PRINCIPLES OF 3D

 

PRINCIPLES OF 3D



3D CAMERA RIGS:

We began by testing with 2 HD Samsung video cameras.


  1. •The distance between the 2 lens apertures was greater than 65 mm

  2. •When these cameras were placed close together they could not be rotated easily to adjust focus (convergence).

  3. •The distance between the cameras is ideally 65 mm which is the average distance between the pupil of the left and right eye.

  4. •Only one output could be seen at a time


We then used 2 HD Flip video cameras.


  1. •Built a new rig to support the 2 flip cams

  2. •Both outputs could be seen at the same time. This allowed us to adjust focus (convergence)

  3. •We marked grid lines on the flip displays using some tape and marker to help us converge on a predetermined plane of focus.

  4. •Everything in front of the plane popped out, everything behind the plane sunk in.

In both cases we had to manually start the 2 cameras at the same time, there was some delay but this could be fixed in post processing. The 2 outputs were merged using anaglyph filters.

 


FOCUSING IN 3D:

The testing we did with the flip cams helped us to better understand plane of focus.

  1. •The human brain can only focus on one object in space at a time, the eye muscles adjust instantly depending on what you are looking at.

  2. •When any content is filmed and displayed in 3D there is a predetermined plane of focus. This is what the creator of the content wanted your eyes to converge on. When this happens 3d works well.

  3. •But when we try to focus on an object that is not in the plane of focus (front/behind) this confuses and puts strain on the eyes. This is why many people strain their eyes when watching 3D; they are focusing on objects that are not in focus in the scene.

  4. •Solution: Since we are generating the content dynamically the convergence of the 2 cameras can also be changed dynamically. We did this by moving the plane of focus to the object selected by the user. This assumes that the user selects what he/she wants to see clearly. This dramatically reduced the strain caused by 3D.

 


TEXT AND MOUSE POINTER IN STEREO 3D:

Initially we had both the text and mouse on a 2D plane right in front of the camera.  As mentioned in the focusing documentation, objects in 2 different planes cannot be viewed clearly at the same time. When you try to look at one plane, you see a double image of the other.


  1. •Solution: We decided to put all the text labels in 3D space. Now the labels appeared on a card just in front of the box it was supposed to point to. This made it easier to read the text.

  2. •Mouse: Due the large amount of data onscreen there was a drop in frame rate of the visualization. We decided not to create a mouse pointer in 3D space, as this would make the flow of the mouse pointer dependent on the frame rate of the visualization. This makes it difficult to select objects when the data is zoomed out, but has very little inconvenience when zoomed in.

< BACK TO DOCUMENTATION PAGE