jimeng-asset-promptor

Generate consistent asset prompts by reading project.json to apply art_style, media_type, and style_keywords. This enforces project style consistency across characters, props, and scenes using the jimeng-4.5 6-element method, outputting structured JSON and MD formats.

The 6-element method structures asset prompts using core subject, detail description, style positioning, scene and logic, technical parameters, and optimization commands. This produces precise, unambiguous prompts for characters, props, and scenes targeting 4K output.

Output asset prompts in JSON and MD formats by providing your project data and assets to the prompt generator. It automatically produces JSON for structured assets data and MD for readable text across all 4K/16:9 template outputs.

Yes, toxicity and validity checks are run automatically during asset prompt generation. This ensures safety and consistency across all generated prompts for characters, props, and scenes while documenting dependencies and components for traceability.

Yes, this applies directly to OpenCode projects with a flat skills directory and 4K output targets. It standardizes the creation of high-quality asset prompts for jimeng-4.5, ensuring consistent style, structure, and safety across all assets.

Maintain art style consistency by defining art_style, media_type, and style_keywords in your project.json. The generator reads these parameters to automatically unify the visual style across all character, prop, and scene prompts.