Not logged in, Join Here! or Log In Below:  
 
News Articles Search    
 


Submitted by Joris Timmermans, posted on July 25, 2000




Image Description, by Joris Timmermans



This is a screenshot of my latest tinkerings with Image Based rendering. It shows the warped image on the left, and the original on the right. The gaps you see are introduced by lack of "visibility" from the original image, I'm currently looking to implement ways of filling these gaps in a satisfying way.

In my test-skeleton you can move around with the keyboard, to show the warpings for different viewpoints.


[prev]
Image of the Day Gallery
www.flipcode.com

[next]

 
Message Center / Reader Comments: ( To Participate in the Discussion, Join the Community )
 
Archive Notice: This thread is old and no longer active. It is here for reference purposes. This thread was created on an older version of the flipcode forums, before the site closed in 2005. Please keep that in mind as you view this thread, as many of the topics and opinions may be outdated.
 
fluffy

July 25, 2000, 02:23 AM

How are you determining the depths? Or are you having POVray put out a depthmap as well?

 
MadKeithV

July 25, 2000, 02:50 AM

I'm not determining the "depth" per se, I'm using POVray's distance to the rendered point to calculate a quantity called "disparity".
For more information, look at http://www.ramartworks.com/ibr/index.html, a very helpful site on Image Based Rendering.

 
malkia/eastern.analog

July 25, 2000, 10:46 AM

Can I get a demo,
I'm very interrested...

 
DirtyPunk

July 25, 2000, 11:26 AM

Okay - Here is a few ideas. First of all, you can use a LDI for warping. An LDI (or layered depth image) contains even information that is occluded, and the depth as well. It takes more memory though. Its one of the fastest ways to take care of it. The other thing you can do is multiple directional renderings for warping.

Note the biggest failure of most occlusion filling techniques ends up being that the user is forced outside the convex hull of the object, LDIs don't have this problem. A set of 3 LDIs (one down each axis) can be enough for a whole scene. In fact such a configuration can represent the same info as a voxel at much lower cost - with normals calculatable easily.

Also note - you cannot warp the specular/reflective properties as well (only the normals). This must be calculated afterwards. (although some IBR techniques, such as lumiagraphs can).

 
fluffy

July 25, 2000, 11:45 AM

By 'depth', I meant 'distance to the point from the camera.' I figured that you'd understand what I meant. :P What I meant was, how are you getting the distance to the point from the camera? Did you hack up POVray to do that, or can it output an image with depth values, or what?

 
Devon

July 26, 2000, 12:33 AM

You *can* hack up pov-ay to do that, without even changing the source. Just go through and set all your objects to regular dull finish, color white, and add a black or white fog that encompasses the extents of the scene....it'sd a fun technique, and you can make those 3D eye pics using it, with some programme like hidden 3d!

 
This thread contains 6 messages.
 
 
Hosting by Solid Eight Studios, maker of PhotoTangler Collage Maker.