OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

Lidarcrafter

LiDARCrafter is a Python research framework for generating and editing dynamic 4D LiDAR scenes. It converts language into an editable 4D layout and uses that layout to guide LiDAR synthesis across time.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Lidarcrafter — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use LiDARCrafter when a research project needs a code framework for language-conditioned 4D LiDAR generation or editing, with a documented nuScenes data-preparation and evaluation workflow.

Who it is for

  • Researchers studying LiDAR sequence generation and dynamic world models
  • Autonomous-driving researchers prototyping synthetic LiDAR scenes
  • Developers exploring language-conditioned 4D scene editing

Documented uses

  • Generate dynamic LiDAR sequences guided by editable 4D scene layouts
  • Explore language-guided edits to LiDAR scenes
  • Prepare nuScenes-derived files for the documented research workflow
  • Train classification and uncertainty models and extract foreground samples for evaluation
Before you choose

Requirements & limits

The repository is public and identified as open source. Its license information is mixed: the README states Apache 2.0 for the work and cautions that specific implementations may have other licenses.

  • The documented workflow is Python-based and requires environment setup from environment.yml.
  • Data preparation is centered on nuScenes and includes a script for generating metadata, scene graphs and CLIP features.
  • The repository lists training, inference and evaluation code, but its model-zoo section says it is still to be updated.
  • The README directs commercial users to review the license documentation for implementation-specific terms.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Prepare nuScenes data for LiDARCrafter

Create the dataset structure and derived files used by the documented workflow.

Before you start

  • A nuScenes dataset directory
  • An environment configured according to the repository's environment.yml

Steps

  1. Link the nuScenes data directory at ./data/nuscenes, as shown in the README.

  2. Run bash scripts/create_data.sh to generate tracking and state information, updated scene-graph pickle files and CLIP features.

Expected outcome

The repository's documented data structure includes nuScenes data, clips and generated information files for training and validation.

Guide limitations

  • This covers data preparation only, not a complete generation or inference run.
  • The README refers to DrivingDiffusion for dataset creation details.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
17 October 2026
Original website title
worldbench/LiDARCrafter · Official homepage title · retained in its source language
How we review sourcesSuggest a correction