← idea.aiCONTENT HISTORY
WHAT CHANGED
Update to idea.ai
Snapshot Nov 16, 2026 · 18:17 UTC · version 0.4.0
Collection source: downloaded plugin package.
First saved snapshot
No earlier snapshot is available to establish a change.
Save idea.ai and check future changes in Following.
Up to 3 plugins free · no card required.
Compare saved observations
Download comparison JSONFull technical diff · 0 changed fields
Full snapshot data
{
"description": "Animate a supplied character or subject using motion from a reference video through idea.ai motion-control models.",
"included_files": [],
"name": "idea-ai-motion-transfer",
"skill_md_contents": "---\nname: idea-ai-motion-transfer\ndescription: \"Animate a supplied character or subject using motion from a reference video through idea.ai motion-control models.\"\n---\n\n# Motion transfer\n\nRead [the shared MCP workflow](../../references/mcp-workflow.md) before calling tools. It defines discovery, file handling, cost authorization, waiting, and delivery for this skill.\n\n## MCP operation\n\nUse free `upload_media` with both motion input slots when the files are on the user’s device, resolve ready uploads and measured metadata, then quote and use `create_video`. Uploading the references does not start animation.\n\n## Establish the two references\n\nIdentify the character/subject image and the motion video. Clarify whether the user wants the clip's movement, camera behavior, and original sound preserved; those are separate choices. Inspect the image and clip when available to describe what can be transferred and flag mismatched framing or visibility.\n\n## Validate the model inputs\n\nInspect Kling V3 Motion Control (`kling_v3_motion_control`) or another available motion-control model. Its inspected schema requires paired `motionImage` and `motionVideo` roles, with a 3–10-second motion video. Validate current duration, format, image dimensions, and size limits before scheduling a generation.\n\nUse the returned `character_orientation`, `keep_original_sound`, and `mode` choices where supported. Do not invent a skeleton, pose track, or character ID. Model controls may not isolate every aspect of motion from camera behavior.\n\nFor user-owned files, declare both required inputs in one authorized upload request. For assets already generated through the MCP, reuse their URLs with the correct roles.\n\n## Assess the result\n\nReview movement timing, body/subject proportions, identity, framing, unintended distortions, and the requested sound treatment. Do not claim frame-accurate motion capture or perfect identity preservation from a completed status alone.\n\nIf the reference clip is outside the model's limits, ask for a suitable excerpt or use separately authorized trimming tools that are actually available. Do not silently submit unsupported media or repeatedly retry it.\n\nDeliver the result with the source references and a concise account of any motion or identity differences that remain.\n"
}SHA-256 of public snapshot: 1bae444c3cc8ec289dae2f946f3e87838358e0339966a625f6c982f42964cc9b