Everyone on the Pixelmator Team agrees that creating new features is the most fun part of work. When developing Pixelmator, we experiment with lots of things (new technologies, code-level tricks, user interface stuff, etc.), and we create lots of mockups, sketches and prototypes…
However, we adopt only a few features that we come up with.
For various reasons, we remove numerous features from our products one or two weeks before they are released to the public. For example, some Mac GPUs do not have some specific features that a feature uses, perhaps the code or UI was complete nonsense or the user interface looks or feels terrible or appears cluttered. In addition, there might be usability issues. We have created hundreds of bad features that are discarded.
We try to ship only those features that are useful and complete. In our opinion, learning from our mistakes only makes us a stronger company.
Some features that we removed the last minute include Shadows and Highlights (in Pixelmator 1.5 codenamed Spider), reflection filter (1.4 Sprinkle), indexed color for GIF and PNG Indexed in save for Web (1.5 Spider), precise numbers (1.3 Tempo), black and white (1.4 Sprinkle), page bounds (1.3 Tempo), gradient map (1.4 Sprinkle), document presets in cover flow, and send to Mobile Me (1.5 Spider).
We removed the aforementioned features for the following reasons:
- Shadows and Highlights was too slow.
- Reflection filter was too heavy and had too many controls.
- Color indexing in save for Web had no future (Web no longer requires it) and added too much clutter to our easy-to-use save for Web feature
- Precise numbers looked cheap.
- Black and White adjustment had the wrong user interface that nobody liked.
- Page bounds was not useful.
- Gradient map was too simple and useless; it was better suited for Microsoft Paint.
- Document presets in cover flow looked wrong.
- Send to Mobile Me had some quality issues.
The removal of indexed color for GIF and PNG resulted in the biggest loss due to the fact that it took us nearly two weeks to develop the quantization engine. Although the indexing looked and worked great, we removed it following the last Pixelmator Spider 1.5 inside pre-release review. We thought that indexing had no future; we still think this. I assure you that I am completely confident that we did the right thing.
There were many more canceled features like those mentioned in Pixelmator. Unfortunately, I am not able to discuss them or show you how they look because some include secret UI objects or will appear in upcoming Pixelmator releases. I hope that you have an idea of how difficult and interesting it is to develop a single feature in any software application. I am sure that Pixelmator is not the only app built in this way. It is also likely that we will cancel some features of the Pixelmator 1.6 Nucleus; we have already canceled the new Crop Tool. It is a good thing, as I am sure that it will be much better in the 1.7 version.
P.S. A day or two ago, we created a fast version of Shadows and Highlights, which we might add to Pixelmator Nucleus 1.6. Further, you can try the slower experimental Shadows and Highlights adjustment now at your own risk by taking the following steps: Press Control + Option + Command + H when in Pixelmator.
Wednesday, 7 April 2010. Posted by Saulius.
