Bonus

This unit used multi-view diffusion as an example to get familiar with the model ecosystem at a high level.

However, multi-view diffusion is just one of many available models used in generative 3D tasks.

Due to the rapid pace of progress in generative 3D, I want to emphasize the importance of getting comfortable with the model ecosystem, allowing you to keep up with the latest research and tools.

To do so, here are some exercises to help you get started:

1. Explore the Model Hub: Check out the Model Hub to see what models are available. You can filter by task, framework, and more.
2. Customize your Demo: In the hands-on, we created a Gradio demo for multi-view diffusion. Try customizing it by, for example, adding a Slider to control the elevation parameter.
3. Create your own Model: If you aren’t familiar with machine learning concepts, follow the NLP Course. Even if you aren’t interested in NLP, this course provides an in-depth introduction to machine learning concepts.

In the next unit, we’ll be diving into the specifics of Gaussian Splatting, an ML-friendly 3D representation and recent hot topic in 3D research.

< > Update on GitHub