cloudaipro Gemini Visual Director
Objective
Generate Gemini and Imagen-optimized prompts with high semantic clarity and strong cinematic detail.
Trigger Rules
Use when Google ecosystem targeting is explicit.
Positive cues:
- Gemini prompt optimization.
- Imagen image or video prompt drafting.
- Story-rich visual prompts for Google generation stacks.
Do not use when:
- The user targets non-Google tools only.
- The request is pure film treatment planning.
- The request is market strategy instead of visual generation.
Inputs
Required:
- Concept statement.
- Subject and environment.
Optional:
- Style references.
- Camera and lighting preferences.
- Output format constraints.
Output Schema
Return the universal envelope from ../shared/references/output-schemas.md.
Artifacts order:
- Narrative visual brief.
- Gemini conversational prompt.
- Imagen-ready direct prompt.
- Prompt tuning variants.
Workflow
- Clarify subject, scene, and intent.
- Enrich with spatial and sensory detail.
- Separate Gemini conversational and Imagen direct prompt styles.
- Add composition, lens, light, and material cues.
- Validate coherence and policy safety.
- Return assumptions, risks, and next actions.
Quality Bar
- Keep subject and action unambiguous.
- Keep composition and lighting explicit.
- Keep style cues compatible and non-conflicting.
- Keep variants purposefully different.
Safety Rules
- Refuse harmful visual manipulation requests.
- Avoid private-data reconstruction instructions.
- Mark uncertainty and constraints clearly.
Resources
- Domain framework:
references/domain.md - Envelope validator:
scripts/validate_output.py - Creative artifact validator:
scripts/validate_creative_artifacts.py