https://github.com/esimov/colidr
Coherent Line Drawing implementation in Go.
https://github.com/esimov/colidr
coherent-line-drawing computer-graphics computer-vision gocv golang non-photorealistic-rendering opencv rendering
Last synced: about 1 month ago
JSON representation
Coherent Line Drawing implementation in Go.
- Host: GitHub
- URL: https://github.com/esimov/colidr
- Owner: esimov
- License: mit
- Created: 2018-08-10T07:38:32.000Z (almost 7 years ago)
- Default Branch: master
- Last Pushed: 2021-04-13T09:13:26.000Z (about 4 years ago)
- Last Synced: 2025-03-27T10:02:54.217Z (about 2 months ago)
- Topics: coherent-line-drawing, computer-graphics, computer-vision, gocv, golang, non-photorealistic-rendering, opencv, rendering
- Language: Go
- Homepage:
- Size: 188 KB
- Stars: 55
- Watchers: 4
- Forks: 6
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- Funding: .github/FUNDING.yml
- License: LICENSE
Awesome Lists containing this project
README
# colidr - Coherent Line Drawing
[](https://github.com/esimov/colidr/actions)
[](./LICENSE)
[](https://github.com/esimov/colidr/releases/tag/v1.0.1)
Implementation in Go of the [Coherent Line Drawing](http://umsl.edu/mathcs/about/People/Faculty/HenryKang/coon.pdf) algorithm developed by Kang et al, NPAR 2007.
## Requirements
- Go 1.10 or higher, but it should work even with a lower version
- OpenCV 3 (tested with 3.4.2)
- [gocv](https://github.com/hybridgroup/gocv) (bundled into the project, since it was extended with missing OpenCV functions needed for the implementation)
- [potrace](http://potrace.sourceforge.net/) - for transforming the bitmap into smooth, scalable image (however this is optional)
## Installation
```bash
$ go get -u github.com/esimov/colidr/
$ go install
```
Another option is cloning the repository and running the `make` file.
```bash
$ git clone https://github.com/esimov/colidr
$ cd colidr
$ make
```
This will generate the binary file.
## Usage
```bash
$ colidr -h
┌─┐┌─┐┬ ┬┌┬┐┬─┐
│ │ ││ │ ││├┬┘
└─┘└─┘┴─┘┴─┴┘┴└─
Coherent Line Drawing CLI
Version: 1.0.1
-aa
Anti aliasing
-bl int
Blur size (default 3)
-di int
Number of FDoG iteration
-ei int
Number of Etf iteration (default 1)
-in string
Source image
-k int
Etf kernel (default 3)
-out string
Destination image
-pt
Use potrace to smooth edges (default true)
-rho float
Rho (default 0.98)
-sc float
SigmaC (default 1)
-sm float
SigmaM (default 3)
-sr float
SigmaR (default 2.6)
-tau float
Tau (default 0.98)
-ve
Visualize Etf
-vr
Visualize end result
```
Feel free to play with the values in order to modify the visual output of the generated (non-photorealistically rendered) image. To obtain higher fidelity results you need to increase the `kernel` value and also the ETF iteration number. Different combinations produces completely different output. The `-di`, `-ei`, `-k` flags are mostly used for fine tuning, on the other hand `-rho` and `-tau` flags could change dramatically the rendered output.
You can also visualize the edge tangent flow if you enable the `-ve` flag. Below is the process illustrated:
| Original image | Edge tangent flow | Coherent line drawing (final output)
|:--:|:--:|:--:|
|  |  |  |
Using the `-pt` flag you can trace the generated bitmap into a smooth scalabe image. You need to have [potrace](http://potrace.sourceforge.net/) installed on your machine for this scope.
Below is an example whith and without the potrace flag activated.
| Normal output | Potrace activated
|:--:|:--:|
|  |  |
The above [image](http://hof.povray.org/images/patio.jpg) was ganareted with the following command:
```bash
colidr -in ~/Desktop/patio.jpg -out ~/Desktop/patio_scene.png -k=1 -sr=2.5 -sm=3.2 -tau=0.9975 -di=1 -aa=1 -ve=1 -vr=0 -pt=1 -ei=1
```
## Sample images
| Rasterized bitmap | Vectorized image
|:--:|:--:|
|  |  |
|  |  |
|  |  |
|  |  |
## Todo
- [ ] GUI support
## Author
* Endre Simo ([@simo_endre](https://twitter.com/simo_endre))
## License
Copyright © 2019 Endre Simo
This project is under the MIT License. See the [LICENSE](https://github.com/esimov/colidr/blob/master/LICENSE) file for the full license text.