We have been working with geodesic domes and projection mapping for years and are finally getting pretty decent with the results.
Depending on the dome and the specific setup there are a few ways to do it. With this post I would like to touch on the method we used and also see if there are any other ways to do it.
First of all we are using multiple projectors. In this case I am showing we used x7 projectors with a 1:1.09 throw.
We then had a "flat" image of the dome struts as a test pattern. This dome had the membrane on the outside so we could leverage the cool architecture and have something to map onto.
The trick is to use enough mesh warping points to adjust all of the triangles and get them super precise. This can be tricky as the mesh warping is in a rectilinear grid while most all geodesic domes are an icosahedral grid. If you throw enough points at it you will eventually get super close.
When it comes to blending the projections either the "soft-edge" approach or blending them along the struts seems to work great.
See attached photo.
////
Okay, next steps.
The above approach can work well though it can also take HOURS.
I have been researching some other ways of doing this and it seems the automatic camera blending is a good way to go. But! Most of these approaches use walled garden types of hardware and software. Specifically proprietary cameras, hardware, and software. I get it is very specific but not being able to easily pipe in a Syphon feed is not cool!
With that in mind does the MadMapper team, or any other user, have a way to achieve any kind of camera based automatic calibration for full domes with a camera?
I recall there is the MadMapper Spacial Scanner but IIRC that does not handle multiple projectors for a dome per se.
Anyways, just thought I would put these threads out there for those in the future. If we improve on the technique we will post here.
Cheers!
Depending on the dome and the specific setup there are a few ways to do it. With this post I would like to touch on the method we used and also see if there are any other ways to do it.
First of all we are using multiple projectors. In this case I am showing we used x7 projectors with a 1:1.09 throw.
We then had a "flat" image of the dome struts as a test pattern. This dome had the membrane on the outside so we could leverage the cool architecture and have something to map onto.
The trick is to use enough mesh warping points to adjust all of the triangles and get them super precise. This can be tricky as the mesh warping is in a rectilinear grid while most all geodesic domes are an icosahedral grid. If you throw enough points at it you will eventually get super close.
When it comes to blending the projections either the "soft-edge" approach or blending them along the struts seems to work great.
See attached photo.
////
Okay, next steps.
The above approach can work well though it can also take HOURS.
I have been researching some other ways of doing this and it seems the automatic camera blending is a good way to go. But! Most of these approaches use walled garden types of hardware and software. Specifically proprietary cameras, hardware, and software. I get it is very specific but not being able to easily pipe in a Syphon feed is not cool!
With that in mind does the MadMapper team, or any other user, have a way to achieve any kind of camera based automatic calibration for full domes with a camera?
I recall there is the MadMapper Spacial Scanner but IIRC that does not handle multiple projectors for a dome per se.
Anyways, just thought I would put these threads out there for those in the future. If we improve on the technique we will post here.
Cheers!
Statistics: Posted by Lumenarius — Tue Mar 26, 2024 10:55 pm
