After installing you will find the settings on the ride side in the "n" panel. First select an output directory and create a bounding box, then create a camera set by hitting "Create Camera Hemisphere / Central Object" or "Create Camera Large Scene / Indoor". Next create a point cloud with "Create and Export Point Cloud" and prepare the scene for rendering with "Prepare Render Settings". Finally render the camera frames as an animation in Blender and export the camera alignments with "Export Camera Files". Now you can use the output directory with the files points3d.ply, transforms_train.json and transforms_test.json as an input for many NeRF and Gaussian Splatting frameworks.

You can use the free version of this addon for personal and research projects at the moment, we may change the licensing and the price in the future.

It is only one last step, you need to start the training process via your preferred framework. We suggest 3DGS or 2DGS. The command to run the training typically looks like "python train.py -s PATH/TO/THE/GENERATED/FOLDER" For more info on how to set up a gaussian splatting framework read this: https://github.com/graphdeco-inria/gaussian-splatting

Please keep in mind, our add-on generates all files needed to train a Splat or NeRF of your scene but you have to start the training yourself.

Go to "Edit - Preferences - Addons" and in the top right corner hit "install from file". Navigate to the .zip and install.

Depending on the scene that can happen. Most likely you have many objects or particle systems. Try to decrease the amount of particles converted to meshes. Decreasing the amount of particles can help as well.

Check the documentation page for a full explanation of the settings.