Z-buffer cameras (Real time and apparently TTL)

Chris Millar · May 9, 2011

I had an idea the other day.. and as always google leads me to others who got in there before me:

http://www.inition.c...D_=53&tab=blurb

http://pro.jvc.com/p...el_id=MDL101309

Still wondering though - what happened to these systems ? Why aren't hearing about them anymore ??

Maybe we are, I just don't know the terminology (?)

Will Earl · May 11, 2011

I think it's most likely a case of a technology that wasn't quite good enough at the time to be of any practical use.

Assuming these were released early to mid 2000s - in which case stereo3D wasn't the big deal it is now, for VFX purposes SD resolution wouldn't be good enough to work with and (if working with depth images in CG is anything to go by) depth keying produces results of varying quality - even in broadcast situations keying via green or blue would have produced better results.

If you shoot in stereo you can extract depth from an image. Applications like Nuke's Ocula http://www.thefoundry.co.uk/products/ocula/ allow you to create depth maps from moving images, while applications like Stereoscan http://www.agisoft.ru/products/stereoscan/ allow you to extract 3d meshes from stereo images.

Chris Millar · May 12, 2011

I think it's most likely a case of a technology that wasn't quite good enough at the time to be of any practical use.

Assuming these were released early to mid 2000s - in which case stereo3D wasn't the big deal it is now, for VFX purposes SD resolution wouldn't be good enough to work with and (if working with depth images in CG is anything to go by) depth keying produces results of varying quality - even in broadcast situations keying via green or blue would have produced better results.

If you shoot in stereo you can extract depth from an image. Applications like Nuke's Ocula http://www.thefoundr...products/ocula/ allow you to create depth maps from moving images, while applications like Stereoscan http://www.agisoft.r...cts/stereoscan/ allow you to extract 3d meshes from stereo images.

I wasn't actually thinking about the usage for 3D output, but of course yeh, makes sense - but in saying that it further begs the question why some group hadn't thrown its chips in this products direction just for the sake of doing it differently... But even for 2D effects it could have been interesting - faking low DOF, mist effects in post etc... all the basic stuff done with Z-buffers with CGI but with real footage ?

Would you agree in principle that some production or industry found a very neat application for it and for want of a better phrase bought it out and have been keeping the process mum since ?

Pretty much if SD can become HD in the time period in question then this process could have also followed a similar increase in resolution and efficacy

Keep in mind I'm no expert on this stuff - just interested in the possibilities of combining technology into new 'bigger than the sum of its parts' ideas (nothing new in that in the cine world though huh )

Will Earl · May 13, 2011

There are plenty of potential applications for it, just look at all the various projects for the Xbox Kinect http://openkinect.org/wiki/Project_ideas and http://openkinect.org/wiki/Gallery to get an idea. Microsoft are even going so far as to release a software development kit for the kinect http://research.microsoft.com/en-us/um/redmond/projects/kinectsdk/.

In terms of filmmaking - certainly it could be used to create z-buffer effects - focus, fog, etc. You could also certainly do very simple mocap, re-lighting or even things like body-reshaping http://www.youtube.com/watch?v=zXSj4pcl9Ao. It's possible you could also create stereoscopic films out of mono footage will a more ease than the process requires now.