Introducing Project Eden: A World Model for Multiplayer and Agents

VAST AI Research introduces Project Eden, a persistent multiplayer world model. Worlds remember your actions, stay consistent across views, and run continuously.

Research
DMiT: Deformable Mipmapped Tri-Plane Representation for Dynamic Scenes

Render dynamic scenes efficiently with DMiT, using deformable mipmapped tri-planes for high-fidelity, multi-resolution novel view generation.

Tripo Team
📅 · 2025/11/27
Research
CharacterGen: Efficient 3D Character Generation from Single Images

See how CharacterGen turns a single image into a high-quality, pose-calibrated 3D character ready for rigging and animation.

Tripo Team
📅 · 2025/11/27
Research
TriplaneGaussian: A new hybrid representation for single-view 3D generation

See how TGS reconstructs high-quality 3D models from a single image in seconds using hybrid Triplane-Gaussian representations and transformers.

Tripo Team
📅 · 2025/11/27
Research
EpiDiff: Enhancing Multi-View Synthesis via Localized Epipolar-Constrained Diffusion

Discover how EpiDiff generates 16 multiview-consistent high-quality images in just 12 seconds using localized epipolar-constrained diffusion.

Tripo Team
📅 · 2025/11/27
Research
Wonder3D: Single Image to 3D using Cross-Domain Diffusion

Explore how Wonder3D converts a single image into a high-fidelity 3D textured mesh in just 2–3 minutes using cross-domain diffusion and multi-view normal maps.

Tripo Team
📅 · 2025/11/27
Research
DreamComposer: Controllable 3D Object Generation via Multi-View Conditions

Discover how DreamComposer generates controllable 3D objects by injecting multi-view conditions into pre-trained diffusion models, enabling high-fidelity novel view synthesis and 3D reconstruction from multiple images.

Tripo Team
📅 · 2025/11/27
Research
SC-GS: Sparse-Controlled Gaussian Splatting for Editable Dynamic Scenes

Explore SC-GS and see how sparse-controlled Gaussian splatting enables editable dynamic 3D scenes, allowing high-fidelity motion synthesis and interactive user-controlled motion editing.

Tripo Team
📅 · 2025/11/27
Research
PI3D: Efficient Text-to-3D Generation with Pseudo-Image Diffusion

Discover PI3D and see how pseudo-image diffusion turns text prompts into high-quality 3D shapes in minutes, leveraging 2D diffusion models for fast and consistent text-to-3D generation.

Tripo Team
📅 · 2025/11/27
Research
CSD: Text-to-3D with Classifier Score Distillation

Explore Classifier Score Distillation (CSD) and see how text-to-3D generation leverages classifier-free guidance for fast, high-quality shape generation, texture synthesis, and mesh editing.

Tripo Team
📅 · 2025/11/27
Research
MV-Adapter: Multi-view Consistent Image Generation Made Easy

Explore MV-Adapter and discover how to generate high-fidelity multi-view images efficiently using any pre-trained text-to-image model, while preserving 3D consistency and versatility under various input conditions.

Tripo Team
📅 · 2025/11/27
Research
SAMPart3D: Segment Any Part in 3D Objects

Explore SAMPart3D and discover how to segment any 3D object into semantic parts at multiple granularities, enabling flexible part editing and scalable 3D understanding without predefined labels.

Tripo Team
📅 · 2025/11/27
Research
DreamCraft3D++: Efficient Hierarchical 3D Generation with Multi-Plane Reconstruction Model

Explore DreamCraft3D++ to see how multi-plane reconstruction and feed-forward 3D generation produce high-quality, textured 3D assets in just 10 minutes, with enhanced geometry and realistic 360° textures.

Tripo Team
📅 · 2025/11/27