Author Topic: What about making a "structure from motion" effect ?  (Read 5258 times)

Offline Memoire

  • *
  • Posts: 2
  • Karma: +0/-0
    • View Profile
on: January 15, 2008, 05:34:50 PM
Hello,

Believe it or not, during my studies in experimental psychology 15 years ago, I did some experiments that were close to those discussed here. "Close", well, I mean it was only 1D (side to side horizontal motion), and I had a little stick on my head linked to a potentiometer to code the motion, so while I was looking at a "window in another room behind the screen", everyone else was looking at something very ridiculous !

My post here is to give you an hint about a "fun" effect that uses the same mechanism as the original demo discussed here.

The point is to have an image that is meaningful only to the one who is making the motion. Everyone else in the room is looking at "noise" on the screen, while the person who wears the glasses can see a 3D form, or see a letter, a number, or a word, etc...

To do this, first choose the 3D form you want to hide in the noise. It should be quite smooth, no peaks and so on.

Then you take a 2D image of randomly distributed white points on a black background, and you give to each of these X, Y points the Z data corresponding to the intersection of the ligne of sight with the 3D object.

Please note I do not talk about texturing the 3D object with some noise, you want to have only dots on a screen, not texture that would shear during the motion etc : you do not want to help the people who do not wear the glasses to figure out what is been displayed.

This is then the 3D data you put in the original demo discussed here. The key is to add a small and slow brownian motion to the X,Y data every frame. This way the 3D effect is not lost for the one who is making the large motion, while the other spectators cannot understand what is happening.

English is not my mother tongue, I hope I was clear enough.

Waiting for your comments.
   





Offline Earlack

  • *
  • Posts: 6
  • Karma: +0/-0
    • View Profile
Reply #1 on: January 19, 2008, 01:12:33 PM
...Not sure to fully understand. I think I understood the method, BUT, I don't understand how it's displayed only to the guy in motion and not to the others. Maybe a schematic representation would help... ? :)



Offline Memoire

  • *
  • Posts: 2
  • Karma: +0/-0
    • View Profile
Reply #2 on: January 20, 2008, 03:23:14 PM
The dots are seen by everyone, but only the guy in motion with the glasses can see something coherent. He will see the 3D structure that pops out of the screen, while the other will only see moving dots. It's a bit like the autostereogram pictures full of dots : if you don't cross the eyes, you see nothing.



Offline Earlack

  • *
  • Posts: 6
  • Karma: +0/-0
    • View Profile
Reply #3 on: January 24, 2008, 03:38:47 PM
AH, ok I got it. I first understood that thoses dots weren't moving, that's why I was confused :D

I think it's possible BUT looks harder than it may at a first look. I mean, if others people must not understand the image, you can't just make the dots follow the guy to form the image ; you must also hade some randomness when placing and choosing what points will be at what position, etc. sounds pretty hard to me. But, of course, I'm no programmer at all. Looks pretty interesting tough.