Skip to content

Text-to-3D

3D Generation

Open Source

  • BE

    Block3D: Efficient Text-to-3D Generation via Block-Wise Diffusion Discovered by AI

    2026

    An efficient text-to-3D generation framework that utilizes block-wise diffusion to achieve high geometric fidelity at a significantly reduced inference cost compared to global diffusion models.

    diffusion efficiency mesh

    GitHub · arXiv

  • HB

    Hunyuan3D-Buffalo 1.0: A Unified Multimodal Model for Scalable 3D Generation, Understanding, and Editing Discovered by AI

    2026

    A unified 3D multimodal framework that combines autoregressive modeling with diffusion-based 3D generation. It supports 3D understanding, text-to-3D generation, instruction-guided 3D editing, and text-grounded part generation within a single architecture.

    3d-editing 3d-generation autoregressive multimodal

    Project · GitHub · arXiv

  • HW

    HY-World 2.0 Discovered by AI

    2026

    HY-World 2.0 is a multi-modal world model framework that generates and reconstructs navigable 3D worlds (meshes or Gaussian Splattings) from text, images, or videos.

    3d-generation scene-generation text-to-3d world-model

    GitHub · arXiv

  • 3R

    3DGen-R1 Discovered by AI

    2025

    A reinforcement learning framework for text-to-3D generation that optimizes global geometric consistency and fine-grained local textures through progressive reward designs.

    cvpr reinforcement-learning text-to-3d

    GitHub · arXiv

  • BR

    BrickGPT

    2025

    Generates physically stable, buildable brick (LEGO-style) structures from text, jointly reasoning about geometry and real-world structural stability (ICCV 2025 Best Paper).

    buildable physics text-to-3d

    GitHub

  • CF

    Consistent Flow Distillation for Text-to-3D Generation Discovered by AI

    2025

    Introduces Consistent Flow Distillation (CFD), a method that leverages the gradient of the diffusion ODE or SDE sampling process to guide 3D generation, addressing the over-smoothing and low-diversity issues of Score Distillation Sampling (SDS).

    diffusion score-distillation text-to-3d

    GitHub · arXiv · Paper

  • DS

    Direct3D-S2 Discovered by AI

    2025

    A scalable 3D generation framework based on sparse volumes and Spatial Sparse Attention (SSA) that dramatically reduces training costs while generating high-resolution 3D shapes.

    3d-generation dit sparse-attention

    arXiv · Paper

  • DR

    DreamDPO

    2025

    Aligns text-to-3D generation with human preferences via Direct Preference Optimization, steering optimization toward more preferred 3D outputs (ICML 2025).

    alignment dpo preferences

    GitHub

  • HO

    Hunyuan3D-Omni Discovered by AI

    2025

    A unified framework for fine-grained, controllable 3D asset generation built on Hunyuan3D 2.1. It accepts point clouds, voxels, bounding boxes, and text prompts to generate high-quality 3D assets with precise structural control.

    3d-assets controllable-generation text-to-3d

    GitHub · arXiv · Hugging Face · Paper

  • PR

    Progressive Rendering Distillation: Adapting Stable Diffusion for Instant Text-to-Mesh Generation without 3D Data Discovered by AI

    2025

    A novel training scheme, Progressive Rendering Distillation (PRD), that adapts pre-trained text-to-image diffusion models into a native 3D generator (TriplaneTurbo) without requiring 3D ground-truth data, enabling high-quality 3D mesh generation in 1.2 seconds.

    diffusion mesh-generation score-distillation text-to-3d

    Project · GitHub · arXiv

  • RC

    Roblox Cube Discovered by AI

    2025

    A 3D foundation model developed by Roblox that generates interactive 3D assets with built-in physical properties and behaviors from text prompts.

    3d-generation foundation-model interactive-3d

    GitHub · arXiv · Hugging Face

  • SE

    SegmentDreamer Discovered by AI

    2025

    SegmentDreamer is a text-to-3D synthesis framework that leverages segmented consistency trajectory distillation to optimize 3D asset generation under classifier-free guidance, improving fidelity and detail.

    3d-generation consistency-model distillation text-to-3d

    GitHub · arXiv

  • T2

    TRELLIS.2 Discovered by AI

    2025

    Introduces a native and compact structured latent representation (O-Voxel) for high-quality 3D generation, capturing complex topologies and detailed appearance.

    3d-generation structured-latents voxel

    arXiv · Paper

  • WT

    Walking the Schrödinger Bridge: A Direct Trajectory for Text-to-3D Generation Discovered by AI

    2025

    Introduces Trajectory-Centric Distillation (TraCe), a text-to-3D generation framework that models Score Distillation Sampling (SDS) as a Schrödinger Bridge problem to learn an optimal transport path between 3D renderings and dynamically estimated target distributions.

    gaussian-splatting schrodinger-bridge score-distillation text-to-3d

    GitHub · arXiv

  • WO

    WorldGen Discovered by AI

    2025

    An open-source 3D scene generation tool that creates complete, explorable 3D environments from text prompts or reference images in seconds. It supports 360° consistent exploration, indoor and outdoor scenes, and outputs both Gaussian splatting and mesh formats.

    3d-scene gaussian-splatting text-to-3d world-generation

    GitHub

  • BO

    Bootstrap3D Discovered by AI

    2024

    A framework that automatically generates high-quality multi-view synthetic images with detailed captions to improve the training and prompt-following capabilities of multi-view 3D diffusion models.

    diffusion multi-view synthetic-data

    GitHub · arXiv

  • CC

    Connecting Consistency Distillation to Score Distillation for Text-to-3D Generation Discovered by AI

    2024

    Analyzes score distillation methods by connecting consistency distillation theory to score distillation, proposing an optimization framework that improves the fidelity and detail of 3D assets generated from text prompts.

    consistency-distillation score-distillation text-to-3d

    GitHub · arXiv · Paper

  • D3

    DIRECT-3D Discovered by AI

    2024

    A diffusion-based 3D generative model trained directly on extensive noisy and unaligned "in-the-wild" 3D assets, enabling direct text-to-3D generation without clean 3D data.

    diffusion text-to-3d wild-data

    arXiv · Paper

  • DR

    DreamCraft3D++ Discovered by AI

    2024

    An advanced hierarchical 3D asset generation framework that extends DreamCraft3D by replacing the slow geometry optimization stage with a feed-forward multi-plane reconstruction model, accelerating generation speed by 1000x while maintaining high-fidelity geometry and texture details.

    3d-generation diffusion text-to-3d

    GitHub · arXiv · Paper

  • DR

    DreamReward Discovered by AI

    2024

    A comprehensive framework that learns and improves text-to-3D models from human preference feedback. It introduces Reward3D, a general-purpose 3D reward model, and utilizes Reinforcement Learning from Human Feedback (RLHF) to generate high-fidelity, multi-view consistent 3D content.

    human-preference rlhf text-to-3d

    GitHub · arXiv

  • PL

    PlacidDreamer Discovered by AI

    2024

    A unified text-to-3D generation framework that uses a single multi-view diffusion model for initialization, multi-view consistency, and text-conditioned generation, while addressing over-saturation with a Balanced Score Distillation algorithm.

    diffusion score-distillation text-to-3d

    GitHub · arXiv

  • TU

    Turbo3D: Ultra-fast Text-to-3D Generation Discovered by AI

    2024

    Presents Turbo3D, an ultra-fast text-to-3D generation system that produces high-quality 3D Gaussian Splatting assets in under one second using a rapid 4-step, 4-view diffusion generator and an efficient feed-forward Gaussian reconstructor.

    fast-generation gaussian-splatting text-to-3d

    arXiv · Paper · Website

  • GA

    GaussianDreamer Discovered by AI

    2023

    A fast text-to-3D generation framework that bridges 2D and 3D diffusion models using 3D Gaussian Splatting. It initializes 3D Gaussians from a 3D point cloud prior and refines them using 2D diffusion models, significantly accelerating the generation process.

    diffusion-models gaussian-splatting text-to-3d

    GitHub · arXiv

  • LI

    LucidDreamer (Interval Score Matching) Discovered by AI

    2023

    A text-to-3D generation framework that introduces Interval Score Matching (ISM) to overcome the over-smoothing and over-saturation issues of Score Distillation Sampling (SDS), producing highly detailed 3D assets.

    high-fidelity score-matching text-to-3d

    GitHub · arXiv

  • PR

    ProlificDreamer

    2023

    Introduces Variational Score Distillation, a particle-based variational framework that improves text-to-3D generation fidelity and diversity over score distillation sampling.

    diffusion nerf text-to-3d

    Project · GitHub · arXiv

  • DT

    DreamFusion: Text-to-3D using 2D Diffusion

    2022

    Introduces score distillation sampling to optimize a NeRF from a pretrained 2D diffusion model, generating 3D objects from text with no 3D training data.

    diffusion score-distillation

    Project · arXiv

  • MA

    Magic3D

    2022

    A coarse-to-fine text-to-3D approach that first produces a low-resolution neural field and then optimizes a textured mesh for higher-resolution content creation.

    diffusion mesh text-to-3d

    Project · arXiv

Proprietary / Industry Tools

  • SA

    Sloyd AI Discovered by AI

    2021

    An AI-powered 3D modeling platform designed to generate game-ready, textured, and customizable 3D assets from text prompts or parametric templates.

    3d-generation game-ready parametric text-to-3d

    Website

  • RH

    Rodin (Hyper3D)

    A commercial 3D generation service with an API producing detailed textured models from text or images for production asset workflows.

    api commercial

    Website

  • TR

    Tripo

    A commercial text/image-to-3D generator known for fast mesh creation, frequently used for rapid game asset prototyping.

    api commercial fast

    Website