text recognition errors with some fonts
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
SikuliX |
New
|
Undecided
|
Unassigned |
Bug Description
Our app has a default set to use Tahoma for certain display elements to guarantee that japanese and chinese characters are properly rendered on the app.
When I use sikuli on Windows (rc2) text recognition with the font set to Tahoma (11).
It fails to find certain text, mainly letter combinations involving an 'l'.
Examples include: ClearBits, and Playlists. I assume it's because the character spacing in so close together.
To reproduce:
1. Launch Miro (nightly or beta build from http://
2. Launch Sikuli IDE
3. Try find("ClearBits") or click("Playlists")
Result: recognition fails - even if searching in a very limited region.
4 Try find("Bits") or 'lists'
Result: recognition successful.
Additional info - I had similar text recognition issues using the default Ubuntu font on linux maverick, however I was able to workaround by changing the system font preferences to something more machine readable.
see consolidated bug 710586