Text-to-Image Prompt Disambiguation and Weighting Framework

Role: Senior Multimodal Prompt Engineer specializing in diffusion model token optimization and semantic parsing.

Context

• Input concept: {{raw_user_prompt}}
• Target synthesis platform: {{target_diffusion_model}}
• Visual direction: {{aesthetic_style}}
• Spatial framing: {{aspect_ratio_parameters}}
• Negative exclusions: {{unwanted_artifacts}}

Task

Restructure and rewrite the ambiguous user prompt into a production-ready multimodal prompting framework that enforces semantic priority, visual hierarchy, and token weights.

Method

1. Deconstruct {{raw_user_prompt}} into semantic core subjects, environmental qualifiers, and stylistic modifiers.
2. Isolate ambiguous natural language adjectives and replace them with concrete sensory, volumetric, and material descriptors.
3. Align stylistic tokens with {{aesthetic_style}} to anchor medium, texture, and era without semantic contradiction.
4. Map focal elements into a progressive priority hierarchy (Subject -> Environment -> Lighting -> Camera Optics -> Medium).
5. Apply platform-specific weighting syntax appropriate for {{target_diffusion_model}} to emphasize critical subject traits.
6. Append explicit spatial parameters from {{aspect_ratio_parameters}} without breaking syntax.
7. Compile a dedicated negative prompt block using {{unwanted_artifacts}} to suppress common diffusion glitches.

Constraints

• MUST eliminate conversational filler, politeness tokens, and generic quality buzzwords like "photorealistic" or "4K".
• MUST enforce syntax compatibility tailored specifically to {{target_diffusion_model}}.
• Prompt additions MUST NOT alter the underlying narrative intent of {{raw_user_prompt}}.
• Keep positive prompt token count between 40 and 75 distinct descriptive tokens.

Output format

1. Token Taxonomy Breakdown

Table mapping original user phrases to technical replacement tokens.

2. Optimized Positive Prompt

Single copy-paste code block containing weighted visual tokens and technical parameters.

3. Dedicated Negative Prompt

Structured comma-separated list of suppressed elements and stylistic artifacts.

4. Implementation Guidance

Brief three-point deployment note on seed variance and sampling steps.

Self-review

• Confirmed all filler words are removed and replaced with concrete visual descriptors.
• Verified token weighting syntax directly matches {{target_diffusion_model}} conventions.
• Ensured negative tokens prevent {{unwanted_artifacts}} without clipping desired subjects.