I was thinking about the challenge of an indy filmmaker getting cheap shallow depth of field. What if you just had a little sensor, something resembling a lav mic, on the person that you wanted in focus. The camera (which would also have a sensor) would then be able to record that distance (onto an SD card or maybe even onto the tape). Then, when you go into your effects software (after effects let's say), there's a plugin that analyzes that data, knows from the sensor data where the subject you want in focus is, and allows you to blur anything in front of or behind it to your heart's content. From a techical standpoint, is that possible?
is this a genius idea or totally stupid?
- Thread starter c.g._eads
- Start date
Batutta
Major Contributor
It's possible with a still camera, don't even need a sensor, and you can refocus to ANY part of the frame. This is wild--
http://graphics.stanford.edu/papers/lfcamera/
Make sure to watch the AVI at the bottom of the page.
http://graphics.stanford.edu/papers/lfcamera/
Make sure to watch the AVI at the bottom of the page.
Rich Lee
I shoot Pirates.
the problem is there would be no way for the sensor to know the size and shape of the object you want in focus...a sensor would just be a single smal point in space relative to the object you would want to keep in focus...
what you are really looking for is something called a Z-buffer image...also known as a depth map.
kinda hard to describe...but basicly its an image that grads off form white to black...with shades of gray in between...te closer to camera the brighter the object...te further away an object if the darker it becomes. or vise versa
http://www.ldv.ei.tum.de/studium/arbeiten/depthmap.png
http://www.ocpstudios.com/images/samprenders/ITTOASE-Depth.jpg
there are a few cameras that can record this sort of image but its a bit dodgy. but it will get better eventualy.
the tech is already in place to use these sorts of images for DOF...its used all the time in the VFX world.
what you are really looking for is something called a Z-buffer image...also known as a depth map.
kinda hard to describe...but basicly its an image that grads off form white to black...with shades of gray in between...te closer to camera the brighter the object...te further away an object if the darker it becomes. or vise versa
http://www.ldv.ei.tum.de/studium/arbeiten/depthmap.png
http://www.ocpstudios.com/images/samprenders/ITTOASE-Depth.jpg
there are a few cameras that can record this sort of image but its a bit dodgy. but it will get better eventualy.
the tech is already in place to use these sorts of images for DOF...its used all the time in the VFX world.
What would be rad..... is a FF unit that can follow the actor automatically.....
MiniMan13
Well-known member
Someone really smart could rig up a IR sensor that measures the distance from the end of the camera to the IR reciver on the actor, then have a small on board computer to calculate the focal distance, and adjust the whip.
Drew Ott
Wish I were banned.
I've thought about something similar to this but probably 200 years ahead of it.
There would be a flash (almost like an electron microscope) and a computer would see where each photon's path ended. From this it seems possible that some amazing computer/software combo could create a 3d image of the environment.
This would be ridiculous (as you could imagine). Especially if you could create a 3d environment with real-life motion.
If you imagine that you could light everything within 3d editing software, amongst thousands of other things.
There would be a flash (almost like an electron microscope) and a computer would see where each photon's path ended. From this it seems possible that some amazing computer/software combo could create a 3d image of the environment.
This would be ridiculous (as you could imagine). Especially if you could create a 3d environment with real-life motion.
If you imagine that you could light everything within 3d editing software, amongst thousands of other things.
Bruce Morgan
Well-known member
This would also be an excellent device for returning freedom to the actor and the director .
Excellent question and answers so far !!!!
Excellent question and answers so far !!!!
Matt Grunau
Adobe Demigod
If the sensor put out a something like a weak radio signal only and the receiver determined signal strength as distance to the camera, it would work. You wouldn't even have to see it. You could have multiple actors all wearing them, each with a different frequency, and the camera could focus from subject to subject with just the push of a button. And it could be done all externally, with only the final signal hitting the cam.
Cool links you provided too.
mikkowilson
Steadi-Guru
Cinetape
You're not so far ahead actually...
Watch this video: http://video.google.com/videoplay?docid=-3656494784112768834&q="3d+video+camera"
Towards the end, they do a live demo, where you actually do get to watch the photons as they travel down a hallway!
- Mikko
If someone could engineer it - amazing. the sensor could effectively be in the focal plane - then plug in the dof in to your NLE and done. Theres probably one thing that wouldnt look quite right. We all know what a blur effect looks like - is it as nice as bokeh on a nice lense? No.
Beat Takeshi
Two-Hip Shakur
You could you a z-buffer kind of signal because if you average the depth of the average person, say like 3 feet to keep it safe, and then tell the z-buffer to start working at 2 feet behind that. Then you just use a regular distance finder to get the closest point of the person to the camera. The average distance should be manual adjustable in feet where you can just expand or slide it back depending on the amount of people in the shot. The only thing is that the width of the person might be hard to get perfectly. Especially where fly away hair comes in where it would be hard to get a reading on.