cyang1 / depixelize Goto Github PK
View Code? Open in Web Editor NEWA C++ implementation of the SIGGRAPH 2011 paper Depixelizing Pixel Art, by Johannes Kopf and Dani Lischinski
A C++ implementation of the SIGGRAPH 2011 paper Depixelizing Pixel Art, by Johannes Kopf and Dani Lischinski
README Author: Cary Yang <[email protected]> --------------------------------------------------------------------------- Overview --------------------------------------------------------------------------- This project is a C++ implementation of Depixelizing Pixel Art, by Kopf and Lischinski, SIGGRAPH 2011, written for my final project in CMU's 15-463, Computational Photography. This code has only been tested on the files in the images/ directory, it is likely to contain bugs. Also, the optimization phase does not appear to work correctly (output does not match paper's results). What follows is instructions on building the code, running the program, and a description of the source files. This README and the CMake build system for this project were largely adapted from projects in CMU's 15-462, Computer Graphics. --------------------------------------------------------------------------- Prerequisites --------------------------------------------------------------------------- There are some libraries required for this to compile, namely OpenCV (for image reading/writing and colorspace conversion) and Boost (for the Voronoi diagram and filesystem utilities). --------------------------------------------------------------------------- Building the Code --------------------------------------------------------------------------- This project uses the CMake build system (www.cmake.org). CMake is a kind of "meta build-system," in that it creates a build system for you. It supports many different build systems, ranging from Unix Makefiles to Visual Studio projects to XCode projects. 1) If you don't have CMake installed, you should install it. If you're running Linux, it should be available through your disto's package manager (apt-get, pacman, ports, etc). Windows and OSX installers can be downloaded from http://www.cmake.org/cmake/resources/software.html 2) Open up a command prompt in the "build" directory and run "cmake ../src". CMake will generate the build system in the build directory. By default, CMake will generate Makefiles on Linux and OSX and a Visual Studio project on Windows. If you'd like to change that, you can pass a different "generator" to cmake with the -G flag. A full list of generators can be found at the bottom of the output generated by running "cmake". For example, "cmake -G 'Xcode' ../src" will generate an Xcode project on OSX. 3) You can now use the build system in the "build" directory. The default target will compile everything for you, and the "install" target will copy the executable to the main "final_project" directory. NOTE: This project has only been tested on OS X Yosemite, compilation on other targets is not gauranteed. WARNING: This project will only compile with GNU-based compilers and will most likely not run on Windows as the proper runtime libraries are not included. --------------------------------------------------------------------------- Running the Program --------------------------------------------------------------------------- ./depixelize [options] input_filename Program Options: -o --output <FILENAME> Puts the resulting image into <FILENAME> (default: [input_filename]_out) -s --scale <INT> Scale the output image by this factor (default: 16) -h --help This message input_filename The input image to depixelize --------------------------------------------------------------------------- Source Files and Directory Structure --------------------------------------------------------------------------- A description of the top-level directories and all source/header file pairs follows. README -- this file images/ -- a set of input images from the original paper (C) Nintendo Co., Ltd src/build/* == Build system stuff. CMakeLists.txt -- Compiler flags can be adjusted by editing the list in here. src/depixelize/ == Project-specific files. main -- Main function, and application backend. pixel_grid -- An implementation of an 8-connected pixel grid that can make itself planar and resolve ambiguities. spline_optimizer -- Extracts and optimizes splines from the Voronoi diagram. color_util -- Various color utility functions. math_util -- Various math utility functions. voronoi_visual_utils -- Voronoi utilities, borrowed from the voronoi_visualizer example in Boost. src/geometry/ == Geometry related classes and functions bspline -- Contains a B-spline implementation that can find its own curvature and be integrated. edge -- Simple undirected-edge made from 2 points. point -- Simple point implementation, with some math utility operators implemented. shape -- A shape, defined by a B-spline edge and a list of internal color points. types -- Various common types. src/render/ == Rendering related classes renderer -- Base interface for all renderers. Defines rendering lists of shapes to arbitrary formats. svg_renderer -- SVG implementation of the renderer. simple_svg_1.0.0 -- An SVG rendering library from the Internets.
Hi, thanks for this repo. Unfortunately, I'm getting segfault or never ending loop on every input...
For input , program running on 100% CPU and never ends. Here is a Valgrind output
==23787== Memcheck, a memory error detector
==23787== Copyright (C) 2002-2013, and GNU GPL'd, by Julian Seward et al.
==23787== Using Valgrind-3.10.0.SVN and LibVEX; rerun with -h for copyright info
==23787== Command: ./depixelize -o out.svg images/smw_mario_input.png
==23787==
==23787== Conditional jump or move depends on uninitialised value(s)
==23787== at 0x6991A28: ippicvGetCpuFeatures (in /usr/local/lib/libopencv_core.so.3.0.0)
==23787== by 0x68242B2: ippicvStaticInit (in /usr/local/lib/libopencv_core.so.3.0.0)
==23787== by 0x53039F8: _GLOBAL__sub_I_deriv.cpp (in /usr/local/lib/libopencv_imgproc.so.3.0.0)
==23787== by 0x4010139: call_init.part.0 (dl-init.c:78)
==23787== by 0x4010222: _dl_init (dl-init.c:36)
==23787== by 0x4001309: ??? (in /lib/x86_64-linux-gnu/ld-2.19.so)
==23787== by 0x3: ???
==23787== by 0xFFF00025E: ???
==23787== by 0xFFF00026B: ???
==23787== by 0xFFF00026E: ???
==23787== by 0xFFF000276: ???
==23787==
==23787== Conditional jump or move depends on uninitialised value(s)
==23787== at 0x6991A28: ippicvGetCpuFeatures (in /usr/local/lib/libopencv_core.so.3.0.0)
==23787== by 0x68242B2: ippicvStaticInit (in /usr/local/lib/libopencv_core.so.3.0.0)
==23787== by 0x5303AA8: _GLOBAL__sub_I_sumpixels.cpp (in /usr/local/lib/libopencv_imgproc.so.3.0.0)
==23787== by 0x4010139: call_init.part.0 (dl-init.c:78)
==23787== by 0x4010222: _dl_init (dl-init.c:36)
==23787== by 0x4001309: ??? (in /lib/x86_64-linux-gnu/ld-2.19.so)
==23787== by 0x3: ???
==23787== by 0xFFF00025E: ???
==23787== by 0xFFF00026B: ???
==23787== by 0xFFF00026E: ???
==23787== by 0xFFF000276: ???
==23787==
==23787== Conditional jump or move depends on uninitialised value(s)
==23787== at 0x6991A28: ippicvGetCpuFeatures (in /usr/local/lib/libopencv_core.so.3.0.0)
==23787== by 0x68242B2: ippicvStaticInit (in /usr/local/lib/libopencv_core.so.3.0.0)
==23787== by 0x5303AF8: _GLOBAL__sub_I_imgwarp.cpp (in /usr/local/lib/libopencv_imgproc.so.3.0.0)
==23787== by 0x4010139: call_init.part.0 (dl-init.c:78)
==23787== by 0x4010222: _dl_init (dl-init.c:36)
==23787== by 0x4001309: ??? (in /lib/x86_64-linux-gnu/ld-2.19.so)
==23787== by 0x3: ???
==23787== by 0xFFF00025E: ???
==23787== by 0xFFF00026B: ???
==23787== by 0xFFF00026E: ???
==23787== by 0xFFF000276: ???
==23787==
==23787== Invalid read of size 4
==23787== at 0x97779A9: ??? (in /usr/lib/x86_64-linux-gnu/libOpenCL.so.1.0.0)
==23787== by 0x9777F28: ??? (in /usr/lib/x86_64-linux-gnu/libOpenCL.so.1.0.0)
==23787== by 0x4010139: call_init.part.0 (dl-init.c:78)
==23787== by 0x4010222: _dl_init (dl-init.c:36)
==23787== by 0x4014C6F: dl_open_worker (dl-open.c:577)
==23787== by 0x400FFF3: _dl_catch_error (dl-error.c:187)
==23787== by 0x40143BA: _dl_open (dl-open.c:661)
==23787== by 0x8B3902A: dlopen_doit (dlopen.c:66)
==23787== by 0x400FFF3: _dl_catch_error (dl-error.c:187)
==23787== by 0x8B3962C: _dlerror_run (dlerror.c:163)
==23787== by 0x8B390C0: dlopen@@GLIBC_2.2.5 (dlopen.c:87)
==23787== by 0x66C9A17: (anonymous namespace)::opencl_fn3<58, int, unsigned int, _cl_platform_id**, unsigned int*>::switch_fn(unsigned int, _cl_platform_id**, unsigned int*) (in /usr/local/lib/libopencv_core.so.3.0.0)
==23787== Address 0x93e7734 is 20 bytes inside a block of size 23 alloc'd
==23787== at 0x4C2AB80: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==23787== by 0x977796A: ??? (in /usr/lib/x86_64-linux-gnu/libOpenCL.so.1.0.0)
==23787== by 0x9777F28: ??? (in /usr/lib/x86_64-linux-gnu/libOpenCL.so.1.0.0)
==23787== by 0x4010139: call_init.part.0 (dl-init.c:78)
==23787== by 0x4010222: _dl_init (dl-init.c:36)
==23787== by 0x4014C6F: dl_open_worker (dl-open.c:577)
==23787== by 0x400FFF3: _dl_catch_error (dl-error.c:187)
==23787== by 0x40143BA: _dl_open (dl-open.c:661)
==23787== by 0x8B3902A: dlopen_doit (dlopen.c:66)
==23787== by 0x400FFF3: _dl_catch_error (dl-error.c:187)
==23787== by 0x8B3962C: _dlerror_run (dlerror.c:163)
==23787== by 0x8B390C0: dlopen@@GLIBC_2.2.5 (dlopen.c:87)
==23787==
For another image
$ ./depixelize -o out.svg Commander\ Keen.png
WARNING: Primary, infinite edge...
Segmentation fault
Any idea, what's wrong? Probably, some incompatibilities in shared libraries...
Hi, thanks for working on this project. Your results are looking very promising!
On line 366 of pixel_grid.cpp
, new_pos
sometimes goes negative. Then it's used as an index into seen
, and I get a segfault.
This seems to be the slowest part of the algorithm.. some ideas to speed it up (keep in mind I don't fully understand the workings of the algorithm):
Threading - Not sure if its possible, but perhaps its possible to do different sections of the image at a time.
Reducing iterations - The randomness of the optimizing seems to result in a high amount of iterations.. if there is a way to remove the random element from it then it should be possible to reduce the needed amount of iterations/guesses, as well as produce more consistent results.
I've already made a pull request that speeds it up quite a bit.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.