Hacker News Reader

Comment by ctdonath | Hacker News Reader

Suggestion: Use "photosynth" techniques to fuse the video frames from the moving cameras into a 3D model of the room.

That is exactly what the video shows.

EDIT: When I read photosynth, I was thinking of Microsoft Photosynth[1], which sketches pictures into panoramas. I should have read the parent comment a little more closely.

1: http://photosynth.net/

It shows multiple angles from one viewpoint being assembled into a panorama. That's very different from taking images from multiple viewpoints and generating a 3D model.

Agreed. Original comment edited.

If these balls are reusable and/or cheap enough, throwing or shooting several of them in and making them and head unit communicate to produce 3D panoramas seems like next logical step. But I'm not sure 3D panoramas are worth it. What is the benefit of having one 3D image instead of just several 2D images of the rapidly changing scene? This technology is not meant to be used in museums.

To get a feeling for the layout and relative positioning of things. Sure it's possible to do by examining several 2D pictures, but interpreting 3D environments is far more natural, and it would allow the operator to rotate and so on to get the best angle available.

Also, since the camera is moving, one doesn't need to use multiple.

The video shows a mockup panorama. You can tell because the perspective is from 5-6 foot, not 2-3 inches.

Reply on news.ycombinator.com