Pre-Alpha

Rapid Synthetic
Dataset Creation

Create high-quality synthetic datasets with a few lines of code. Instantly see results and generate production-ready labeled datasets.

$ go install github.com/Rapid-Vision/rv@latest

See It In Action

script.py

1import rv
2
3class BasicScene(rv.Scene):
4    def generate(self):
5        self.get_world().set_params(sun_intensity=0.03)
6        cube = (
7            self.create_cube()
8            .set_location((1, 0, 0.5))
9            .set_scale(0.5)
10            .set_tags("cube")
11        )
12        sphere = (
13            self.create_sphere()
14            .set_location((-1, 0, 1))
15            .set_shading("smooth")
16            .set_tags("sphere")
17        )
18        plane = self.create_plane(size=1000)
19        empty = self.create_empty().set_location((0, 0, 1))
20
21        self.get_camera().set_location((7, 7, 3)).point_at(empty)

Single tool to preview and run

RV simplifies synthetic dataset creation with just two core commands that handle everything from preview to production.

Interactive Preview

rv preview script.py

Open a live Blender window that automatically updates as you modify your script. See your scene changes in real-time with immediate visual feedback.

Instant feedback on code changes
Interactive 3D visualization

Production Rendering

rv render script.py -n 1000 -p 3

Generate thousands of randomly varied, fully labeled images ready for machine learning. Complete with bounding boxes, classes, and masks.

Batch dataset generation
Automatic labeling (boxes, classes, masks)

Everything You Need for Synthetic Data

Built on Blender's powerful 3D engine with a simple command interface designed for data scientists and ML engineers.

Simple API

Create meshes, set up cameras, and distribute objects with intuitive Python commands.

Visual Feedback

See your changes instantly in an interactive 3D window that updates as you code.

Smart Labeling

Automatically generate bounding boxes, class labels, and segmentation masks.

Raytracing incuded

Leverage the full power of Blender's rendering engine and 3D capabilities.

Open Source

Free, open-source tool with an active community and extensible architecture.

Production Ready

Generate thousands of high-quality images suitable for training ML models.

Our Sponsors

Ready to Create Your First Dataset?

Try Rapid Vision out to generate high-quality randomized labeled datasets for your next computer vision project.