Important

NeMo 2.0 is an experimental feature and currently released in the dev container only: nvcr.io/nvidia/nemo:dev. Please refer to NeMo 2.0 overview for information on getting started.

Training with Predefined Configurations

DreamFusion trains a Neural Radiance Field (NeRF) using a 2D diffusion model as a prior. The recommended configuration can be found in the conf/training/nerf/dreamfusion directory. You can access and modify the parameters to customize the hyperparameters according to your specific training requirements.

DreamFusion-DMTet

DreamFusion-DMTet fine tunes a DreamFusion checkpoint using a DMTet model. The recommended configuration can be found in the conf/training/nerf/dreamfusion-dmtet directory. You can access and modify the parameters to customize the hyperparameters according to your specific training requirements. You will need to provide a pretrained model using DreamFusion in the model.resume_from_checkpoint parameter, and the initial tetrahedral grid using the model.renderer.quartet_fileresume_from_checkpoint parameter.