diff --git a/README.md b/README.md index 439d546..de411cb 100644 --- a/README.md +++ b/README.md @@ -1,7 +1,7 @@ # Cube: Generative AI System for 3D

- + Cube 3D teaser showing generated 3D assets

@@ -15,10 +15,10 @@ towards this vision, we hope to engage others in the research community to addre ## May 2026 Update: CubePart
- - - - + CubePart project page + CubePart arXiv paper + CubePart models on Hugging Face + CubePart demo on Hugging Face
https://github.com/user-attachments/assets/bafcad1d-6846-4db3-b176-e079e2c84f32 @@ -31,22 +31,22 @@ Check out more details [here](https://github.com/Roblox/cube/tree/main/cubepart) ## July 2025 Update: Cube 3D v0.5 ✨
- - - - - + Roblox blog post + Cube 3D arXiv report + Cube 3D model on Hugging Face + Cube 3D interactive demo on Hugging Face + Cube 3D Google Colab demo
With the v0.5 model, we introduce two upgrades to the auto-regressive base model for 3D geometry generation from text: *higher fidelity 3D compositions* and *bounding box conditioning*. The example gif below shows the model's capacity to generate 3D shapes capturing mixtures of concepts expressed in text, for example *mechanical lobster with mechanical tank treads*. The v0.5 model also shows significantly better text adherence, for example the *lowpoly paper craft victorian rabbit*.

- + Animated examples of 3D compositions generated from text prompts

With bounding box conditioning, we observe novel 3D generations where the model balances between the two conditioning inputs -- text prompt and global aspect ratio. In the gif below, notice how the model creatively interprets the *seashell* or *tall pagoda* prompts into distinct 3D shapes. The model sometimes struggles when the bounding box is too extreme for a given prompt, for example the *cat*, where it can produce disconnected components or generates it along a diagonal to fit the bounding box constraints.

- + Animated examples of 3D shapes generated with bounding box conditioning

For a technical overview of the methods behind these two improvements, please refer to our latest v3 report on arXiv. The latest model was trained on an additional ~2.8 million synthetic 3D assets. We introduced several refinements to VQ-VAE architecture and training procedures, and increased the VQ-VAE latent length from 512 to 1024 to increase generation fidelity. @@ -74,11 +74,11 @@ The output will be an `.obj` file saved in the specified `output` directory. ## March 2025 Launch: Cube 3D v0.1
- - - - - + Roblox blog post + Cube 3D arXiv report + Cube 3D model on Hugging Face + Cube 3D interactive demo on Hugging Face + Cube 3D Google Colab demo
Cube 3D is our first step towards 3D intelligence, which involves a shape tokenizer and a text-to-shape generation model. We are unlocking the power of generating 3D assets and enhancing creativity for all artists. Our latest version of Cube 3D is now accessible to individuals, creators, researchers and businesses of all sizes so that they can experiment, innovate and scale their ideas responsibly. This release includes model weights and starting code for using our text-to-shape model to create 3D assets. diff --git a/cubepart/README.md b/cubepart/README.md index 5210130..d59eae7 100644 --- a/cubepart/README.md +++ b/cubepart/README.md @@ -1,10 +1,10 @@ # CubePart: An Open-Vocabulary Part-Controllable 3D Generator
- - - - + CubePart project page + CubePart arXiv paper + CubePart models on Hugging Face + CubePart demo on Hugging Face