Copyright (C) 2015-2022 Filip Szczerek ([email protected])
version 0.6.5 (2022-04-10)
This program comes with ABSOLUTELY NO WARRANTY. This is free software, licensed under GNU General Public License v3 or any later version and you are welcome to redistribute it under certain conditions. See the LICENSE file for details.
- 1. Introduction
- 1.1. Processing and display back ends
- 2. User interface features
- 3. Supported image file formats
- 4. Image processing
- 4.1. Brightness normalization
- 4.2. Lucy–Richardson deconvolution
- 4.3. Unsharp masking
- 4.4. Tone curve
- 5. Saving/loading settings
- 6. Batch processing
- 7. Image sequence alignment
- 7.1. High-contrast features stablization (phase correlation)
- 7.2. Solar limb stabilization
- 8. Misc
- 9. Known problems
- 10. Downloading
- 11. Building from source code
- 11.1. Building under Linux and similar systems using GNU (or compatible) toolchain
- 11.1.1. Building under Ubuntu 18.04
- 11.1.2. Packaging
- 11.1.3. Building for macOS
- 11.2. Building under MS Windows
- 11.3. UI language
- 11.1. Building under Linux and similar systems using GNU (or compatible) toolchain
- 12. Acknowledgements
- 13. Change log
Homepage: https://greatattractor.github.io/imppg/
ImPPG (Image Post-Processor) performs common image post-processing tasks, in this order:
- brightness levels normalization
- non-blind Lucy–Richardson deconvolution with a Gaussian kernel
- blurring or sharpening via unsharp masking
- tone curve adjustment
(all the operations are optional). The processing is performed using 32-bit (single-precision) floating-point arithmetic. Settings for all the above steps can be saved to a file and used for batch processing of multiple images.
ImPPG can also align an image sequence, with possibly large and chaotic translations between images. This can be useful, for example, when preparing a solar time-lapse animation, where subsequent frames are offset due to inaccurate tracking of the telescope mount. Other possible applications are smoothing out of terrestrial landscape time-lapses or preparing raw frames (with serious image jitter) for stacking.
The alignment is performed (with sub-pixel accuracy) either via phase correlation, which automatically aligns on the most contrasty features, or by detecting and stabilizing the solar limb.
Starting with version 0.6.0, ImPPG contains two processing & display back ends: "CPU + bitmaps" and "GPU (OpenGL)". The latter is enabled by default and offloads all processing (except image alignment), as well as view scrolling and scaling, onto the GPU. On most systems it is faster than CPU processing by a factor of several or more.
Processing back ends can be switched via menu Settings/Back end
.
-
The processing controls panel, initially docked on the left of the main window, can be undocked or docked on the right.
-
Both the processing controls panel and the tone curve editor can be closed and later restored via the "View" menu or toolbar buttons.
-
Changes to the L–R deconvolution, unsharp masking settings and the tone curve are applied only to the current selection within the image (bounded by an inverted-brightness or dashed frame). To process the whole image, use the
Select (and process) all
option fromEdit
menu or the corresponding toolbar button. -
The current selection can be changed at any time by dragging the mouse with left button pressed. Changing the selection cancels all previously started processing (if any).
-
The smaller the selection, the faster the processing. For fine-tuning of L–R deconvolution’s sigma, it is recommended to use moderately small areas; this way moving the sigma slider will show the results almost instanteneously. The refreshing is slightly slower when the current zoom factor is not 100%.
-
The view can be zoomed in/out by using toolbar buttons,
View
menu items or mouse wheel. -
The view can be scrolled by dragging with the middle or right mouse button.
-
Size of toolbar icons can be changed via the
Settings
menu.
Accepted input formats: BMP, JPEG, PNG, TIFF (most of bit depths and compression methods), TGA and other via the FreeImage library, FITS. Image is processed in grayscale and saved in one of the following formats: BMP 8-bit; PNG 8-bit; TIFF 8-bit, 16-bit, 32-bit floating-point (no compression or compressed with LZW or ZIP), FITS 8-bit, 16-bit or 32-bit floating-point.
Output images produces by the sequence alignment function are saved as uncompressed TIFF with number of channels and bit depth preserved (except 8-bit palettized ones; those are converted to 24-bit RGB). Input FITS files are saved as FITS with bit depth preserved.
An illustrated processing tutorial can be found at http://greatattractor.github.io/imppg/tutorial/tutorial_en.html
ImPPG can automatically adjust the input image’s brightness levels (before all other processing steps) to fall into the user-supplied range of min
to max
(specified as percentage of the full brightness; 0% = black, 100% = white). The darkest input pixels will get the min
value, the brightest: the max
value.
Both min
and max
can be less than 0% and larger than 100% (this may clip the histogram). It is also allowed for max
to be less than min
; this will invert brightness levels (create a negative; also possible via tone curve adjustment, see sec. 4.4).
Normalization may be useful when processing a sequence of astronomical stack images captured during variable air transparency (i.e. with varying brightness levels).
Access by:
menu: Settings
/Normalize brightness levels...
ImPPG performs image sharpening via non-blind Lucy–Richardson deconvolution using Gaussian kernel. The kernel’s width is specified by Gaussian sigma; increasing this value makes the sharpening coarser.
Recommended number of deconvolution iterations: 30 to 70. Specify 0 to disable L–R deconvolution.
The Prevent ringing
checkbox enables an experimental function which reduces ringing (halo) around over-exposed (solid white) areas (e.g. a solar disc in a prominence image) caused by sharpening.
Access by:
Lucy–Richardson deconvolution
tab in the processing controls panel (on the left of the main window)
Unsharp masking can be used for final sharpening (independently of L–R deconvolution) or blurring of the image. The sigma parameter specifies the Gaussian kernel’s width; the larger the value, the coarser the sharpening or blurring. Amount
specifies the effect’s strength. Value < 1.0 blurs the image, 1.0 does nothing, value > 1.0 sharpens the image.
In this mode the amount varies, depending on the input (unprocessed) image brightness. For areas darker than threshold − transition width
the amount is set to amount min
. For areas brighter than threshold + transition width
the amount is set to amount max
. Between those values the amount changes smoothly from amount min
to amount max
.
An example situation where this mode can be useful is stretching of faint solar prominences with the tone curve. Typically this will also underline the image noise (especially if L–R was applied), as the darker image areas have small signal-to-noise ratio. By setting amount max
to a sharpening value (>1.0) appropriate for the solar disc interior, but amount min
to 1.0 (i.e. no sharpening) and threshold
to fall somewhere at the disc/prominences transition, the faint prominences and background are left unsharpened. They can be even smoothed out by setting amount min
to <1.0, while keeping rest of the disc sharpened.
Is is also allowed to set amount min
> amount max
.
Access by:
Unsharp masking
box in the processing controls panel (on the left of the main window)
The tone curve editor allows changing the input-output mapping of brightness levels. The tone curve is only applied after all the other processing steps. The histogram shown in the editor window’s background corresponds to the output (just the current selection) of previous processing steps before the application of the tone curve.
Access by:
menu: View
/Panels
/Tone curve
or the corresponding toolbar button
All settings described in 4.1-4.4 can be saved to/restored from a file (in XML format). A settings file is needed in order to use the batch processing function.
Access by:
menu: File
or the corresponding toolbar buttons
Multiple images can be processed using the same settings specified by an existing settings file (see sec. 5). Output images are saved in the specified folder under names appended with _out
.
Access by:
menu: File
/Batch processing...
The Input image files
list in the Image alignment
dialog contains the input files in the order they are to be aligned as a sequence. List items can be removed and shifted using the buttons in the upper right corner.
Enabling Sub-pixel alignment
results in a smoother animation and less image drift. Saving of output files will be somewhat slower; sharp, 1-pixel details (if present) may get very slightly blurred.
The output files are saved under names suffixed with _aligned
. FITS files are saved as FITS, the remaining formats: as TIFF files.
A general-purpose method. Attempts to keep the high-contrast features (e.g. sunspots, filaments, prominences, craters) stationary. In some cases it may be undesirable, e.g. in a multi-hour time-lapse of a sunspot nearing the solar disc’s limb; phase correlation would tend to keep the sunspot stationary, but not the limb.
Detects and matches the solar limb position. The more of the limb is visible in input images, the better the results. It is recommended that the images to align are already sharpened. Requirements:
- the disc has to be brighter than the background
- no vignetting or limb darkening exaggerated by post-processing
- the disc must not be eclipsed by the Moon
Operations such as full/partial brightness inversion (creating a negative) or applying a “darkening” tone curve can be applied after the alignment.
Access by:
menu: Tools
/Align image sequence...
ImPPG stores certain settings (e.g. the main window’s size and position) in an INI file, whose location is platform-dependent. On recent versions of MS Windows the path is %HOMEPATH%\AppData\Roaming\imppg.ini
, where %HOMEPATH%
usually equals C:\Users\<username>
(but if OneDrive is enabled, the file may be located there). On Linux the path is ~/.imppg
.
-
Starting with wxWidgets 3.1.5 on Linux, GL Canvas uses EGL by default. If the GLEW library used for building ImPPG is not built with EGL support, the call to
glewInit
will fail. Solution: either use GLEW built with EGL, or build wxWidgets adding-DwxUSE_GLCANVAS_EGL=OFF
to its CMake invocation. -
ImPPG remembers and restores the positions and sizes of the main program window and the tone curve editor window. It may happen (rarely) the desktop environment interferes with window positioning and e.g. the tone curve window remains off-screen. In that case, use the menu command
Settings
/Reset tone curve window position
, or delete the ImPPG configuration file (see section 8). -
(OpenGL mode, MS Windows 8.1, Intel HD Graphics 5500, driver 10.18.14.5074) Cubic interpolation mode is broken (a shader compiler bug, or perhaps a texture read count limitation); use "View/Scaling method/Linear" instead. Note that the same hardware works fine under Linux (Fedora 29 5.3.11-100, Mesa 18.3.6).
-
(wxGTK 3.0.2, Fedora 20) All the toolbar buttons react correctly, but their displayed state may be incorrect after using the
View
menu items/close boxes of tone curve editor and processing controls panel. Cause: sometimes check tool’s Toggle() method does not work. -
(wxGTK) As of 2/2015, some of GTK themes function incorrectly (e.g. “QtCurve”, but not “Raleigh”). In ImPPG this may manifest as follows:
File open
dialog does not show any files after it is opened; the files do appear when the dialog is resized by the user- program crashes when trying to select the output directory in Batch dialog using the
Choose folder
dialog - program hangs when trying to change file type in the
Open image file
dialog
Solution: change the GTK theme to "Raleigh" (e.g. in Fedora use the "GTK+ Appearance" tool).
ImPPG source code and binaries for MS Windows and Ubuntu 18.04 (x86-64) can be downloaded from: https://github.com/GreatAttractor/imppg/releases
Building from source code requires a C++ compiler toolchain (with C++17 support), CMake, Boost libraries v. 1.57.0 or later (though earlier versions may work) and wxWidgets 3.0 (3.1 under MS Windows). Support for more image formats requires the FreeImage library, version 3.14.0 or newer. Without FreeImage the only supported formats are: BMP 8-, 24- and 32-bit, TIFF mono and RGB, 8 or 16 bits per channel (no compression). FITS support (optional) requires the CFITSIO library. Multithreaded processing requires a compiler supporting OpenMP.
To enable/disable usage of CFITSIO, FreeImage and GPU/OpenGL back end (they are enabled by default), edit the config.cmake
file.
To remove any created CMake configuration, delete CMakeCache.txt
and the CMakeFiles
folder.
Note: CMake relies on the presence of the wx-config
tool to detect and configure wxWidgets-related build options. Sometimes this tool can be named differently, e.g. in Fedora 23 with wxGTK3 packages from repository it is wx-config-3.0
. Older versions of CMake may not accept it. This can be remedied e.g. by creating a symlink:
$ sudo ln -s /usr/bin/wx-config-3.0 /usr/bin/wx-config
Download the source code manually or clone with Git:
$ git clone https://github.com/GreatAttractor/imppg.git
In the source folder, run:
$ mkdir build
$ cd build
$ cmake -G "Unix Makefiles" -DCMAKE_BUILD_TYPE=Release ..
This creates a native Makefile
. Unless you edit config.cmake
again, from now on there is no need to run CMake.
To compile ImPPG, run:
$ make
You will find imppg
executable in the build
folder.
To install ImPPG system-wide, run from the build
folder:
$ sudo cmake -P cmake_install.cmake
To uninstall, run:
$ cat install_manifest.txt | sudo xargs rm
To use a different installation prefix, add -DCMAKE_INSTALL_PREFIX=<my_dir>
to the initial CMake invocation.
The following packages are needed for building under Ubuntu 18.04:
git cmake build-essential libboost-dev libwxgtk3.0-gtk3-dev libglew-dev pkg-config libccfits-dev libfreeimage-dev
The default GCC version (7.x) is too old. Install and enable GCC 8 (example instructions: https://linuxize.com/post/how-to-install-gcc-compiler-on-ubuntu-18-04/
). (Do not choose GCC 9, otherwise the built binary will not run on a clean Ubuntu 18.04 due to too old a version of libstdc++
.)
After building imppg
, it can be either installed as in section 11.1, or a Debian package can be created and installed with apt
(see 11.1.2).
In order to create a binary package, modify appropriately the final statement in CMakeLists.txt
:
include(packaging/ubuntu_20.04.cmake)
before calling CMake and building (see the packaging
subdirectory for possible targets). After a successful build, the package can be created by running:
$ cpack
Note that building needs to be performed in an environment corresponding to the selected target system, so that proper shared objects are linked to (“environment” = a full system installation, a Docker image, or similar).
Note: macOS build and support is still work-in-progress
To build ImPPG for macOS you will need Xcode and Homebrew installed.
OpenMP is not supported yet as Apple has disabled OpenMP in clang toolchain distributed with Xcode. As of September 2021 the build method has been verified on macOS Big Sur 11.6 and Xcode 12.5.1.
Install following libraries with Homebrew:
$ brew update
$ brew install boost cfitsio cmake freeimage glew mesa pkg-config wxwidgets
Now follow Linux build steps:
$ mkdir build
$ cd build
$ cmake -G "Unix Makefiles" -DCMAKE_BUILD_TYPE=Release ..
$ make -j8
$ make install
$ imppg
The provided CMakeLists.txt
supports the MSYS2 build environment. With manual configuration, other toolchains can also be used (e.g. MS Visual Studio).
In order to build ImPPG (64-bit) under MSYS2, follow its installation instructions at http://www.msys2.org/. Then open the MSYS2/MinGW64 shell (after default installation: c:\msys64\mingw64.exe
) and install the ImPPG's dependencies by running:
$ pacman -S git mingw-w64-x86_64-cmake base-devel mingw-w64-x86_64-toolchain mingw-w64-x86_64-boost mingw-w64-x86_64-cfitsio mingw-w64-x86_64-freeimage mingw64/mingw-w64-x86_64-glew mingw64/mingw-w64-x86_64-wxmsw3.1
Download the ImPPG's source code manually or clone it with Git:
$ git clone https://github.com/GreatAttractor/imppg.git
In the ImPPG source folder, run:
$ mkdir build
$ cd build
$ cmake -G "MSYS Makefiles" -DCMAKE_MAKE_PROGRAM=mingw32-make -DCMAKE_BUILD_TYPE=Release ..
This creates a native Makefile
. Unless you edit config.cmake
again, from now on there is no need to run CMake.
To compile ImPPG, run:
$ mingw32-make
You will find imppg.exe
executable in the build
folder. It can be run from MSYS2 shell, from the ImPPG source folder:
$ build/imppg.exe
To run ImPPG from Windows Explorer, the subfolders images
, pl
, shaders
and all the necessary DLLs must be placed at the same location as imppg.exe
. See the MS Windows binary distribution (imppg-win64.zip
) for an example.
ImPPG supports multiple user interface languages using the wxWidgets built-in internationalization facilities. All translatable strings in the source code are surrounded by the _()
macro. Adding a new translation requires the GNU gettext
package and consists of the following steps:
- extraction of translatable strings from sources into a PO file by running:
$ xgettext -k_ src/*.cpp src/*.h -o imppg.po
-
translation of UI strings by editing the
msgstr
entries inimppg.po
-
converting
imppg.po
to binary form by running:
$ msgfmt imppg.po -o imppg.mo
-
placing
imppg.mo
in a subdirectory with name equal to the language code (e.g.pl
,fr-ca
) -
adding the language to the arrays inside
c_MainWindow::SelectLanguage()
(main_window.cpp
)
Binary distribution of ImPPG needs only the MO (binary) language files. Beside the imppg.mo
file(s), also the wxWidgets translation is needed (for strings like standard menu items, e.g. “Open”; control captions, e.g. “Browse” etc.). Those can be found in <wxWidgets_source_root>/locale
. On Windows, the wxWidgets file <language>.mo
has to be available as <imppg_root_folder>/<language>/wxstd3.mo
. On operating systems with a common location for all .mo
files (e.g., Linux), it is sufficient to have a wxWidgets installation.
Russian and Ukrainian translations: Rusłan Pazenko. German translation: Marcel Hoffmann.
0.6.5 (2022-04-10)
- Bug fixes
- Invalid behavior of the Brightness Normalization dialog for certain combinations of system language and regional settings
- Saving and loading of settings files for certain combinations of system language and regional settings
0.6.4 (2021-10-30)
-
New features
- German translation
-
Enhancements
- File type filter is remembered when opening and saving images
0.6.3 (2021-04-13)
-
Enhancements
- Opening an image file by dragging and dropping onto the ImPPG window
-
Bug fixes
- Crash when batch processing in CPU mode with deringing enabled
0.6.2 (2020-07-04)
- New features
- Russian and Ukrainian translations
0.6.1 (2020-01-25)
-
Bug fixes
- Invalid batch processing results in OpenGL mode
-
Enhancements
- Tone curve window position reset command
0.6.0 (2019-12-23)
-
New features
- GPU (OpenGL) back end for much faster processing
-
Enhancements
- View scrolling by dragging with the right mouse button (previously: with the middle button)
- Zooming in/out with the mouse wheel (previously: Ctrl + mouse wheel)
0.5.4 (2019-02-02)
-
New features
- Configurable look of the tone curve editor
-
Enhancements
- Last loaded settings file name shown in toolbar
0.5.3 (2017-03-12)
- Bug fixes
- Fixed filling the list of recently used settings under Windows
0.5.2 (2017-01-07)
- Bug fixes
- Fixed drawing of histogram and tone curve
0.5.1 (2016-10-02)
-
New features
- List of most recently used settings files
-
Enhancements
- High-resolution toolbar icons
- Improved tone curve drawing performance on high-resolution displays
0.5 (2016-01-02)
-
New features
- Adaptive unsharp masking
-
Enhancements
- Numerical sliders can be scrolled with cursor keys
- Processing panel width is preserved
- Using CMake for building
0.4.1 (2015-08-30)
-
Enhancements
- Numerical sliders use 1-pixel steps instead of hard-coded 100 steps
- Output format selected in batch processing dialog is preserved
- Unsharp masking not slowing down for large values of sigma
- Increased the range of unsharp masking parameters
-
Bug fixes
- Invalid output file name after alignment if there was more than one period in input name
- Crash when a non-existing path is entered during manual editing
- Program windows placed outside the screen when ImPPG was previously run on multi-monitor setup
- Restored missing Polish translation strings
0.4 (2015-06-21)
-
New features
- Image sequence alignment via solar limb stabilization
- FITS files support (load/save)
- Zooming in/out of the view
-
Enhancements
- View scrolling by dragging with the middle mouse button
- Logarithmic histogram setting is preserved
-
Bug fixes
- Tone curve in gamma mode not applied during batch processing
0.3.1 (2015-03-22)
- New features
- Polish translation; added instructions for creating additional translations
0.3 (2015-03-19)
-
New features
- Image sequence alignment via phase correlation
-
Enhancements
- Limited the frequency of processing requests to improve responsiveness during changing of unsharp masking parameters and editing of tone curve
-
Bug fixes
- Incorrect output file extension after batch processing when the selected output format differs from the input
0.2 (2015-02-28)
-
New features
- Support for more image file formats via FreeImage. New output formats: PNG/8-bit, TIFF/8-bit LZW-compressed, TIFF/16-bit ZIP-compressed, TIFF/32 bit floating-point (no compression and ZIP-compressed).
-
Enhancements
- Enabled the modern-look GUI controls on Windows
-
Bug fixes
- Selection border not marked on platforms w/o logical raster ops support (e.g. GTK 3)
0.1.1 (2015-02-24)
- Bug fixes
- Blank output files after batch processing when L–R iteration count is 0
0.1 (2015-02-21)
- Initial revision