![]() |
2010-01-09
, 00:59
|
Banned |
Posts: 291 |
Thanked: 42 times |
Joined on Dec 2009
|
#22
|
![]() |
2010-01-09
, 01:02
|
Banned |
Posts: 291 |
Thanked: 42 times |
Joined on Dec 2009
|
#23
|
Sure, sounds very interesting. If I understand correctly then the effect is done purely in software by image processing the user-facing camera image, finding the eyes, and estimating the relative location of the camera w.r.t. the user's head, and feeding that info to a 3D app as the 3D rendering viewpoint.
![]() |
2010-01-09
, 01:03
|
Posts: 127 |
Thanked: 50 times |
Joined on Sep 2009
@ Denmark
|
#24
|
![]() |
2010-01-09
, 01:04
|
|
Posts: 355 |
Thanked: 245 times |
Joined on Jul 2009
@ Northern VA
|
#25
|
![]() |
2010-01-09
, 01:06
|
Banned |
Posts: 291 |
Thanked: 42 times |
Joined on Dec 2009
|
#26
|
![]() |
2010-01-09
, 01:08
|
|
Posts: 3,404 |
Thanked: 4,474 times |
Joined on Oct 2005
@ Germany
|
#27
|
Be as sceptical as you want. The Apple innovation train has its steam at full throttle.
![]() |
2010-01-09
, 01:09
|
Banned |
Posts: 291 |
Thanked: 42 times |
Joined on Dec 2009
|
#28
|
Skeptical, I will be Mr. Phonics. The fact that you mentioned Cameron then Avatar, just shows you're grand staging some truly weak viewpoint. You do understand that special lenses must be worn to come even close to the 3D nature of that film. A secondary accessory to use a device has never been received well. If it's using the camera the same effect can be seen, watching George Washington's eyes on a dollar bill as you move it around. Umm yeah awesomeness indeed OrangeChode.
![]() |
2010-01-09
, 01:11
|
|
Posts: 3,404 |
Thanked: 4,474 times |
Joined on Oct 2005
@ Germany
|
#29
|
The latter. Still, a HUGE awesomeness to see depth on a screen without using special glasses, don't you think?
![]() |
2010-01-09
, 01:15
|
Banned |
Posts: 291 |
Thanked: 42 times |
Joined on Dec 2009
|
#30
|
This means it could be done on the N900 (or other phones) as well, as long as they have user-facing cameras. Maybe OpenCV might have some routines that could help with programming this. It's an interesting problem. I think you can assume an average size for a human head, then given the position and spacing between the eyes, you can extrapolate the size and position of the head; given known camera parameters this allows you to compute where the head is in 3D space relative to the camera. Sounds like something where there should be enough research papers (or existing code) to get it going on the N900.
Edit: Using face tracking would allow you to get the image-space position of the head; eye tracking allows you to get the orientation of the head and eyes; an assumed size of the head + known camera parameters allow you to calculate the 3D location/rotation of the head relative to the camera. Again, sounds doable.
Last edited by Otaku; 2010-01-09 at 01:02.