3D & Worlds

SC

Streaming Code World Model

Model

1

Research code that streams video generation chunk by chunk from a rough 3D proxy, built on Code World Model.

M4

Matrix-Game 4.0

Model · Skywork

2

Real-time playable world model from Skywork pairing a game engine and coding agent with a few-step causal video model.

FL

FlowHMR

Model

1

Monocular video motion capture that combines flow matching with physics-based RL for plausible 3D human motion.

AT

Any Time, Any Space (ATAS)

Research

1

Research framework for long, consistent video generation using a persistent 3D world state as memory.

IV

InfiniSplat v2

Model

1

Feed-forward 3D Gaussian Splatting model that reconstructs scenes from arbitrary input views, with or without camera poses.

RE

ReLift

Model · Nvidia

1

Feed-forward model that repairs lifted 2D tracks and depth into accurate dense 3D trajectories from monocular video.

U3

UniForge 3D

Model

1

Research model that builds complete 3D assets from sparse, unposed images by jointly doing reconstruction and generation.

GE

GenIA

Model · Meta

1

Meta research framework that aligns a frozen SAM 3D model to input images or video at inference time for 3D reconstruction.

LL

Lucid Lyra

Model · Nvidia

1

NVIDIA research model that turns a single image into a queryable 3D Gaussian scene with semantics and object instances.

MU

Multics

Model

1

Research multi-player video world model with hierarchical communication, trained on 4 players and run with up to 10.

HA

HairDeck

Model

1

Research model that generates animation-ready stylized hair cards and spring-bone chains from a single character image.

T3

T3lescope

Research · Preferred Networks

1

Generative method from Preferred Networks that reconstructs high-resolution 3D scene meshes from posed multi-view images.

MS

Mira Scene

Model

1

Model that reconstructs a single image into a 3D scene with separate, individually usable objects.

Meta Horizon tools that turn a text prompt into a 2D or 3D mobile game, via a phone app (Create) and a browser (Studio).

JA

Jing and Dao

Model · XGen Labs

1

Research preview from XGEN Labs: Dao, a persistent world-state engine, and Jing, a per-character experience model.

Geometry-guided video-to-video model from Viggle that re-renders existing footage along a new camera path and timing.

LW

Lingbot World 2

Model · Robbyant

2

Real-time interactive world model from Robbyant that generates controllable, long-horizon environments.

FI

Fire3D

Model

1

Feed-forward system that turns an RGB image or video into editable, textured per-object 3D meshes in under a minute.

Image-based model that predicts depth maps, surface normals and albedo from a single regular image.

AT

Atlas

Model · World Labs

3

World Labs' world model that generates and reconstructs explorable 3D worlds from images, text, video or 3D input.

LU

Lucida

Research · ByteDance

2

ByteDance research pipeline that turns indoor captures into editable, simulation-ready 3D scenes of separate object meshes.

HW

H3 World

Model

1

Research framework that turns the Minimax H3 video generator into an interactive world model controlled by key presses.

SO

Solaris

Model · Runway

1

Runway's real-time interactive video model, built on Gen-4.5, that conditions frames on user clicks and drags.

GW

GWM Worlds 2

Model · Runway

1

Runway's real-time interactive world model generating 720p/24fps video with audio, steered by text actions and camera input.