Research map / 3D vision / 3D generation
WonderTurbo: Generating Interactive 3D World in 0.72 Seconds
Chaojun Ni, Xiaofeng Wang, Zheng Zhu, Weijie Wang, Haoyun Li, Guosheng Zhao, Jie Li, Wenkang Qin, Guan Huang, Wenjun Mei
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add StepSplat instead of FreeSplat as geometry model (ablation) | WonderTurbo · Novel view rendering quality | CLIP score (CS): 27.65 -> 28.65 | Improved |
| add StepSplat instead of FreeSplat as geometry model (ablation) | WonderTurbo · Novel view rendering quality | Q-Align: 3.1543 -> 3.7253 | Improved |
| add StepSplat instead of DepthSplat as geometry model (ablation) | WonderTurbo · Novel view rendering quality | CLIP score (CS): 27.32 -> 28.65 | Improved |
| add StepSplat instead of DepthSplat as geometry model (ablation) | WonderTurbo · Novel view rendering quality | CLIP aesthetic (CA): 6.7432 -> 7.3243 | Improved |
| add depth guided cost volume (ablation) | StepSplat · Novel view rendering quality | CLIP score (CS): 27.72 -> 28.65 | Improved |
| add depth guided cost volume (ablation) | StepSplat · Novel view rendering quality | Q-Align: 3.4361 -> 3.7253 | Improved |
| add incremental fusion (ablation) | StepSplat · Novel view rendering quality | CLIP score (CS): 27.87 -> 28.65 | Improved |
| add incremental fusion (ablation) | StepSplat · Novel view rendering quality | CLIP consistency (CC): 0.9654 -> 0.9732 | Improved |
| add FastPaint 2-step inpainting (ablation) | WonderTurbo · Novel view rendering quality | CLIP score (CS): 27.82 -> 28.65 | Improved |
| add FastPaint 2-step inpainting (ablation) | WonderTurbo · Novel view rendering quality | Q-Align: 3.7146 -> 3.7253 | No clear effect |
| use WonderTurbo instead of WonderWorld | WonderTurbo · Scene generation time (H20 GPU) | total time (s): 11.05 -> 0.72 | Improved |
| use WonderTurbo instead of WonderWorld | WonderTurbo · Novel view rendering quality | Q-Align: 3.7823 -> 3.7253 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- LucidDreamer — LucidDreamer: Towards High-Fidelity Text-to-3D Generation via Interval Score Matching (CVPR 2024) · better on 6 of 6 shared results
- DreamScene360 — DreamScene360: Unconstrained Text-to-3D Scene Generation with Panoramic Gaussian Splatting (ECCV 2024) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- WonderWorld: Interactive 3D Scene Generation from a Single Image (CVPR 2025 · similarity 0.93)
- Interactive3D: Create What You Want by Interactive 3D Generation (CVPR 2024 · similarity 0.88)
- WonderJourney: Going from Anywhere to Everywhere (CVPR 2024 · similarity 0.86)
- Architect: Generating Vivid and Interactive 3D Scenes with Hierarchical 2D Inpainting (NeurIPS 2024 · similarity 0.85)
- InstaInpaint: Instant 3D-Scene Inpainting with Masked Large Reconstruction Model (NeurIPS 2025 · similarity 0.85)
- WorldAct: Activating Monolithic 3D Worlds into Interactive-Ready Object-Centric Scenes (NeurIPS 2026 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
