Tweet
I came up with an idea earlier this morning- it's probably nothing new. But I was wondering why one couldn't take a stereoscopic image (let's say from a stereoscopic camera) and automatically 'convert' the objects within the scene into 3D models. I realize that this wouldn't be completely 180 degrees - but it would work for alot of things that you don't need backfaces for. Right now it's possible to export from Max to 3D using some third party programs.
-
stereoscopic photography is used to get very detailed and geometrical correct 2d AND 3d drawings of fassades.
It's a standard in GIS processing... -
You might be interested to know there's software availble to do this sort of thing. Haven't used any myself, but they all seemed fairly impressive when I've seen them demonstrated at trade shows etc...
Here's a link to the first product I could find: http://www.photomodeler.com/Comment
-
We have it nad it ai't all that great....i'm no expert at the software but i found it to be more trouble then it was worth.-----Dwayne D. Ellis-----Comment
-
For some reason last opinion seems to me to be true. As I figured out from photomodeller's tours, most work is be done by hand, not automatically. So, I believe, it's just another form of money grabbing.Comment
-
I've seen photomodeler and im not refering to that (which involves taking a series of 'regular' photos. I believe they have a demo you can try.
But I'm talking about developing 3d models from stereoscopic images- something that should be much quicker and simpler without backfaces. As for GIS- excuse my ignorance but what do you mean by that?Comment
-
GIS = geographical information system
When you get a drawing from your site with all informations about height and landscape it was created with a gis system like arcgis ( http://www.esri.com/ )
Beside the fact that this is a bigger structure , a similar system can be used to get 2D drawings from a fassade, including 3d informations!
The software is very high priced ..the hardware too.
If you only try to model a 3d object like a plasticduck then go with photomodeler.
The problem with stereoscopic pictures is, that you get UV-coord and have to calculate from them all three XYZ coord...
a more interressting system is this technique:
http://www.gisdevelopment.net/aars/a...m/phm002.shtml
..have a look at page 7
Comment
-
Comment
-
Hmmmm. Thanks for the heads up. My brother uses ArcGis in conjunction with his research- I will have to ask about it.
As for the links Dschaga. They're interesting and thanks. The first one is similar to the way photomodeller works. The second one is more on the mark with what I had in mind- but obviously need more work to be done.Comment
-
This is pretty common, I have seen it in a lot of medical apps. basically makes a depth map.Two heads are better than one ...
....but some head is better than none.....Comment
-
the problem with this technique is that the accuracy decreases as you go further in depth.
if you want to calculate the distance of a point thats 200m away your 2 cameras have to be much farther apart than for an object 1m away.
for objects closeby its working and already used
the main point is that you have to use the difference in perspective between 2 pictures to calculate the depth
for objects close to the camera, the change in perspective is quite large even when moving the camera only a bit (1m)
for objects far away the difference is minimal, making things very complicated
an example of how its used (they move the camera up instead, but the technique is the same):
http://www.spheron.com/digitalforens...asurement.htmlComment
Comment