Decomposing an input image into layers via "color unblending"
Clone or download
Latest commit 7f3b9c6 Jan 23, 2019
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
docs/images Add images for README.md Dec 19, 2018
examples Add example data Dec 17, 2018
external Add submodules Dec 17, 2018
resources Add main source files Dec 17, 2018
unblending-cli Add main source files Dec 17, 2018
unblending-gui Add comments Jan 22, 2019
unblending Add comments Jan 23, 2019
.gitmodules Add submodules Dec 17, 2018
CMakeLists.txt Hide a NLopt's build option Jan 5, 2019
Dockerfile Add Dockerfile Dec 17, 2018
Doxyfile Update Doxyfile Jan 16, 2019
LICENSE.Commercial Add README and LICENSEs Dec 17, 2018
LICENSE.LGPLv3 Add README and LICENSEs Dec 17, 2018
LayerInfos.md Separate README files Dec 25, 2018
README.md Update dependencies list Jan 10, 2019

README.md

Unblending

Unblending is a C++ library for decomposing a target image into a set of semi-transparent layers associated with advanced color-blend modes (e.g., "multiply" and "color-dodge"). Output layers can be imported to Adobe Photoshop, Adobe After Effects, GIMP, Krita, etc. and are useful for performing complex edits that are otherwise difficult.

An example of image decomposition. Input image courtesy of David Revoy.

  • unblending is a library that provides the main functionality for the use from other programs.
  • unblending-cli is an executable with a command line interface to use the main functionality.
  • unblending-gui is an executable with a graphical user interface to use the main functionality with interactive specification of necessary input parameters.

Publication

Yuki Koyama and Masataka Goto. 2018. Decomposing Images into Layers with Advanced Color Blending. Comput. Graph. Forum 37, 7, pp.397--407 (2018). (a.k.a. Proceedings of Pacific Graphics 2018) DOI: https://doi.org/10.1111/cgf.13577

This library is provided mainly for researchers who want to use or extend the method described in the above paper. We also expect that this library is useful for developers who plan to use the method in their software.

Demo

Unblending Web App: https://unblending.ongaaccel.jp

Dependencies

Prerequisites

Eigen and Qt5 need to be installed beforehand. macOS users can use brew for this purpose:

$ brew install eigen qt5

Build Using CMake

Unblending is organized using cmake and is built by the following commands:

$ git clone https://github.com/yuki-koyama/unblending.git --recursive
$ mkdir build
$ cd build
$ cmake ../unblending
$ make

This process builds a static library (e.g., libunblending.a) and a command line interface (CLI) named unblending-cli.

The CLI can be used by the following command:

$ ./unblending-cli/unblending-cli [-o <output-dir-path>] [-w <target-image-width>] <input-image-path> <layer-infos-path>

The GUI allows you to interactively specify necessary parameters. Currently the GUI is tested on macOS only (pull requests are highly appreciated).

GUI. Input image courtesy of David Revoy.

Build and Run Using Docker

If you use docker, you can easily build the CLI by docker build:

$ git clone https://github.com/yuki-koyama/unblending.git --recursive
$ cd unblending
$ docker build -t unblending-cli:latest .

Then, you can use the CLI by docker run:

$ docker run --rm --volume $(pwd):/tmp -it unblending-cli:latest [-o <output-dir-path>] [-w <target-image-width>] <input-image-path> <layer-infos-path>

For example, the following command generates results into a sub directory named output:

$ docker run --rm --volume $(pwd):/tmp -it unblending-cli:latest -o /tmp/output -w 120 /tmp/examples/magic.png /tmp/examples/magic.json

Note: this typically takes around 10--30 seconds with consumer-level laptops.

LayerInfos File Structure

See LayerInfos.md.

Licensing

Unblending is dual-licensed; You may use this software under either LGPLv3 or our commercial license. See the LICENSE files for details.

Contributing

Pull requests are highly welcome. Please be aware that any contribution to this repository will be licensed under the above license condition.

Authors

  • Yuki Koyama
  • Masataka Goto

Copyright

Copyright (c) 2018 National Institute of Advanced Industrial Science and Technology (AIST) - koyama.y@aist.go.jp