Implement SID_ZImagePromptGenerator with externalized config
Features:
- 6-stage agentic pipeline for Z-Image prompt generation
1. Classification (shot framing + photography genre)
2. Metadata extraction (dimensions, aspect ratio)
3. Attribute mapping (rule-based schema selection)
4. Detailed analysis (LLM structured extraction)
5. Prompt composition (narrative generation)
6. Z-Image recommendations (resize, settings)
- Externalized configuration in YAML:
- 11 shot framing types (ECU to VLS)
- 56 photography genres across 6 categories
- Complete attribute schemas for body, clothing, pose
- Content detail levels (minimal to explicit/NSFW)
- Prompt templates and focus overrides
- Utility module with:
- Config loading with caching
- Output cleaning (removes meta-tags, markdown)
- Image hashing for deterministic caching
- Z-Image recommendation generation
- Attribute schema builders
- Node features:
- 18 input parameters
- 4 outputs (prompt, structured_data, metadata, debug_log)
- Seed modes (fixed, randomize, increment, decrement)
- Prompt modes (analyze, enhance, priority, override)
- Focus area toggles for token budget allocation
- Detail levels (Quick, Standard, Deep)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
