Loading video...

Video Failed to Load

Go Home

WorldCam Interactive Autoregressive 3D Gaming Worlds with Camera Pose as a Unifying Geometric Representation paper:

_akhaliq's profile picture

AK

526,753 subscribers

15,267 views • 6 months ago •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

5 Comments

Bohan Zeng's profile picture
Bohan Zeng6 months ago

Excellent work! We will add this method in our framework ( ) later

NirD's profile picture
NirD6 months ago

fwiw their camera pose predictor needs 512x512 minimum resolution, breaks hard on anything smaller even with padding

Latent Node's profile picture
Latent Node6 months ago

Looks great, we also did something recently using mobile gameplay videos -

Victor's profile picture
Victor6 months ago

camera pose as the unifying representation is such a smart architectural choice — instead of fighting the geometry, you make it the organizing principle. curious how this handles discontinuous environments vs more open-world settings

Chain Alpha's profile picture
Chain Alpha6 months ago

another metaverse attempt? We'll see about that 🍿.

Related Videos

CS:GO + Adobe + Wan2.1 = WorldCam. Interactive autoregressive 3D gaming worlds. > AI now generates playable 3D worlds live > You move the mouse, the AI builds the map instantly > Turn around, and it remembers exactly what was there No traditional game engines. Just neural networks hallucinating reality.
0:40

CS:GO + Adobe + Wan2.1 = WorldCam. Interactive autoregressive 3D gaming worlds. > AI now generates playable 3D worlds live > You move the mouse, the AI builds the map instantly > Turn around, and it remembers exactly what was there No traditional game engines. Just neural networks hallucinating reality.

Wildminder

67,899 views • 6 months ago

COCOON x LOST IN CULT🪲 Warp between worlds with Cocoon: Design Works in collaboration with Geometric Interactive, Annapurna Interactive & writer Rachel Watts🤝 Pre-order the Deluxe Edition before 16th December @ 6pm BST 🚨
0:40

COCOON x LOST IN CULT🪲 Warp between worlds with Cocoon: Design Works in collaboration with Geometric Interactive, Annapurna Interactive & writer Rachel Watts🤝 Pre-order the Deluxe Edition before 16th December @ 6pm BST 🚨

Lost In Cult

13,544 views • 11 months ago

Built a web app with AI using Three.js + WebGL that turns motion, mechanics and symmetry into an interactive 3D experience featuring a gear mechanism, gyroscope, turbine, coil and geometric structures. Workflow ↓
0:24

Built a web app with AI using Three.js + WebGL that turns motion, mechanics and symmetry into an interactive 3D experience featuring a gear mechanism, gyroscope, turbine, coil and geometric structures. Workflow ↓

Techartist

24,529 views • 2 months ago

Google DeepMind just dropped Genie 2. AI can now create diverse, interactive 3D worlds from a single image or text. Gaming will never be the same. 10 wild examples: 1. Long video generation on the fly
1:01

Google DeepMind just dropped Genie 2. AI can now create diverse, interactive 3D worlds from a single image or text. Gaming will never be the same. 10 wild examples: 1. Long video generation on the fly

Min Choi

373,295 views • 1 year ago

HY-World 2.0 A Multi-Modal World Model for Reconstructing, Generating, and Simulating 3D Worlds paper:
1:04

HY-World 2.0 A Multi-Modal World Model for Reconstructing, Generating, and Simulating 3D Worlds paper:

AK

13,879 views • 5 months ago

Step into a new era of gaming with Seele AI! 🚀 Create immersive 3D worlds from a single sentence. #SeeleAI #AIGaming #GameChangers SEELE AI
0:34

Step into a new era of gaming with Seele AI! 🚀 Create immersive 3D worlds from a single sentence. #SeeleAI #AIGaming #GameChangers SEELE AI

Daffydre🎶

76,462 views • 1 year ago

Fable 5 is an absolute beast at threejs. It's the best LLM at generating 3D simulations/worlds. Watch how I combine it with gpt-realtime-2 to generate interactive educational 3D worlds DAIR.AI. Comment below if you want to see a breakdown.
4:14

Fable 5 is an absolute beast at threejs. It's the best LLM at generating 3D simulations/worlds. Watch how I combine it with gpt-realtime-2 to generate interactive educational 3D worlds DAIR.AI. Comment below if you want to see a breakdown.

elvis

38,616 views • 3 months ago

🔥1-min Interactive Video Generation with Multimodal Control🔥 Towards *long-context world model*, #LongVie is an end-to-end autoregressive framework for controllable ultra-long video generation - Page: - Paper: . Thanks AK
2:12

🔥1-min Interactive Video Generation with Multimodal Control🔥 Towards *long-context world model*, #LongVie is an end-to-end autoregressive framework for controllable ultra-long video generation - Page: - Paper: . Thanks AK

Ziwei Liu

14,213 views • 1 year ago

Generative 3D worlds are amazing! Here is a generative AI video showing fine grained camera control. Camera movements are all hand places within a generated 3D world.
0:52

Generative 3D worlds are amazing! Here is a generative AI video showing fine grained camera control. Camera movements are all hand places within a generated 3D world.

Brittani Natali

21,061 views • 1 year ago

omg the pose of lisa’s 3D arms is a representation of a thai dance, she’s so proud of her roots 🇹🇭
0:24

omg the pose of lisa’s 3D arms is a representation of a thai dance, she’s so proud of her roots 🇹🇭

lisa loops

38,390 views • 5 months ago

Mind blown by this new CVPR 2026 paper: tttLRM🤯 It introduces a Test-Time Training layer for long-context autoregressive 3D reconstruction with linear complexity. Multiple image observations are compressed into fast weights that keep updating during inference, forming an implicit scene representation that can be decoded into explicit formats such as 3D Gaussian Splats. Progressive 3D reconstruction from streaming inputs is getting wild🥳It’s interesting to see how AI inference and real-world capture are starting to converge.
0:20

Mind blown by this new CVPR 2026 paper: tttLRM🤯 It introduces a Test-Time Training layer for long-context autoregressive 3D reconstruction with linear complexity. Multiple image observations are compressed into fast weights that keep updating during inference, forming an implicit scene representation that can be decoded into explicit formats such as 3D Gaussian Splats. Progressive 3D reconstruction from streaming inputs is getting wild🥳It’s interesting to see how AI inference and real-world capture are starting to converge.

KIRI Engine - 3D Scanner App

16,949 views • 6 months ago

We’re open-sourcing HY-World 2.0, a multimodal world model that generates, reconstructs, and simulates interactive *3D worlds* from text, images, and videos. Outputs can be integrated into game engines and embodied simulation pipelines. Key highlights: 🔹 One-click world generation Turn text or image into interactive 3D worlds automatically. 🔹 Pipeline-ready 3D outputs Editable 3D worlds for Unity and Unreal Engine, with standard 3D exports including mesh, 3DGS, and point clouds. 🔹 Unified world model system One model family for world generation and reconstruction across synthetic and real-world scenes. 🔹 Interactive character mode Explore generated 3D worlds in real time with physics-aware movement and collision support. ✨ Apply for access: 🔗 GitHub: 🤗 Hugging Face: 📄 Technical Report:
1:04

We’re open-sourcing HY-World 2.0, a multimodal world model that generates, reconstructs, and simulates interactive *3D worlds* from text, images, and videos. Outputs can be integrated into game engines and embodied simulation pipelines. Key highlights: 🔹 One-click world generation Turn text or image into interactive 3D worlds automatically. 🔹 Pipeline-ready 3D outputs Editable 3D worlds for Unity and Unreal Engine, with standard 3D exports including mesh, 3DGS, and point clouds. 🔹 Unified world model system One model family for world generation and reconstruction across synthetic and real-world scenes. 🔹 Interactive character mode Explore generated 3D worlds in real time with physics-aware movement and collision support. ✨ Apply for access: 🔗 GitHub: 🤗 Hugging Face: 📄 Technical Report:

Tencent Hy

372,797 views • 5 months ago

3D scene reconstruction works great until the camera never sees part of the scene. ArtiFixer from NVIDIA Research is an open autoregressive model that fills in the missing geometry that other methods leave blank. #SIGGRAPH2026 paper, code + demo:
0:29

3D scene reconstruction works great until the camera never sees part of the scene. ArtiFixer from NVIDIA Research is an open autoregressive model that fills in the missing geometry that other methods leave blank. #SIGGRAPH2026 paper, code + demo:

NVIDIA AI

155,754 views • 3 months ago

Check-out 📢DUSt3R📢 - a new 3D reconstruction model that works with no prior info on camera calibration nor viewpoint poses! Outperforms SoA monocular & multiview depth estimation & relative pose estimation. Paper, demo, videos (& soon code!) available
2:07

Check-out 📢DUSt3R📢 - a new 3D reconstruction model that works with no prior info on camera calibration nor viewpoint poses! Outperforms SoA monocular & multiview depth estimation & relative pose estimation. Paper, demo, videos (& soon code!) available

NAVER LABS Europe

17,160 views • 2 years ago

If you’ve been curious about the intersection of AI, 3D and gaming, then this 18 min video is for you: 0:00 Status Quo 1:54 Next Gen 2:55 Dynamic 3D Scenes 6:01 Large Scale 3D 11:10 Interactive 3D Worlds 14:07 Films 🤝 Games 16:35 Jensen's Prediction
18:00

If you’ve been curious about the intersection of AI, 3D and gaming, then this 18 min video is for you: 0:00 Status Quo 1:54 Next Gen 2:55 Dynamic 3D Scenes 6:01 Large Scale 3D 11:10 Interactive 3D Worlds 14:07 Films 🤝 Games 16:35 Jensen's Prediction

Bilawal Sidhu

61,498 views • 1 year ago

𝗜𝗻𝘁𝗿𝗼𝗱𝘂𝗰𝗶𝗻𝗴 𝗨𝗩𝗚𝗦 We introduce 𝗨𝗩𝗚𝗦, a new 2D representation of 3D Gaussian Splatting (3DGS) that leverages spherical mapping. Website: Paper:
0:23

𝗜𝗻𝘁𝗿𝗼𝗱𝘂𝗰𝗶𝗻𝗴 𝗨𝗩𝗚𝗦 We introduce 𝗨𝗩𝗚𝗦, a new 2D representation of 3D Gaussian Splatting (3DGS) that leverages spherical mapping. Website: Paper:

Nikolaos Sarafianos

10,752 views • 1 year ago

✨ Presenting MicKey (#CVPR2024, Oral) ✨ We regress and match 3D camera coordinates rather then 2D key points, all in metric space. Gives you a scaled relative pose between two RGB images. Paper: Page:
0:36

✨ Presenting MicKey (#CVPR2024, Oral) ✨ We regress and match 3D camera coordinates rather then 2D key points, all in metric space. Gives you a scaled relative pose between two RGB images. Paper: Page:

Eric Brachmann

70,428 views • 2 years ago

Interactive 3D posing and smart lighting prompts. With this nodepack in ComfyUI, it's now easier for you to control the image's pose and lighting:
1:09

Interactive 3D posing and smart lighting prompts. With this nodepack in ComfyUI, it's now easier for you to control the image's pose and lighting:

Daxiong (Lin)

13,330 views • 8 months ago

A totally new level of pose control in ComfyUI - VNCCS. - full 3D Pose Studio for character posing & lighting, - multi-pose, body generator, pose gallery; - vision-guided QWEN Detailer + camera controls.
1:04

A totally new level of pose control in ComfyUI - VNCCS. - full 3D Pose Studio for character posing & lighting, - multi-pose, body generator, pose gallery; - vision-guided QWEN Detailer + camera controls.

Wildminder

165,729 views • 8 months ago

Meet LA-Pose. Our latest model taking Wayve another step towards generalization at scale. LA-Pose employs large-scale self-supervised learning, building strong motion representations for 3D perception from 10.2 million unlabeled driving video snippets, unlike today's strongest approaches that often depend on expensive, carefully curated 3D supervision. With only a lightweight pose head and limited labelled data, LA-Pose achieves: 📷 State-of-the-art camera pose estimation 🌎 Strong zero-shot generalization across diverse driving scenarios 🏷️ Orders of magnitude less labelled data than fully supervised 3D approaches Our full blog post: Explore the full paper here:
1:00

Meet LA-Pose. Our latest model taking Wayve another step towards generalization at scale. LA-Pose employs large-scale self-supervised learning, building strong motion representations for 3D perception from 10.2 million unlabeled driving video snippets, unlike today's strongest approaches that often depend on expensive, carefully curated 3D supervision. With only a lightweight pose head and limited labelled data, LA-Pose achieves: 📷 State-of-the-art camera pose estimation 🌎 Strong zero-shot generalization across diverse driving scenarios 🏷️ Orders of magnitude less labelled data than fully supervised 3D approaches Our full blog post: Explore the full paper here:

Wayve

37,042 views • 5 months ago