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 :)

Lytro "infinity" Selective DOF in Post

Scott Miller

Scott Miller

Well-known member
Joined
Jul 2, 2009
Messages
1,192
Reaction score
0
Points
0
Location
Carlsbad, Ca - San Diego/OC
http://allthingsd.com/20110621/meet...s-to-sharpen-focus-of-entire-camera-industry/

Very interesting tech! When I worked @ Adobe, I saw a tech demo of something similar, but much less polished. Make sure to play with the examples at the bottom of the article.

Found on this Twitter feed:

dps_normal.jpg

digitalps DP School (DPS)



Would U like a camera that you can focus AFTER you've taken the shot? Check out this soon to be released Lytro Camera! http://bit.ly/lTxa2F
 
I noticed this too. Neat. In fact, the technology should also allow you to reposition the camera in post -- at least within a limited area, and change the aperture. I'm curious to see what kind of resolution they can get out of it. The main issue with a microlens array is that the pixels under each microlens has to record the image separately, so you need a lot more sensor pixels than output pixels.
 
Sam Eilertsen said:
I noticed this too. Neat. In fact, the technology should also allow you to reposition the camera in post -- at least within a limited area, and change the aperture. I'm curious to see what kind of resolution they can get out of it. The main issue with a microlens array is that the pixels under each microlens has to record the image separately, so you need a lot more sensor pixels than output pixels.
Click to expand...
Mmmm imagine this tech in a 617 Epic.
Ng, licence it already dude!
 
Very cool. I wasn't aware that Ren Ng had devised a way to maintain resolution. Previous examples of plenoptic cameras resulted in *massive* resolution loss. So delivering high resolution output from a small camera would be a real imaging breakthrough.

One of the downsides of other previous lightfield sensors (and it appears this one as well) is that the focus is pretty Gaussian and you lose those nice clean Bokehs. But I bet you could fix that with some improved code. Just imagine, you shoot with a spherical lens and get anamorphic defocus in post. Now *that* would be something new and groundbreaking.

I guess we'll have to wait and see what kind of aliasing, haloing and artifacts this produces--but it's certainly a taste of the future where lens, lighting and framing on set is just meta-data and everything is ultimately 'captured' and rendered later.
 
everything gavin mentioned, of course.

for now (besides it not being out) the biggest problem would be the ability to capture high enough resolution imagery at enough frames/second. i'm sure there may be other problems to solve before this type of tech is ready for movies... we'll see.
 
I can't see a huge demand for this on the pro level any way. If you can get your focus and framing right the first time, you don't need it. Although potentially it could allow you to "put" the camera somewhere you can't physically place it.
 
I Bloom said:
Are there any other tradeoff's? Any reason why this tech couldn't be used for movies?

IBloom
Click to expand...

Not that I'm aware of, although the algorithms that reconstruct the image might artifact differently based on slightly different images. So halos might flicker for instance.


Sam Eilertsen said:
I can't see a huge demand for this on the pro level any way. If you can get your focus and framing right the first time, you don't need it. Although potentially it could allow you to "put" the camera somewhere you can't physically place it.
Click to expand...


You're looking at it all wrong. It's more than just choosing focus in post (although the option to perfectly nail an action sequence on a long lens would be nice, not to mention eliminating the need for a focus puller period could save you some $$) you have to start thinking about what this creatively offers you. You could do a split focus or tilt shift effect potentially with less artifacts.

You also get some primitive depth cues and could potentially use it to extract mattes for grading.

You could be shooting "wide open" in a dark scene but still have a deep depth of field. You could have an infinite depth of field. You could stop down for exposure but still have shallow DOF without using an ND. If you had a large sensor camera like the RED 617 you could shoot 3D with a single lens. You could get rough 3D geometry of your scene perfectly matched to your RGB color image. That means you could wander around set with a camera and capture the entire set as a color 3D model. The possibilities are endless and far beyond just adjusting focus--although there is a lot that could be done creatively with that too.
 
Sam Eilertsen said:
I can't see a huge demand for this on the pro level any way. If you can get your focus and framing right the first time, you don't need it. Although potentially it could allow you to "put" the camera somewhere you can't physically place it.
Click to expand...

focus pulling isn't always easy, i've seen it hold up production. think about the crazy steadicam moves and everything else (like you mentioned, tight spaces, etc). i think it would be a nice option to be able to do focus interactively on set, but then have the ability to very easily adjust it... that could be cool.
 
IMO this could theoretically be a better way to do 3D motion picture imaging.

Now we need a micro-holographic optical computer with optical storage to manipulate the data and record it.
 
Hold on a second I'll enhance it. http://www.youtube.com/watch?v=Vxq9yj2pVWk

I've seen some of the best ACs in the industry have problems pulling a shot where camera movement, subject movement and a wide open lens together just make it humanly impossible.

Consider this: as sensor size increases pulling focus becomes increasingly more problematic. It may get to the point where you need something like this, just to work in a particular resolution.

IBloom
 
I Bloom said:
Consider this: as sensor size increases pulling focus becomes increasingly more problematic. It may get to the point where you need something like this, just to work in a particular resolution.
Click to expand...

Hopefully the sensors get more sensitive so that a decent stop is achievable for a manageable DOF even on the bigger sensors.
 
I posted this in the other thread, but thought I'd bring it to where the real party is.

Mike P. said:
It's like the FCPX of digital photography sensors!!!...

All joking aside, my only concern is that it ends up costing a small forture to get it into bigger APSc or FF sized sensors, where post-focusing makes much more sense (artistically and functionally). Great tech, but still probably 10years out before it makes its way to (low-res) video cameras if at all possible... My guess is it'll be in phone cameras sooner rather than later; they really should have sold out to one of the bigger guys as they could have pumped the tech into their entire consumer line-up.
Click to expand...
 
Like Gavin said, it's about more than focus. And yes, it can be used for 3D -- at the launch party, they had a bunch of 3D TVs set up with pictures they'd taken -- single camera, single lens 3D with impressive separation. Way better than I would've thought possible, had to see it to believe it (there's also a brief clip showing off the parallax possible in the system on techcrunch -- http://techcrunch.com/2011/06/21/ly...m-photography-with-50m-in-venture-funds-tctv/).

I'm pretty sure this will become standard or semi-standard at all levels of production, if only because there's diminishing returns on what you can do with all those extra pixels. Capturing a light field opens up some amazing, revolutionary things you can do with the image.

As I mentioned in the other thread, rumor has it that this technology was shopped to Red at some point, but the company declined it. Maybe we'll see it in the future, though.
 
Very fascinating technology. Is it just me or do all the photos seem soft unless the focus is close? Looking at the company site, I see many images with sharp focus, but only up close. Pretty soon, we won't need cameras at all, just click on the time and space button and the image will appear from a big eye in the sky.
 
Not acting as spoiler.
Just checking if this is our big guy!
Amazing new technology.
Check their site out.
Lytro!
Lytro on June 22, 2011 at 2:11 pm said:

Thanks Jared – will keep both your initial and follow up posts “un-public.”
 
I had this idea the other day...

What if you could make 3D video where your eye movement is tracked so that whatever you're looking at is in focus?
That would be quite immersive I think. :)
 
Luis de la Cerda said:
I had this idea the other day...

What if you could make 3D video where your eye movement is tracked so that whatever you're looking at is in focus?
That would be quite immersive I think. :)
Click to expand...

It would require one helluva of a processor.. but I love it! That's a genuinely unique and interesting application I've never heard someone suggest before. I wonder if it would feel 'right' in practice. Unfortunately though it still wouldn't fix the "Focus Point/Convergence Point" discrepancies unless you also shifted the convergence based on where your eyes tracked. That I imagine though would feel weird and will require some sort of advanced optics in the future.
 
You must log in or register to reply here.
Back
Top