How to adapt the code for new scenes with poses?

[EchoTHChen]

I have several datasets(wide-baseline) which are captured about 1.0 meter between two neighbor views. I have the poses of the datasets but I do not know how to adapt the code to run on my own data with poses. Could you tell me how to revise the code?

[tedyhabtegebrial]

Hi, you will have to update the following functions: cartesian_2_spherical, spherical_2_cartesian, spherical_2_equi, and equi_2_spherical in the src/utils.py file. These functions need to change based on: the way the coordinate systems are defined in your dataset and the way $\theta$ and $\phi$ are defined on the Equirectangular projection image. For the coordinate system details in our existing datasets, please check datasets.

[tedyhabtegebrial]

Lastly, make sure you are importing and add your new dataset class in the get_dataset function in data/init.py.

[EchoTHChen]

Thanks!