Skip to main content

A 3D scanning application

Project description

# A Ciclop 3D Scanner CLI

## Introduction

I got a Ciclop 3D scanner, figured it can’t work well on modern Linux distributions, started to hack…

This scanner is fully opensource, uses double lasers and is quite decent for the price. You can find some for [around $100]( on the web. This software only targets Linux users.

## Features

  • Makes any scanner work on Linux (currently Ciclop is supported, any can be added in few hours)

  • Colored two lasers scanning done in 2.5 minutes (in bright condition)

  • Fully automated calibration done in 30s !

  • Manual laser segmentation mode (to avoid calibration mistakes)

  • Two passes scanning mode for difficult objects

  • User friendly CLI

## Installation

On Archlinux system:

% yaourt -S –needed opencv python-numpy python-scipy % wget % unzip % cd thot-master % python build % sudo python install

> If you prefer using python2, just replace “python” with “python2” on the lines above

> You may also want to install guvcview to setup the webcam before starting the app

## Usage

[![Quickstart video](](

After installing all dependencies, unpack sources and open a terminal emulator:

% cd thot-master % ./

Calibrate software (after changing exposure to get the pattern recognized):

Scan Bot> cam_exposure 3000 Scan Bot> calibrate

Scan 3D object

Scan Bot> scan

## Advanced usages

You can also call the application from the command line:

$ thot recalibrate

Reconstruct previously scanned mesh (to test new calibration):

$ thot make

A special “exec” command allows command chaining from the command line:

$ thot exec calibrate, keep_laser 1 , make

Rebuild mesh from previous scan, keeping first laser information only:

$ thot exec keep_laser 1, make

Rebuild mesh with different line detection threshold

$ thot exec algop threshold 5, make

Images are saved into capture/ folder Pointcloud is saved as capture.ply

## Example

Result of a 2 lasers scan (unprocessed point cloud):

![Tux]( ![Tux side](

## Future / TODO

  • Make “first start” commands:
    • make separate shots for camera

    • calibrate camera from the shots

    • ask user to place the pattern on the surface

    • start calibration

  • Blender integration

  • compute normals or mesh directly

## Dependencies

  • opencv

  • numpy

  • pyserial

  • prompt_toolkit

  • scipy

# Troubleshooting

## “Mesh in a mesh” issue

If you have a mesh inside another mesh, this is probably a camera calibration issue: take many shots with the shot command (changing position of chessboard each time to cover maximum surface), then use calibrate_shots to compute camera calibration again.

Scan Bot> advanced Scan Bot> shot Scan Bot> shot Scan Bot> shot Scan Bot> shot Scan Bot> shot Scan Bot> calibrate_shots

After this you will need to do standard calibration again (platform and lasers), in case you didn’t change the setup and you calibrated the scanner recently, you can avoid capturing pictures by typing:

$ thot recalibrate

## Meshe is randomly distorted, but I can see two different shapes

This is probably a laser calibration issue, some lines are probably badly detected. You can fix it by discarding the frames that are not well analyzed:

$ thot recalibrate_manual

In case you want to capture the pictures again, just type:

$ thot calibrate_manual

## Bugs

  • no release yet: may not work out of the box, consider unstable

  • camera value getters are broken

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

thotus-0.1.1.tar.gz (30.8 kB view hashes)

Uploaded Source

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page