lcqysl/gems/text-rendering

Skill: Text Rendering

Description

This skill should be triggered when the user's request contains specific quotes, slogans, names, or any intended text that needs to appear accurately and aesthetically within the generated image.

• Trigger when: The user uses quotation marks (e.g., "Welcome"), or phrases like "a sign saying...", "a logo with the word...", "text on the shirt," or "the title is...".
• Do NOT trigger when: The request is purely for abstract imagery, textures, or scenes where no specific legible characters are required.

Instructions

When this skill is active, rewrite the user's prompt by applying the following six principles to ensure high-fidelity text generation and professional typographic layout:

1. Spatial Layout & Planning (Positional Clarity)

• Explicitly define the position of the text using natural language (e.g., "centered on the top third," "aligned to the bottom right," "wrapped around the center").
• Create a clear hierarchy if there are multiple text elements.

1. Line-Level Decomposition (Structural Breakdown)

• If the text is long, break it into distinct lines. Do not let the model "guess" the line breaks.
• Describe the structure: "A two-line inscription where the first line reads '...' and the second line reads '...'".

1. Contextual Carrier & Integration (The Surface)

• Clearly define what the text is written on (e.g., a neon sign, a weathered wooden plank, a sleek digital screen, or embroidered on a silk fabric).
• Ensure the text interacts with the carrier's texture (e.g., "the letters are deeply carved into the stone," "the neon light casts a pink glow on the surroundings").

1. Typographic Style & Materiality (Visual Identity)

• Specify font characteristics: (e.g., "bold sans-serif," "elegant cursive calligraphy," "distressed typewriter font," "3D chrome letters").
• Define colors, lighting, and materials (e.g., "glowing gold letters," "matte black ink," "translucent frosted glass characters").

1. Explicit Signaling (Syntactic Reinforcement)

• Use strong, directive verbs such as "rendered," "inscribed," "embossed," "spelled out," or "stenciled."
• Always wrap the target text in double quotation marks to signal its importance to the diffusion model's encoder.

Output Format

Return ONLY the final enhanced prompt text. Do not include any conversational filler, introductory remarks, or prefixes like "Enhanced prompt:".