame in focus, just as you would have at the moment you took the photo.

You'd need very fast, very light-sensitive sensors for any photos of moving objects or you might get some CMOS "jello". But it could be done.

Of course the big question is WHY. If you want everything in focus, use a tiny lens. Walaah! Everything in focus. Most P&S people wanting to post snapshots to Facebook are quite content with that.

If you used a max focus P&S camera but you wanted to give the impression of a more "artsy" shallow depth of field shot, that can be done in post. You can fake depth of field by cutting out objects in Photoshop and blurring them on either side of whatever point you are trying to imply is the focus point of the picture (foreground and background objects). A lot of work for fakery that a real photographer would spot instantly, but possible. Unfortunately, the perspective of the lens would give it away for those in the know.

Those people who actually know what shallow depth of field can do (more serious amateur and semi-pro/pro photographers) probably know what they want to focus on within a given shot.

Sounds like a solution in search of a problem to me, at least in regards to stills. If they could really make it work fast, without CMOS "jello", it might make a great technology for video capture. If you could capture images fast enough (24fps x 100 exposures each) and if you could still have the right motion "smear" that a standard 1/48 exposure gives you, then, in post, you could time out your rack focus shots any way you want. Downside--the poor 1st Assistant Cameraman would quickly be out of a job--no need for a focus puller if you'll be pulling it in post!
 
jtara said:
But, is that true?

You've got extra pixels to capture angle information. That means smaller pixels. Smaller pixels = less surface area = less sensitivity.

Now, sure, with a conventional sensor, you can combine pixels and average to reduce noise. That is, it doesn't really matter if you have, say, 1000 big pixels in a row or 5,000 small ones. You can combine cells, add-up the values, and average, and you'll get close to the same sensitivity (minus extra dead space between the smaller pixels!) with of course high noise and clipping in the extreme highlights and shadows.

Can the processing algorithm both extract the angle information AND meaningfully average values to increase sensitivity?
Click to expand...

I'm not sure I truly understand what you're saying.

First, I'm not sure if the sensor is actually flat. From my understanding, you would need at least two translucent sensors one in front of the other to be able to determine the angle of the incoming light. Poke a spear through a sheet of paper, it's hard to tell the angle if you only have 2D info, poke a spear through two sheets of paper, you now have the offset that can give you the angle. If that's the case then there would not really be less pixels per area, as the sensor that gives you the angle is independent from the one that gives you colour. The other way they could be doing it is by having multiple sensors per microlens on the sensor. Basically having a bunch of little domes on the sensors and you would be able to tell the angle of the incoming light by checking the light intensity at different spots inside that dome. If that was the case, all of the sensors inside that dome would still have usable data if you were to do some noise reduction. But I really have no idea how they do it. [just checked wiki it's a bunch of microlenses]

But I can see your point, if you only have 1 megapixel on a light field sensor and have 10 on a regular sensor, would the offset of being able to use your lens wide open be worth it compared to using some noise reduction method? I'm not sure how modern cameras lower the noise with higher ISO settings but what the averaging you suggest actually lowers the sensitivity of the camera, you still have as many pixels but they are averaged so you're not getting an equivalent image. You're loosing clarity with such a noise reduction method.

Bottom line is how what pixel density can you get on a lytro camera compared to a regular camera and if those additional pixels can actually give you more light info than using a wide open aperture all the time.

Not sure if we can really find the answer but still interesting to see where this technology will lead us.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back