Would desperately need advice about a simple video installation

Hello kind people of VVVV!

I have what I think is probably a relatively simple idea:

  1. A static wide angle camera on the side wall of a room recording people as they pass through the room.

  2. A projector (or a large screen) projecting the recorded image on the same side wall creating a kind of “mirror” effect.

Pretty simple up to here I just need to connect the camera and projector and deal a bit with setting up appropriate lighting, angles, etc.

BUT HERE IS THE CATCH:

I would like the projected image always to be the average (median value) of the last 30 seconds of recording.

So if you just pass through the room you are almost invisible you just see the motionless objects projected clearly. If you stand still for a certain amount of time your “reflection” slowly starts to become visible (for example after 15 sec. you would be kind of 50% transparent. and after 30 second of standing motionless you would be completely “reflected”).

I would call it something like: Be here now… (the idea is just to get people to stop and be still for some time and then to slowly discover their own reflection. a kind of mindfulness helper.)

UNFORTUNATELY I HAVE NO PROGRAMMING EXPERIENCE AND WAS NOT ABLE TO FIND THIS KIND OF EFFECT ON EXISTING VIDEO/VJ SOFTWARE. COULD SOMEONE PLEASE HELP ME??

Any suggestions on software, points in the right direction, existing scripts, or offers to collaborate on the “software” side of the project would be most appreciated!!

THANKS IN ADVANCE!!

JASA

Should be pretty easy to do. Some texture queue and averaging. Can patch youvsomething when im back home if someone else didnt allready did it ;)

How do you plan to do averaging part?

in DX9 we have neither floating point textures nor texture arrays to do it single pass

in DX11 I have difficulties with Texture2D queue

echo (dx9 filter) does that its in the addon pack afaik

Just simple low-pass filter. How can’t I guess that? Amazingly simple stuff %)

My quick solution is so complex and resource-hog in comparison to this.
I use DX9ToDX11 node to bring video in DX11 world which has floating point textures. Because DX9ToDX11 doesn’t support spreads, I take only first and last frames in queue. And then I add new frame and substract the oldest one.

Anyway attaching it. Do not use it, Echo node is so much better.

awful way to do echo (7.7 kB)