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.
Data Preparation
Note
It is the responsibility of each user to check the content of the dataset, review the applicable licenses, and determine if it is suitable for their intended use. Users should review any applicable links associated with the dataset before placing the data on their machine.
DreamFusion relies on a pretrained 2D text-to-image diffusion model to perform text-to-3D synthesis, thereby eliminating the need for a training dataset. We support Stable Diffusion as the backend diffusion model. Depending on your chosen backend implementation, it will be necessary to set up the model checkpoints.
These checkpoints can include:
Hugging Face pipeline: the checkpoint will be automatically downloaded at runtime.
NeMo pipeline.
DreamFusion-DMTet
Note
It is the responsibility of each user to check the content of the dataset, review the applicable licenses, and determine if it is suitable for their intended use. Users should review any applicable links associated with the dataset before placing the data on their machine.
Similar to DreamFusion, DMTet relies on a pretrained 2D text-to-image diffusion model to perform text-to-3D synthesis and doesn’t require an external database. However, the network requires three components:
Diffusion model
A pretrained DreamFusion checkpoint, used to initialize the DMTet network.
Initial tetrahedral grid: can be generated using this repo, or downloaded from NGC.