I tried extracting non-metric camera parameters with colmap and running gaussian_splatting.py and got bad results. The mesh is all over the place and the render result is not output from the model, but from the ground truth.
I ran it back to the original 3dgs and confirmed that the camera parameters are fine, is the 3dgs you put together only assuming a metric based camera?
And what is the error that the ground truth is showing up in the render image?
I tried extracting non-metric camera parameters with colmap and running gaussian_splatting.py and got bad results. The mesh is all over the place and the render result is not output from the model, but from the ground truth.
I ran it back to the original 3dgs and confirmed that the camera parameters are fine, is the 3dgs you put together only assuming a metric based camera?
And what is the error that the ground truth is showing up in the render image?