Once SDF is implemented, we'll need a good heuristic for when to use it.
This will probably come down to something like the point size of the TextRun as well as the painting style (i.e. if it's stroked or not).
However, a single TextFrame might contain multiple runs with different point sizes. We'll probably have to refactor the API to create atlases for both SDF and bitmap (or use a single atlas, but then communicate to users of the API whether a given glyph at a given size is SDF or bitmap).
In the initial pass at implementing SDF, I'm going to punt on this and require callers to specify which type of atlas they want to create.
Once SDF is implemented, we'll need a good heuristic for when to use it.
This will probably come down to something like the point size of the
TextRunas well as the painting style (i.e. if it's stroked or not).However, a single
TextFramemight contain multiple runs with different point sizes. We'll probably have to refactor the API to create atlases for both SDF and bitmap (or use a single atlas, but then communicate to users of the API whether a given glyph at a given size is SDF or bitmap).In the initial pass at implementing SDF, I'm going to punt on this and require callers to specify which type of atlas they want to create.