How Autodesk's Project Bernini Enhances Generative AI Capabilities in 3D Modeling
What is Generative AI?
Generative AI refers to algorithms that can generate new content based on the data they’ve been trained on. Unlike traditional AI, which is often used to classify or predict data, generative AI can create something new. This technology has been used in various fields, from generating human-like text with models like GPT-4, GPT‑4o, and tools like DALL·E to creating realistic image generation, advanced data analysis, web browsing, and more.
According to Autodesk’s 2024 State of Design & Make report, 78% of business leaders think AI will boost their industry, and 79% agree that AI will foster creativity within their field. In this blog post, we’ll dive deep into the technology behind Autodesk’s Project Bernini, and explore the benefits it brings to the table.
The Technology Behind Autodesk’s Project Bernini
Autodesk is well-known for its pioneering work in 3D modeling and design software. From AutoCAD to Revit, the company has consistently pushed the boundaries of what’s possible in digital design. Their latest innovation, Project Bernini is an experimental generative AI model designed to quickly create functional 3D shapes from various inputs such as 2D images, text, point clouds, and voxels. The model is trained on a diverse dataset of 10 million shapes, including CAD models and organic forms, and aims to enhance 3D modeling accuracy and productivity across industries like architecture, product design, entertainment, and more.
Autodesk intently focused on generating functional 3D structures because the articles built or manufactured from the outputs of these models must work in the real world, serving the designer’s purpose.
A reference to the example below would be a water pitcher. Many other 3D generative models might produce shapes that look like pitchers with textures that improve their superficial appearance in a very specific lighting environment. However, the Project Bernini model generates shape and texture separately and does not confuse or meld those variables. So, water pitchers generated by our model are hollow in the middle and could hold water, as any real-world pitcher would need to do.
Key Features and Applications
- Generative AI Model: Project Bernini uses a generative AI approach to create multiple design variations, enhancing creativity and functionality in 3D modeling.
- Inputs: The model quickly generates functional 3D shapes from a variety of inputs including 2D images, text prompts, point clouds, and voxels.
- Training Data: The model was trained on a dataset of 10 million shapes, including CAD models and organic forms.
- Industry Applications: The technology is intended for use in architecture, product design, entertainment, filmmaking, and other fields where functional 3D structures are required.
- Current State: The model is still experimental and not available for public use. Autodesk intends to improve the technology and investigate collaborations to advance its use in practical design and manufacturing applications.
Future Development
- Integration with Existing Products: Autodesk has not announced when Project Bernini will be integrated with its existing products or services.
- Training on Larger Datasets: The technology will become more valuable and impressive when trained on larger, higher-quality professional datasets, such as those for car designs or game characters.
Conclusion
The Autodesk team is currently working on refining their new AI technology. They are also actively reaching out to potential partners to further study and improve the performance of generative models for design and making. While Project Bernini leads in generative AI, its models will gain greater utility and appeal when trained on larger, higher-quality professional datasets. The Bernini model is designed to offer efficient training and effortless fine-tuning, facilitating this enhancement.
Stay tuned for future updates as we continue to research cutting-edge AI technology that helps our website visitors design and make anything.
