ImageFlow technical blog

This is the first version of Imageflow I regard as stable. Frido and me tried to test it thoroughly over the past week and hopefully we missed nothing. On the other hand, if we do missed something, we have the best reasons to continue development.

Description

ImageFlow is a node-based Macro-Editing interface for ImageJ.

More details in the information sheet.

Documentation on the Unit-Element XML-Specification.
The new version brought some slight updates to the specification. The changes are noted in the document.

Installation

Requires: Java Runtime Environment 1.5
Tested on Windows XP, Ubuntu and Mac OS X 10.4

Run as standalone application:

1. Download the archive


2. Extract into a directory on your harddrive


3. Launch the Imageflow_.jar

Run as ImageJ-Plugin:

1. Download the archive and extract the included Imageflow-folder.


2. Navigate to the plugins-directory of your ImageJ-installation


3. Copy the Imageflow-folder into ImageJ's plugin-folder


4. Restart ImageJ and a new menu-item for Imageflow appears in the Plugins-Menu.

Download

Download Imageflow 1.0 for Windows, Linux and Mac OSX

Download Imageflow 1.0 Java Sources

As always, I appreciate bugreports or other feedback, especially regarding the direction this tool should take.

Credits for this go to Johannes at Tudor.

Obviously I'm using ImageJ as the basis for my programm. I have been looking for a way to start ImageJ headless, having an instance without the WindowFrame, which starts up automatically when creating a new ImageJ-instance. I asked a few colleagues here at Tudor and we dived into the topic. In it's constructor ImageJ has a call for the show()-method, which instantly displays the ImageJ-window. show() has been deprecated for 2 Java-versions and so we thought, well let's override it locally. We empty the show()-method and on testing, nothing happened. Nothing like in "nothing showed up". We dived deeper. We couldn't use setVisible(boolean) the replacement of show() anymore, looked in the java-source only to find, that setVisible() does nothing more than using the old deprecated methods of show(). Nice one!

So we hacked a boolean-variable, which defines outside the class, whether or not the show should do something or not. Here is the code. Enjoy!

class StartIJheadless {

static boolean hide = true;

public static void main(String[] args) throws Exception {

ImageJ ij = new ImageJ() {

@Override
public void show() {
if (!hide)
super.show();
}

};
// ImageJ starts without GUI

hide= false;
ij.setVisible(true);
// it shows up
}
}

I posted a request in the mailinglist of ImageJ to introduce a clean way to start IJ headless. This can't be the solution.

I had an interesting talk yesterday, with an interesting question. What's the direction?
Honestly that's a very good question. I have the feeling ImageFlow is a very nice tool, but I don't really know how it is used and therefore in which direction I should concentrate.

Speaking with Andreas and Frido here at Tudor we have many ideas, hopefully one for a proper direction as well.

So working on the programm for a week now, we get quite a few things done for the first stable version.

I added a few new Unitelements. for example for merging RGB-channels, for FFT, Saving and more. Adding new units is kind of a in-between task, if I get bored, I add a new one.

Alot of work has been spent fixing the algorithm for sorting the units to generate the final macro. I introduced Frido to the topic and explaining the algorithm helped finding a few issues. Rubber duck debugging works.

Frido also fixed a problem with the GUI. You can now actually resize the sidepane on the left.
Also new is to display a grid on the workspace and to align all units to this grid. To sum up the obvious new features is to collaps units to take up less space on the workspace.



Here is a screenshot, work continues with debugging.

I have a few news.

I arrived in Luxembourg today to start a 4 months internship at the Research Institute Henri Tudor. I'll work here for the next 4 months. My work will be on some of the projects in progress here, as well as continued development of Imageflow. I repeat, after finishing my thesis about the programm, I can continue it's development and I'll be joined by Friedrich Maiwald, a colleague of mine from Berlin. So twice the fun.

During the week we will brainstorm for the next version and lay out a few milestones. First task will certainly be to get the first stable version based on the march-beta out. There are still a few quirks I'd like to address, but I don't want to be postpone anymore. So hopefully the new version and the related sources will be out soon.

I will write more often now.