Text Processing Pipeline

• Modern Text Processing Pipeline .

  • Well explained.

Segmentation

• Divides the input text into several substrings called “runs”.

• Runs are sequences of characters that share a common 'direction' (left-to-right or right-to-left) as well as a common script.

• A “hard line break”, like an explicit newline or carriage return character, signals the end of a run.

• Segmentation only produces bounds between runs. It does not modify the input text in any way.

Shaping

• Instead of drawing an ASCII / UTF-8 character directly, a shaping engine is used to perform operations that make the text compatible with various languages.

• Operations :

  • Substitution :

    • Necessary for ligatures.

      • For Latin languages this is not necessary, but languages like Arabic, etc., are completely wrong without ligatures.

  • Reordering :

    • Important for some characters outside the Latin alphabet.

  • Positioning :

    • Depending on the context, characters may want to move slightly.

    • One example of this is Kerning .

      • Bringing one character closer to another is an example of this, being super important for cursive fonts.

    • Depends on directionality.

      • Latin:

        • Left to Right -> Top to Bottom.

      • Arabic:

        • Right to Left -> Top to Bottom.

      • Japanese (Hiragana, Katakana, Kanji):

        • Top to Bottom -> Right to Left.

• Libs :

  • HarfBuzz.

  • Uniscribe.

Kerning

• Adjustment of horizontal space between specific pairs of glyphs (for example, “A” followed by “V”).

• Kerning tables in font files specify fine-tuning offsets so that letter combinations appear optically balanced.

• Applying :

  • Shaping Engines :

    • It's the primary mechanism for applying kerning in many modern text rendering systems.

  • Manual :

    • A layout engine or rendering system can apply kerning adjustments independently of the shaping engine, either using font metrics directly or user-defined adjustments.

  • Design Tools :

    • Software like Adobe Illustrator or Figma can allow manual kerning (user overrides), separate from the shaping engine's automatic adjustments.

  • Web Browsers / CSS :

    • In web environments, the font-kerning  and letter-spacing  CSS properties may influence kerning behavior, even disabling shaping-based kerning.

• Kerning Data can reside in :

  • GPOS  table :

    • More flexible, modern positioning.

    • OpenType.

  • kern  table

    • Legacy.

    • TrueType.

Rasterization

• The process of converting a vector font to a bitmap font.

• Challenges :

  • Hinting :

    • Adjusting glyph shapes to align with pixel grids for clarity at small sizes.

  • Anti-aliasing :

    • Smoothing jagged edges using grayscale or subpixel techniques.

  • DPI Scaling :

    • Scaling fonts appropriately for the screen’s physical density (dots per inch).

Layout

• Text layout is the process of wrapping lines to fit a certain width, and optionally making the text “fit better” through a bunch of adjustments, like justification and hyphenation.

• These adjustments are typically very dependent on the script being processed.

Hinting

• Instructions embedded in an outline font (TrueType or PostScript) that tell the rasterizer how to align strokes to the pixel grid at small sizes.

• Proper hinting reduces artifacts like uneven stroke widths or blurry edges when rendering on low-resolution screens.