Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

To buy or not to buy, Lytro Light Field camera.

W

Wayne Morellini

Well-known member
Joined
Jan 3, 2008
Messages
6,157
Reaction score
0
Points
0
I want one so bad but they're only shipping to US. I'd send you the money to get me one and ship it to me in Egypt if you would. Seems like an awesome toy.
 
Apparently the rez on this thing isn't great, but its a sign that things are possible in such a small package...

but given what Jannard just posted above, think RED will be able to push this and if it lands in Scarlet in the future...wow...i hope to god that Canon have a game plan to compete because given the stuff from Adobe then you mix it with RED...I think you get a drug that youll never want to stop using
 
Wayne Morellini said:
The Light Field camera is here from Lytro.

www.steves-digicams.com/news/the_lytro_light_field_camera_is_here.html
https://www.lytro.com/camera

It does not seem to have a video mode.
Click to expand...

Yeah, have been looking at that for a while and my buddy in fact put two on the list for me, the RED ones with higher capacity was to take delivery of them next week..

But after Jim's post...

Jannard said:
One thing I will say is that we are working on multiple focus and depth map (think Lytros) but with wicked high resolution... The prototype is incredible. Don't tell Gizmodo or Engadget. :-)

Jim
Click to expand...


I'll wait... ;)
 
Mohamed Younis said:
I want one so bad but they're only shipping to US. I'd send you the money to get me one and ship it to me in Egypt if you would. Seems like an awesome toy.
Click to expand...

Sorry, I'm also not in the US Mohamed. Pity about the us only, too nsny companies seem to be doing it lately.
 
Jannard said:
http://www.reduser.net/forum/showthread.php?65086-I-want-to-tell-you.../page2

Jim
Click to expand...

Old, he he, you should have hired me, would be a few years earlier on these things.

You can do this in software (see my writeup on how to do it years ago when the Light Feild technique was first released). Intel also implemented a 2d to 3d conversion before I got around to implementing something similar. However, while this might be made to work, I didn't like my chances of getting the quality level up enough all the time. I am considering at least three point sampling now, at least 4 for professional, Lytro probably uses 9 channels (not the same as points, for everybody else that is reading).

So, what else could Red be up to? There are variouse ways to get a depth map (can't remember my own techniques from years ago) (somebody came up with an adequate technique in the late 90's or early 2000's). However I just remember a real beauty of a solution I came up with a few years ago, so I better be careful how I answer this one. Laser can also be used.

What can be done in hardware. Key tech, motion compensation etc.

Taking advantage of high frame rates with a high speed focus mechanism, you can shift focus between frames to glean depth information. Now you compensate for shifting motion in the frame to correctly interpret the focus. I claim the invention.

A mechanism, like the isee3d mechanism is, that shifts the view can give 3d. But I like a nice round focus aperture. Isee3d is a cheap base line 3d option for people that don't want, need, or can't afford, to resort to a higher quality twin camera solution.

So, did I hit the spot at all Jim?

I just realised, I could save myself a bit of hassle with this. Could apply this to pro video, think I might contact viimagic.


Thanks again.


Wayne.
 
If only they got real about it and included a grade video.
 
It's a fun toy, but I can't think of any useful applications for it. Take a look at the analysis by other engineers and camera designers, for example:

http://forums.dpreview.com/forums/read.asp?forum=1018&message=39633412

If they designed it to allow two stops of DOF adjustment (from f/11 to f/45 in FF35-equivalent DOF), then the resolution is only 0.7 MP. If they designed it to allow only one stop of DOF adjustment (which is unlikely, since that makes the difference from a normal camera incredibly small), then the resolution would be 2 MP.

It's a novelty item.
 
Oh, the intent of the idea has great promise, the implementation though, I am yet to examine.

One day you may not need to do much with the camera, it may mostly be done in post. What this really means until then, is that the professional can select what he recommends, and it canbe touched up in post. On small productions a few people could be replaced with adjustments left to post. That is the possible intent of techniques like these.
 
As far as I can tell, the Lytro isn't even as advanced as the Adobe multi-focus / lightfield demo from around 2 years ago. It wasn't a new idea or concept then, either.
 
Well New way of achieving it maybe, however..
 
Deanan said:
nope
Click to expand...

I forgot one, well that I am willing to talk about, but it might have patent issues. Some years ago I came up with a method to store uncompressed HD to VHS tape using existing technology, also for SD cameras to do HD. It involved pixel shifting to reveal higher resolution. I think around the time I considered focus shifting, I also considered using shifting pixels to reveal 3D depth. You see, the closer the object the more it shifts relative to objects further away revealing depth. So using, a pixel or less of shift, not very noticeable, you can calculate distances, and just stack slice the image planes in the frame (or use before mentioned methods to calculate distances on an object). It is basically like when we focus by contrast. The more dramatic the shift from distance the more the pixel value changes. Now, if you want to get rid of the little blur from the shift scan, you use a little motion compensation or and shift more slowly.

This shift might be in an arc aswell as on a horizontal and or vertical.

Using these techniques I could get a 8k+ image with, a single large, or small, pixel pad ( not that it would not be desirable unless motion is compensated for). You could use a mechanism to scan the image on the stationery pixel. So many camera mechanisms, so few good ones.

Of course you can use the natural shift and motion in the scene to get extra resolution and depth information. The previous analysis techniques enhance this. They started using these techniques to restore old film years after, however at magnitudes more processing power. I claim these versions of these inventions.
 
Jannard said:
http://www.reduser.net/forum/showthread.php?65086-I-want-to-tell-you.../page2

Jim
Click to expand...

Hi again Jim.

I'm just looking for a cheap personal camera to carry around in the car, rather than dragging the scarlet around everywhere. My car has an amusing feature, a long distance remote that if you bump the button in your pocket unlocks the car or pops the boot. Rather nice feature for a GM Cruze turbo diesel. So I rather have something cheap (or possibly go down to the dealer ship and annoy them by asking for two big holes in the boot that I can stick a 4 inch plus pad lock through (for internet publicity shots on how well their remotes work), why can't people design things properly these days) ;) .


The Lytro mechanism has some good video work features though. You literally can have a camera that you can take out your pocket and hit record for any sudden infield action scene and worry about focus in post. Probably will be advantageous to infield journalists and war correspondence teams.

At the moment there is only a few cameras to consider. Apart from this one (not pocketable and no video, but I had considered approaching them about doing a cheap professional version before). The Sinclair Design POCO series, that has a credit card sized fullhd p60 and 3D cameras. But problem is it uses OmniVision sensors. If only they had used a top grade Sony sensor or similar size it would have been fine, or the greater than fullhd resolution modes. However, I can assure you that many of his products look like items you would carry around yourself. If you ever were going to do a cheap camera, or redesign the cases, he would be worth calling.

www.iainsinclair.com/poco_pro_tech_spec.html

The top Sony pocket camera is another one.

There is supposed to be greater than fullhd resolution android ambarella Ione based solutions to come.

Past this cheap setup (now Sanyo is taken over) we have the Sony nex-5n and gh2 (or newer) If only the hackers had concentrated on shd 10 bit raw or 4:2:2, I might have reconsidered the scarlet.

These are worthy little 'playmates' to the Scarlet as personal cameras, not ideal, but that is why they are not making Scarlets.

So Jim, this Light Feild like feature, is that an upgrade, or firmware upgrade, to first gen Fixed Scarlets (or is it only mount Scarlets)? Because, THAT, would be so cool. I had considered doing a lens adaptor for the fixed Scarlet and using a isee3D enabled lens in high frame rate mode before I heard this?

Thanks again.


Wayne.
 
I assume we'll be able to get Depth maps out of lytro eventually. It could make set surveying interesting if you had essentially a pocketable kinect. The sensor is so small though I have doubts about its accuracy.
 
There are ways around this Gavin, see techniques I came up with above. But your set surveying, that is an idea I had in mind years ago. I did an analysis of production workflow and figured how to replace every body in additional to the single operator/director and talent (my AI research is not active, though I am getting flashes of ideas at the moment in how to do it simply, even more simply again now, but too simple). A few years ago I turned my attention to figuring out how to completely set a set up exactly the way it was, even automated, allowing it to be easily packed away (slid away even). Carousel rotational stage was really interesting, allowing half a dozen or more sets to be set up in a fraction of the space. I could probably easily figure a way to automate set construction, I have ideas flashing through my mind at the moment, Intel and StarTrek fans will be happy ;) . The main problem in the lesser 3d set, is lighting interaction, I came up with a light sampling solution years ago to increase, there's that tooheys extra dry stag beer commercial on again, so cool. To increase accuracy and interactivity so the set did nit look fake. My proposal for a public domain 3d object library, would allow all set objects that did not need to, or had limited interaction, to be computer generated. If somebody invested a few million, I estimate I could generate billions a year. I have researched things far in advance of anything Red is considering. Why settle for cameras when you can make more money for equipment revolutionizing all areas of production and delivery. My treated h health levels are up at the moment and I'm getting bored of getting to do nothing.
 
You must log in or register to reply here.
Back
Top