Word Tracer Editor

When we first started working on Word Tracer - Learn Chinese (WT-LC from now on), we were going to do a few hundred characters.  Each character was generated using Photoshop.  As we got into it, we decided to increase the character offerings to 1500 characters and ran into a problem.  Using Photoshop-generated character, each character takes up about 60KB of space. Multiply that with 1500 characters, that's 90MB just for character assets alone.  Clearly this approach isn't scalable.  We want our users to be able to download our app quickly, and ideally over the 3G network.  The current requirement for 3G download is 20MB max.  Another issue is that in the future we may expand our offering to 3000 characters or beyond, using the Photoshop approach means the app size will approach 200MB.  So we decided to generate the character via code.  The end result is a slight compromise on the oveall character quality (If you look at the characters on WT-LC's Home screen, you will see what a Photoshop-generated character looks like), but the resulting app is much smaller (14.9 MB currently).  There is another side benefit from going with code-generated approach.  The guide markers you see on each character have to be placed manually.  To help with that, we have created an editor to help with this (see picture above)  When we were using Photoshop approach, the editor has to deal with two sets of files - the character assets and the marker placement data.  Now the editor only needs to load the marker placement data file and generate the appropriate character in code, which greatly simplifies collaborations (we have five people responsible for placing the markers on all 1500 characters).  Over the course of the development, the editor became more and more powerful as we find ways to cut down on time to place the markers.  Because we didn't establish an agreed-upon standard in marker placements, we were getting inconsistent placements from different people.  For example, for the same character radical, one person placed three markers while another person placed four.  After discovering the inconsistency problem, we established a standard and spent additional two weeks re-repositioning all markers.

Our near future plan is to release Word Tracer Editor for download so users can generate their own character data.  So instead of five people placing markers for weeks like we did for WT-LC, the upcoming release of Word Tracer for other languages (Japanese, Korean, Hebrew, Thai, etc.) will be community-generated.