I have very important question for me. I would like to use Autodesk Reality Capture API in my app. I read the documentation to API but I did not find it. I know the position of camera and i would like to send this information to Reality Capture API. For example circle was divided in 24 parts. So I know that each photo was made every 15 degrees. Is there any parameter which gives me possibility to provide the position of camera?
Asked
Active
Viewed 72 times
1 Answers
0
There is no way of passing this kind of information to Reality Capture API (at least no official way) and even if it is debatable, there is not much use for such input.
Roughly speaking, the engine will “stitch” the given images based on common pixels/regions/patches. For complex objects, each 15 degrees might not be enough to capture the complex geometry and you will have to add more photos aiming that specific region.
The main benefit is that you can process your images, get the result, see the missing or low detail spots, take a bunch of photos of those specific spots and add them to the project, process your project again and repeat till you get a satisfying result. From this perspective, the "rule" of photos taken each 15 degrees will break very fast.
If you are getting wrong results, 80% of the time (again the Pareto principle) this is caused by missing the scenetype parameter, which defaults to aerial, when usually people expects to use the object type.
Check The Hitchhiker's Guide to ... Reality Capture API for more details.
denis-grigor
- 505
- 5
- 9