whiteboard-infographic-video-renderer

Whiteboard Infographic Video Renderer

Overview

Use this skill to turn a calibrated whiteboard board package into a reviewable video output: AI narration, subtitles, an editable HyperFrames project, a preview MP4, and synchronization keyframes. The renderer does not create or recalibrate the board image; it consumes the control layer produced upstream.

Workflow

1. Read references/contracts.md before changing paths, JSON fields, or timing behavior.
2. Confirm the project package has script/voiceover_segments.json, a board_manifest.json, a motion_plan.json, and a consumable board visual asset. Single-board packages usually use board.png; multi-board packages may use per-board board.png/asset.localPath or asset.kind=url.
3. Run the bundled renderer script from this skill directory:

node scripts/render_whiteboard_project.mjs --project-dir /path/to/project-output --quality standard

For faster iteration:

node scripts/render_whiteboard_project.mjs --project-dir /path/to/project-output --quality draft

1. Check video/renderer_report.json, audio/voiceover_timing.json, sync/action_timing.json, sync/camera_plan.json, sync/action_camera_qa_report.md, video/preview.mp4, video/hyperframes/, and video/keyframes/keyframe_manifest.json.
2. Treat validate and inspect failures as blocking. lint warnings may be reported if there are no lint errors and the generated video artifacts are complete.

Multi-Board Workflow

Use the multi-board entry when upstream D produces board_index.json and combined_motion_plan.json:

node scripts/render_multi_board_project.mjs \
  --project-dir /path/to/project-output \
  --board-root /path/to/board \
  --voiceover /path/to/voiceover_segments.json \
  --quality standard

Rules:

• Read board_index.json and combined_motion_plan.json.
• Treat combined_motion_plan.json as the full-video timeline.
• Preserve each board directory's motion_plan.json as a local control package; do not interpret local start=0 values as full-video time.
• Switch the visual board by segment.boardId.
• Use each board's consumable visual asset as the visual layer: local board.png/asset.localPath, or asset.kind=url for cloud/object-storage model outputs.
• Use board_manifest.json, annotation_manifest.json, and combined_motion_plan.json as the control layer.
• Add renderer-level action rhythm and camera strategy only after measured audio timing exists; do not push these fields back into C's semantic plan.