Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Feature Request] Import Point Cloud/Gaussian Splatting Cameras #1179

Open
equilento opened this issue Jul 26, 2024 · 4 comments
Open

[Feature Request] Import Point Cloud/Gaussian Splatting Cameras #1179

equilento opened this issue Jul 26, 2024 · 4 comments

Comments

@equilento
Copy link

I was wondering if it would be possible to import cameras (positions) inside PlayCanvas since I think that would greatly with positioning movement.

@slimbuck
Copy link
Member

Hi @equilento,

Where are these camera positions stored?

Gaussian splat PLY files have the ability to store such data actually, but training software doesn't write out this data unfortunately.

@equilento
Copy link
Author

I can export cameras from Metashape that correlate with point cloud from which gaussian splatting is generated.
Here are the export cameras options:
Screenshot 2024-07-29 at 14 47 14
And this is how exported Cameras and Point Cloud look like (I use a simple script) which I then import inside Postshot:
Screenshot 2024-07-29 at 14 52 29
Do you think anything can be done with this?

@slimbuck
Copy link
Member

We could perhaps add support for loading cameras.json files. These are the json versions of cameras.bin file. This would be quite a general solution, because you can use colmap itself to convert the bin files to json format.

I'm not sure if Metashape supports the json format though, are you able to confirm either way?

@equilento
Copy link
Author

I don't think Metashape supports json format, but if the conversion from bin to json is easy in colmap that's a solution I guess.
Let me just say why I think this would be a good feature. If one knows camera positions then they can be used as camera reference which could be very helpful for virtual tours for example. That way, we can just use one height (camera trajectory) and there's no need to have the whole scene covered with multiple camera heights, which in result would give us bigger areas. It is not a problem if the area covered is all the same height, because then we can just use camera tripod, know the height which is not changing and set that as y coordinate. The problem is if the area is not uniform in height, but that could be resolved with importing cameras.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

3 participants