Metadata Documentation

Overview

The metadata.csv file contains detailed information about each generated text sample, including typography, styling, and visual properties.

Fields

1. sample_id

• Type: String
• Format: sample_XXXXX (5-digit zero-padded number)
• Description: Unique identifier for each sample
• Example: sample_00042

2. text

• Type: String
• Description: The actual text content rendered in the image
• Possible Values:
  • English letters (a-z, A-Z)
  • Digits (0-9)
  • Symbols (!@#$%^&*()_+-=[]{}|;:',.<>?/~`)
  • Spaces
• Length Range: 5-20 characters (after trimming)
• Example: Hello World!

3. font_family

• Type: String
• Description: The font family used for rendering
• Possible Values:
  • Arial
  • Georgia
  • Times New Roman
  • Courier New
  • Verdana
  • Tahoma
  • Trebuchet MS
  • Impact
  • Calibri
  • Cambria
  • Comic Sans MS
  • Consolas
• Example: Arial

4. font_size

• Type: Integer
• Unit: Pixels (px)
• Description: The font size of the text
• Range: 140-160 px
• Example: 132

5. font_weight

• Type: String
• Description: The weight (thickness) of the font
• Possible Values:
  • normal - Regular weight
  • bold - Bold weight
• Example: bold

6. font_style

• Type: String
• Description: The style of the font
• Possible Values:
  • normal - Upright text
  • italic - Slanted text
• Example: italic

7. color

• Type: String (Hex color code)
• Format: #RRGGBB
• Description: The color of the text
• Range: #000000 to #888888 (dark to medium gray tones)
• Example: #3A5F7D

8. outline_thickness

• Type: String (Float) or Empty
• Unit: Pixels (px)
• Description: The thickness of the text outline/stroke
• Possible Values:
  • Empty string (no outline applied)
  • 1.0 to 2.0 (outline thickness range)
• Probability: ~50% chance of having outline
• Example: 1.5 or `` (empty)

9. outline_color

• Type: String (Hex color code) or Empty
• Format: #RRGGBB
• Description: The color of the text outline
• Possible Values:
  • Empty string (no outline applied)
  • #999999 to #FFFFFF (light gray to white)
• Example: #E5E5E5 or `` (empty)

Sample Row

sample_id,text,font_family,font_size,font_weight,font_style,color,outline_thickness,outline_color
sample_00001,Data Science,Arial,128,bold,normal,#2C3E50,1.8,#FFFFFF
sample_00002,AI & ML,Georgia,135,normal,italic,#1A2332,,

Notes

• Empty fields for outline_thickness and outline_color indicate no outline was applied
• All color values use uppercase hexadecimal format
• Font sizes are uniformly distributed within the specified range
• Text content may contain mixed English letters, numbers, and symbols