Future of photography: Imaging by light vector recording.

Started Apr 9, 2013 | Discussions
looper1234
Regular MemberPosts: 314
Like?
Future of photography: Imaging by light vector recording.
Apr 9, 2013

The camera has being around for a long time, from the first pin hole, to modern d-SLRs, they all record images by focusing light on to a plane containing light sensitive material.

This method can record two very important details on the light: color and intensity; together with the position they fall on the light sensitive material, an image can be made.

But now a university has constructed a camera that could record light in the form of a vector, that is, information about its color, intensity and direction of travel.

When I heard this news i felt this could have literally earth shattering effect on photography. The most obvious and relevant point being 'focusing' will be a thing of the past.

with 3D vector of lights recorded, any 2D plane can be constructed to produce a photo at that focus. Imagine a sensor capable of recording all lights in all directions around you, and store them in vector form; from this massive collection of data, all possible photos that could be taken at that point with a traditional camera can be constructed.

There could be a day when photography no longer holds the meaning that it does today. I feel this new type of camera is not far, exciting as it is, this could spell the end for many things we enjoy.

One day soon, photography will just be bring a little ball sensor out on a stick, record everything there is to record, and produce what you want in-front of a computer as post processing.

scared ? excited ? share your thoughts.

Doug Pardee
Veteran MemberPosts: 7,518Gear list
Like?
Re: Future of photography: Imaging by light vector recording.
In reply to looper1234, Apr 9, 2013

The Lytro hasn't exactly been a raging success: http://www.dpreview.com/news/tag/lytro

One of the fundamental issues of light-field photography is the extremely low resolution despite the number of sensor elements required. The Lytro delivers images of around 1 megapixel.

-- hide signature --

The open-source LightZone Project: http://lightzoneproject.org/

 Doug Pardee's gear list:Doug Pardee's gear list
Olympus Stylus 1030 SW Canon EOS 350D Canon EF-S 17-85mm f/4-5.6 IS USM Adobe Photoshop Elements 10 LG G2 +20 more
Reply   Reply with quote   Complain
Joseph S Wisniewski
Forum ProPosts: 33,996Gear list
Like?
No, just a passing fad...
In reply to looper1234, Apr 9, 2013

looper1234 wrote:

The camera has being around for a long time, from the first pin hole, to modern d-SLRs, they all record images by focusing light on to a plane containing light sensitive material.

Actually, no. The pinhole doesn't focus anything, it works by vectors. But that's not important...

This method can record two very important details on the light: color and intensity; together with the position they fall on the light sensitive material, an image can be made.

But now a university

Four universities, going back more than a decade.

has constructed a camera that could record light in the form of a vector, that is, information about its color, intensity and direction of travel.

When I heard this news

The "news" is that two different companies, Lytro an Raytrix put them into production. Both flopped. We'll get to why in just a second.

i felt this could have literally earth shattering effect on photography. The most obvious and relevant point being 'focusing' will be a thing of the past.

Something that is not true can be neither "obvious" or "relevant".

with 3D vector of lights recorded, any 2D plane can be constructed to produce a photo at that focus.

Not "any plane", because that would require an infinite number of infinitely thin rays.

Imagine a sensor capable of recording all lights in all directions around you, and store them in vector form; from this massive collection of data, all possible photos that could be taken at that point with a traditional camera can be constructed.

Again, no. When you move from the infinite number of rays needed to do what you describe, to the finite number that is part of reality, the whole thing sort of breaks down. For example, the most popular paper on the subject, by Ren Ng from Stamford, describes his experiments with a camera that took 169 rays per pixel (a 13x13 matrix, more or less). To do that, he required a 16mp (4000x4000) sensor to produce a 0.083mp (288x288) image. The decimation of the sensor into 13x13 ray packets meant that he could refocus the camera's f4 lens over the DOF range of an f26 lens (since it was a medium format camera, that wasn't that much range).

The tradeoff between the "refocus" ability (or 3D ability) and resolution (and low light ability) is what insures that this technology will never be more than a novelty.

Ren Ng started a company to commercialize it, but to make it affordable, he lowered the megapixel count a bit and lowered the decimation a lot, to get a higher resolution output. The end result was a camera with a lot less of the "lightfield magic" that his earlier papers promised combined with a much lower resolution than a phone camera or even a $11 keychain camera.

There could be a day when photography no longer holds the meaning that it does today.

Of course. Photography doesn't hold the meaning today that it did 10 years ago. It won't hold the meaning 10 years from now that it holds today. That is obvious.

I feel this new type of camera is not far,

You are incorrect. Sorry.

exciting as it is, this could spell the end for many things we enjoy.

One day soon, photography will just be bring a little ball sensor out on a stick, record everything there is to record, and produce what you want in-front of a computer as post processing.

"Photography will just be" nothing of the sort. The vast majority of users don't want to post process. The "one day soon" workflow is that they compose with the device at hand (probably a phone) and share with their friends immediately (probably on facebook).

scared ? excited ? share your thoughts.

Already did.

-- hide signature --

Rahon Klavanian 1912-2008.
Armenian genocide survivor, amazing cook, scrabble master, and loving grandmother. You will be missed.
Ciao! Joseph
www.swissarmyfork.com

 Joseph S Wisniewski's gear list:Joseph S Wisniewski's gear list
Canon EOS 5D Mark II Nikon D90 Nikon D2X Nikon D3 Nikon D100 +43 more
Reply   Reply with quote   Complain
looper1234
Regular MemberPosts: 314
Like?
Re: No, just a passing fad...
In reply to Joseph S Wisniewski, Apr 9, 2013

just a passing fad...

so says the kodak executive who took a pass on digital photography

Reply   Reply with quote   Complain
Joseph S Wisniewski
Forum ProPosts: 33,996Gear list
Like?
The difference between "someone" and "everyone"
In reply to looper1234, Apr 9, 2013

looper1234 wrote:

just a passing fad...

so says the kodak executive who took a pass on digital photography

Which executive would that be, the one who authorized the invention of the practical digital camera, the one that started producing the first industry's first DSLRs, the one who made Kodak the number one brand of aerospace, scientific, and industrial sensors, or the one that made Kodak the world's number one producer of consumer digital cameras for years?

Your analogy is wrong on multiple levels. As you can see after a minute's thought, there is no "kodak executive who took a pass on digital photography". They died because they had a deeply entrenched corporate culture that was formed at a time when their main technology field (photochemistry) advanced at a much slower rate than electronics. They couldn't adapt to the rate of progress that the more experienced electronics companies could, and fell behind.

The second level of wrongness is a confusion of cause and effect. A lot of companies that enthusiastically embraced digital photography also failed.

The third level is practicality. While Kodak was moving slowly, they were moving. Everyone was moving, because everyone skilled in technology forecasting could make pretty decent projections on how digital would evolve over time. Those same predictive techniques show that lightfield cameras really will be nothing more than niche products.

You described something that isn't just marketing, to do what you predict, you have to break laws of physics. Physics is pretty unbreakable.

-- hide signature --

Rahon Klavanian 1912-2008.
Armenian genocide survivor, amazing cook, scrabble master, and loving grandmother. You will be missed.
Ciao! Joseph
www.swissarmyfork.com

 Joseph S Wisniewski's gear list:Joseph S Wisniewski's gear list
Canon EOS 5D Mark II Nikon D90 Nikon D2X Nikon D3 Nikon D100 +43 more
Reply   Reply with quote   Complain
looper1234
Regular MemberPosts: 314
Like?
Re: The difference between "someone" and "everyone"
In reply to Joseph S Wisniewski, Apr 9, 2013
You described something that isn't just marketing, to do what you predict, you have to break laws of physics. Physics is pretty unbreakable.

Why is modeling light vectors is impossible ?  which law of physics is broken by collecting light information as vectors and recording them ?

to there is only a finite amount of light at any given location; a constructed model based on the vectors collected would be enough to allow post processing to mimic the effect of traditional cameras within such a space.

i am talking about software aperture, software lens, software shutter speed.

the only difficulty of this process is method of collection, how to construct an accurate model. If your opinion is that this can never be done, or by your own words "physically impossible" than I think the Kodak executive analogy is spot on for you.

Reply   Reply with quote   Complain
looper1234
Regular MemberPosts: 314
Like?
Re: Future of photography: Imaging by light vector recording.
In reply to looper1234, Apr 9, 2013

I didn't want to sound adversarial, but i believe what I described is a plausible terminus to the technical aspect of photography.

i don't care that someone tried to make a cheap buck using immature technology, or some paper i can't even comprehend.

the technical aspect of photography is not hard to explain; collect light information, use said information to construct image. In the old days we do so in darkrooms, now we have lightroom and who knows.

all the rest is mere technicalities.

Reply   Reply with quote   Complain
GodSpeaks
Forum ProPosts: 10,273Gear list
Like?
Re: Future of photography: Imaging by light vector recording.
In reply to looper1234, Apr 9, 2013

looper1234 wrote:

scared ? excited ? share your thoughts.

Just pie in the sky BS.

Remember LYTRO.

-- hide signature --

The greatest of mankind's criminals are those who delude themselves into thinking they have done 'the right thing.'
- Rayna Butler

 GodSpeaks's gear list:GodSpeaks's gear list
Minolta DiMAGE 7i Panasonic Lumix DMC-LX3 Panasonic Lumix DMC-TS4 Panasonic Lumix DMC-TS5 Kodak DCS Pro SLR/n +40 more
Reply   Reply with quote   Complain
Joseph S Wisniewski
Forum ProPosts: 33,996Gear list
Like?
which law of physics is broken
In reply to looper1234, Apr 9, 2013

looper1234 wrote:

You described something that isn't just marketing, to do what you predict, you have to break laws of physics. Physics is pretty unbreakable.

Why is modeling light vectors is impossible ?  which law of physics is broken by collecting light information as vectors and recording them ?

to there is only a finite amount of light at any given location; a constructed model based on the vectors collected would be enough to allow post processing to mimic the effect of traditional cameras within such a space.

I'm not sure what "to there" means, but you're pretty close to explaining why it won't work.

"there is only a finite amount of light at any given location"

That's the whole ticket. To emulate a lens, you have to have a lot of rays. Ren Ng decimated his lens by a factor of 13 squared, and that cut both his pixel count and his light gathering ability down by a factor of 169.

The decimation required to even begin to simulate that "software lens" is on the order of 10,000, and there simply aren't enough photons in that "finite amount of light at any given location" to do it.

i am talking about software aperture,

Within limits.

software lens,

No. There still a physical lens in a plenoptic camera, a good one. The physical lens has to exceed the capabilities (and therefore, the size and weight) of every lens that the plenoptic camera tries to emulate through software. Have a look at the Lytro, it's essentially all lens.

software shutter speed.

Nope. There's nothing magic about plenoptics that negates the simple fact that adding an electronic shutter to a sensor cuts your low light sensitivity in half. They've been around for decades, but you seldom see anyone try to use one on a camera. It's a compromise no one wants to deal with.

the only difficulty of this process is method of collection, how to construct an accurate model.

That's like saying the only difficulty of anti-gravity cars is anti-gravity. You can't "model" what you don't have.

If your opinion is that this can never be done, or by your own words "physically impossible" than I think the Kodak executive analogy is spot on for you.

If you think that analogy is appropriate, then you do not want to know what I think of you.

I'm done wasting time with you. Have a nice life.

-- hide signature --

Rahon Klavanian 1912-2008.
Armenian genocide survivor, amazing cook, scrabble master, and loving grandmother. You will be missed.
Ciao! Joseph
www.swissarmyfork.com

 Joseph S Wisniewski's gear list:Joseph S Wisniewski's gear list
Canon EOS 5D Mark II Nikon D90 Nikon D2X Nikon D3 Nikon D100 +43 more
Reply   Reply with quote   Complain
Joseph S Wisniewski
Forum ProPosts: 33,996Gear list
Like?
If you cannot comprehend the technology...
In reply to looper1234, Apr 9, 2013

looper1234 wrote:

I didn't want to sound adversarial, but i believe what I described is a plausible terminus to the technical aspect of photography.

i don't care that someone tried to make a cheap buck using immature technology, or some paper i can't even comprehend.

If you cannot comprehend the technology, then what is your "belief" that it is "plausible" based on?

The people who created it say that it's not possible to use that technology in the manner you describe.

the technical aspect of photography is not hard to explain; collect light information, use said information to construct image. In the old days we do so in darkrooms, now we have lightroom and who knows.

all the rest is mere technicalities.

Not unless you have some idea how to address them. That's what "technicalities" are, the technical details. If there is no technology to solve a problem, then there are no technicalities of how to implement it.

-- hide signature --

Rahon Klavanian 1912-2008.
Armenian genocide survivor, amazing cook, scrabble master, and loving grandmother. You will be missed.
Ciao! Joseph
www.swissarmyfork.com

 Joseph S Wisniewski's gear list:Joseph S Wisniewski's gear list
Canon EOS 5D Mark II Nikon D90 Nikon D2X Nikon D3 Nikon D100 +43 more
Reply   Reply with quote   Complain
Piginho
Regular MemberPosts: 305
Like?
Re: Future of photography: Imaging by light vector recording.
In reply to looper1234, Apr 9, 2013

looper1234 wrote:

One day soon, photography will just be bring a little ball sensor out on a stick, record everything there is to record, and produce what you want in-front of a computer as post processing.

scared ? excited ? share your thoughts.

Who wants to take photos with a ball on a stick held in the air?

Seriously though, if you wanted to be able to do as you describe and then say, pick your focal length (effectively by cropping) and pick you focal distance, all from an area encompassing 360 degrees in all planes, to an acceptable level of resolution, you would surely need receptors which would be much smaller than the wavelength  of the light reaching them.

Although I haven't done any calculations to back this up I suspect that the laws of physics will always prevent you getting the level of functionality that you would desire, together with acceptable resolution.

So am I "scared ? excited ?"  Neither, as I'm certain that it will never provide anything useful for me.

Reply   Reply with quote   Complain
looper1234
Regular MemberPosts: 314
Like?
Re: which law of physics is broken
In reply to Joseph S Wisniewski, Apr 9, 2013
I'm done wasting time with you. Have a nice life.

lol, i think I've being more than polite to you considering this is a forum for "adults"; although sometimes i truly wonder.

if your final argument is to compare light vector modeling to anti-gravity, than i can only describe your position as desperate.

I can make a simple model for a black dot in 3D space and simulate taking photographs of it right now. I don't see any difficulty in that.

since you said its impossible to do this, I've decided to outline a plausible way to obtain real life model details: one can simply mount such custom senors on fixed and known locations to obtain to relative deferential to solve for spatial coordinates. 3 sensors to fix a position, 3 more to fix another to get the vector.

the fact that you still mention lens means you are still trapped in your own narrow mind: just as a simulation is enough to predict accurately a projectile's trajectory so can the same be done to lights; any light traveling thru a lens will have its behavior modeled and its results simulated. Your argument keeps brings up the failures of a camera that i don't even know or care.

I am sure most of the time you get enormous satisfaction dissecting technical details where there are none. And to point them out to people, to fill your own ego. I have to say you have really spent a lot of effort.

Perhaps you had a bad day, and want to vent, and my post look sufficiently newbie enough to serve your purpose

~but know this~

you just got served by a 19 year old who uses English as a second language.

Reply   Reply with quote   Complain
nnowak
Regular MemberPosts: 384
Like?
Re: which law of physics is broken
In reply to looper1234, Apr 9, 2013

looper1234 wrote:


I can make a simple model for a black dot in 3D space and simulate taking photographs of it right now. I don't see any difficulty in that.

since you said its impossible to do this, I've decided to outline a plausible way to obtain real life model details: one can simply mount such custom senors on fixed and known locations to obtain to relative deferential to solve for spatial coordinates. 3 sensors to fix a position, 3 more to fix another to get the vector.


~but know this~

you just got served by a 19 year old who uses English as a second language.

I think you just proved the futility of this concept with your own "simulation".  To capture the vector information for a single black dot, you needed to use 6 sensors.  What if you want to know what color a non-black dot is along with that vector information?  You are now up to 6 red, 6 green, and 6 blue sensors for a minimum total of 18 sensors.  And that is just for a single dot.

I'm not sure what you photograph, but my pictures are usually more detailed than a single dot.  To capture the vector information, you either need to drastically increase the amount of data collected, or drastically reduce the effective resolution.  The miniscule value of vector information is far outweighed by the downsides from collecting it.

To look at it another way, you would need to collect massive amounts of data to completely define a scene in 3D space.  You then throw away the majority of that data when you define a single 2D focal plane and create a final image.  That is not an efficient use of hardware or time.  it is so much simpler to just get focus set right from the start.

I think you just "got served" by yourself.

-- hide signature --

Nick

Reply   Reply with quote   Complain
wfektar
Contributing MemberPosts: 694
Like?
Re: Future of photography: Imaging by light vector recording.
In reply to looper1234, Apr 9, 2013

What does "light vector recording" even mean?

This method can record two very important details on the light: color and intensity; together with the position they fall on the light sensitive material, an image can be made.

Color and intensity define a perfectly good vector.

But now a university has constructed a camera that could record light in the form of a vector, that is, information about its color, intensity and direction of travel.

"Direction of travel" is well known -- towards the camera. You know just where because it's an imaging system.

with 3D vector of lights recorded, any 2D plane can be constructed to produce a photo at that focus. Imagine a sensor capable of recording all lights in all directions around you, and store them in vector form; from this massive collection of data, all possible photos that could be taken at that point with a traditional camera can be constructed.

OK this sounds like plenoptics. You're selecting a tiny fraction of the total information collected, but you have to collect all that information because at the time of capture you haven't decided which fraction you want to use. It's competing against the "standard" approach, which is to decide beforehand and maximize those information resources towards that decision.

scared ? excited ? share your thoughts.

Why the breathless enthusiasm right now? A hundred years later we're still stuck with the information bandwidth issue. Bandwidth has gone up but not nearly as much as expectations -- that is, from a photographer's perspective, to get something acceptable, instead of proof of concept. We're nowhere near that threshold.

Reply   Reply with quote   Complain
looper1234
Regular MemberPosts: 314
Like?
Re: Future of photography: Imaging by light vector recording.
In reply to wfektar, Apr 10, 2013

Stand up, look around, hows does that 3D model look to you ? that was from just two sensors. (I am assuming you still have two eyes.) so six sensors are plenty.

The brain constructs 3D models from 2D retinal signals.  we can do this even without  the mapping data on surrounding terrains; Why do so many scream, this can't be done, when they have being doing it themselves since the dawn of time.
with more sensors and more accurate terrain data, a model of great resolution and fidelity can be created, greater than the model in your brain.

laser dot matrix mapping, satellite mapping; the technology is all there to replicate the same type of modeling that the brain does.

Chances are its already happening in the military.

predetermined or dot matrix mapped 3D model of surroundings, coupled with visor sensory information from each squad member, integrated and extrapolated into a 3D model.

it is only a matter of time before this technology can provide better resolution and fidelity than our eyes and brains can.

Reply   Reply with quote   Complain
Mahmoud Mousef
Senior MemberPosts: 2,363
Like?
Re: No, just a passing fad...
In reply to looper1234, Apr 10, 2013

looper1234 wrote:

just a passing fad...

so says the kodak executive who took a pass on digital photography

http://www.imaging-resource.com/PRODS/DC210/DC210Acgi.HTM

Kodak were definitely one of the digital camera pioneers.

I don't know enough about their situation to even guess why they aren't a major player in cameras, but I know they made sensors for some big names, including Olympus.

The very first digital camera I saw selling in stores was a Kodak.

Reply   Reply with quote   Complain
Detail Man
Forum ProPosts: 15,271
Like?
Re: Future of photography: Imaging by light vector recording.
In reply to looper1234, Apr 10, 2013

looper1234 wrote:

Chances are its already happening in the military.

Now I understand why the phrase "military intelligence" is considered by some to be oxymoronic.

... it is only a matter of time before this technology can provide better resolution and fidelity than our eyes and brains can.

Perhaps we are here providing living proof that it has already eclipsed our cognitive capacities ? ...

Reply   Reply with quote   Complain
looper1234
Regular MemberPosts: 314
Like?
Re: Future of photography: Imaging by light vector recording.
In reply to Detail Man, Apr 10, 2013

Some thought just popped up in my head.

I am also sure that Google is perusing exactly this.

internal 3D mapping of buildings by their "glasses" in the form of laser dot matrix + differential sensor readings from the two "eye" sensors.

now imagine thousands uploading these exact information to their server, with this information; they can construct a model that could rival any the brain could construct.

A collective vision, a single brain with millions of eyes.

street view would become real view.

Reply   Reply with quote   Complain
wfektar
Contributing MemberPosts: 694
Like?
Re: Future of photography: Imaging by light vector recording.
In reply to looper1234, Apr 10, 2013

SkyNet!

But you still haven't addressed the information resource allocation issue. Which is the backbreaker for plenoptics.

Reply   Reply with quote   Complain
Detail Man
Forum ProPosts: 15,271
Like?
Re: Future of photography: Imaging by light vector recording.
In reply to looper1234, Apr 10, 2013

looper1234 wrote:

Some thought just popped up in my head.

I am also sure that Google is perusing exactly this.

internal 3D mapping of buildings by their "glasses" in the form of laser dot matrix + differential sensor readings from the two "eye" sensors.

now imagine thousands uploading these exact information to their server, with this information; they can construct a model that could rival any the brain could construct.

A collective vision, a single brain with millions of eyes.

street view would become real view.

You realize, don't you, that the last ice age thawed around 11,500 years ago, 90% of the planet's history is one of being buried under hundreds of feet of ice (except perhaps at the equator), and that all of our bold and foxy efforts are up against a 500,000 year optical design cycle(?):

Source: http://www.detectingdesign.com/images/HumanEye/Muller%20Glial%20Cells.jpg

Perhaps we had best hurry up before we roast the planet, plunging us into the next ice age ?

Reply   Reply with quote   Complain
Keyboard shortcuts:
FForum MMy threads