This article on NewsForge talks about efforts to create an open-standard Photoshop-like file format, capable of storing raster (pixel-based) images, text, paths, etc. This is interesting to me because such a format would be a candidate for the XVG and/or Icon Machine 4 file format, instead of the SVG-based format that I currently have in mind.

I was disappointed to find that the PSD file format isn’t open (though according to the article it used to be). You can use QuickTime to get at individual layer images, but that doesn’t get you everything. So I’m going to be following OpenRaster (a tentative title that doesn’t appear on that wiki page) and similar efforts, and see if I end up using a resulting standard or just some of their ideas.

As if I didn’t have enough going on already, I’ve started on a new project: an SVG-based Cocoa drawing application, tentatively titled XVG. Actually the focus will be more on creating it as a library, with a sample application to show off the library’s features.

This will be an open source project, and my plan is to use it in Icon Machine 4.0. In effect, I’ll be releasing a large part of Icon Machine as open source. Hopefully other people will participate and find uses for it in other applications.

As I think I’ve described before, I want to go with SVG for Icon Machine because it has nearly all the features I use in Photoshop when I’m creating my own icons - especially filters (I think the only thing that’s missing is inner/outer glow). Another advantage of choosing a public standard such as SVG is that it provides a well thought out and well defined feature set that I can use as a target.

The goal will be to make it what is officially termed a Conforming SVG Generator and a Conforming Static SVG Viewer (except I don’t plan to support conditionals). If someone else wanted to join the project and add the animation and interaction features, that would be fine with me. I plan to make the project officially public once I have the basic shapes working for reading, writing, rendering, and editing… unless someone wants to join in sooner than that.

If you’re wondering how this relates to the Inkscape project, there are two major differences. First, XVG will be Mac OS X native, using Cocoa and Core Graphics (Quartz). Second, Inkscape is intended specifically to be an SVG authoring tool, including the ability to view the SVG source for the drawing. My goal with XVG is to have a drawing program whose feature set and file format corresponds to that of SVG, but without exposing those implementation details to the user. That’s why I don’t plan to slavishly follow the definition of a Conforming Static SVG Viewer. This, of course, is because Icon Machine would have no cause to expose its SVG underpinnings. But again, if someone with different needs were to join the project, I’m sure we could work something out.

MacWorld today posted a Mac OS X Hints article about using custom icons, in which Icon Machine gets a little recognition. It’s always fun to see my name - or my app’s name - in lights now and then.

Work on Icon Machine 4 continues. Another cool things I’m realizing about Cocoa bindings is that not only do they make it easy to quickly throw an interface together, but you can also keep them around under-the-hood as you transition from your prototype UI to the final version. This way you can mess around with the UI without disturbing your data-management stuff. I assume this is how Delicious Library was developed. There are still some things that confuse me about creating my own bindings; when I figure those details out I’ll post what I’ve learned.

I’m finally taking a serious crack at Windows icon importing/exporting in Icon Machine. My plan had been to use FreeImage, since it seemed to be a good open-source implementation. The first obstacle I ran into was that the open source nature of the code made it tricky to get it to work in CodeWarrior. While I had converted Icon Machine to Mach-O, I hadn’t made the switch to XCode yet.

Now I’m in a position to try this again. Things are still complicated, though. 32-bit icons are working reasonably well, but indexed images are trickier. I’ve been tempted to just dump indexed images since they seem old-fashioned, but then there would be lots of icon files that Icon Machine couldn’t open. So I decided to go all the way.

The next problem is translating between the FreeImage API and the Quartz and NSImage APIs. At this point I’m not entirely sure how to get the transparency data for an indexed image, and it’s enough to get me thinking I may as well eliminate the FreeImage layer. The purpose of FreeImage is actually to give a common interface to a variety of image formats, so using it for only one format isn’t quite so useful.

There’s a bunch more stuff planned for Icon Machine 4. Originally, I designed it closely around the Mac OS icon format. Now I’m rethinking it, with a more generalized definition of what an icon is. This ought to enable me to keep up with things for a while yet.

Anybody else an Art of Noise fan?

Anyway, I just posted a sort of emergency stealth update: Icon Machine 3.2.1. Unless you’re Japanese and/or haven’t registered yet, the fixed problems won’t affect you.

Icon Machine has just been upped to version 3.2. Mostly minor modernizations. It uses the Mac OS X font panel for the text tool, with all the fancy typography settings involved, plus Quartz text drawing. It’s also Carbon Event driven, which basically means it uses less CPU time when it’s idle. One bug that was fixes is saving to a file server - you no longer get a weird error code when you try to do that.

This version also requires Mac OS X 10.2; if you’re still using OS 9 or 10.1, you’ll have to stick with version 3.1.

And before anybody asks, there is one problem with the font panel: you can’t select Helvetica Oblique or Courier Oblique. It’s actually a bug in Mac OS X.

Every once in a while someone buys Icon Machine when they really shouldn’t have. Usually it has been because they didn’t realize that it’s a Mac-only app and they need something for Windows. Sometimes they don’t try it out before buying it, and discover later that it doesn’t meet their needs.

It’s a surprisingly difficult task to put the right words in the right places to keep this from happening. It didn’t occur to me at first that anyone would think Icon Machine runs on Windows. I need to take a step back and reevaluate things from the viewpoint of a total novice. Make no assumptions. Walk myself through the download and first run, closing my eyes to all the knowledge I take for granted.

It all seems very Zen.

One more bug I want to fix before releasing Icon Machine 3.2: saving to file servers. It’s a weird thing. It successfully creates the file, but then it gets a permissions error when it tries to write to it. Why I can do one but not the other doesn’t seem to make sense. It seems like such a simple thing.

There are of course lots of other improvements I want to make, and I’m still deciding how gradual or dramatic the transition will be from version 3 to version 4. I’ve thought about throwing in support for 256x256 images, but since not even the Finder uses them yet I think I might as well wait.