ai-video-concept-art

name	ai-video-concept-art
description	Write one concept image request and local concept manifest.

AI Video Concept Art

Artist Stance

Act as a concept artist preparing one image request for a short video: preserve the director's intent, make the selected subject visually specific, and make the requested image useful for later keyframe generation.

Workflow

1. Read one target concept item from preproduction/generation_plan.yaml under concepts. It must include concept_id and reason.
2. Read only the matching source record from preproduction/concepts.yaml, including its type, plus preproduction/visual_style.yaml, relevant preproduction/shots.yaml usages, and the concept reason.
3. Decide the useful presentation for this one concept. If multiple views or detail insets are useful, request them as panels within one image.
4. Write art/<type>s/<concept_id>/image_request.yaml.
5. Write art/<type>s/<concept_id>/manifest.yaml.

Required Outputs

Create or update these outputs for this one concept:

• art/<type>s/<concept_id>/image_request.yaml
• art/<type>s/<concept_id>/manifest.yaml

Resources

Use this skill's bundled template at skills/ai-video-concept-art/assets/templates/image_request.yaml for the image request.

Use this skill's bundled template at skills/ai-video-concept-art/assets/templates/concept_manifest.yaml as the structure reference for manifest.yaml.

The image request must set output_path to art/<type>s/<concept_id>/concept_v001.png.

Validate with this skill's bundled validator at skills/ai-video-concept-art/scripts/validate_outputs.py, passing --project <project> --concept-id <concept_id>. Fix any reported issue before returning.