Add Motion

Animates a still image into a short video clip using Kling 3.0 Pro.

This skill covers: product shots, packshots, clothing shoot outputs, styled images, or any still you want to bring to life.

This skill does not cover:

• Voiceovers or talking-head content — use /ugc or /product-ugc for that
• Long-form video — clips are 3–12 seconds

Output saved to ./brands/[brand-name]/motion/[output-name]/.

Step 1 — Select brand

Scan ./brands/ for subfolders that contain a brand-identity/visual-guidelines.md file.

• One brand found: use it automatically and confirm: "Using brand: [name]"
• Multiple brands found: list them and ask which to use
• None found: tell the user to run /brand first

Step 2 — Select the source image

Ask: "Which image would you like to add motion to?"

If it's a previously generated image, ask them to point you to it. Suggest checking these folders:

brands/[brand-name]/product-shoot/
brands/[brand-name]/packshots/
brands/[brand-name]/styled/
brands/[brand-name]/clothing-shoot/
brands/[brand-name]/ugc/

If it's an external image they want to upload, create the uploads folder and tell them where to put it:

mkdir -p brands/[brand-name]/motion/uploads

"Drop your image into brands/[brand-name]/motion/uploads/ and let me know the filename."

Wait for confirmation before continuing.

Step 3 — Describe the motion

Ask: "Describe the motion you'd like added. Be specific — direction, speed, camera movement, subject behaviour. Examples:

• 'Slow upward drift with a subtle parallax depth effect'
• 'The product rotates 90 degrees clockwise, slow and smooth'
• 'Camera slowly pulls back from the subject, slight lens breathing'"

Realism patterns (selfie / UGC / handheld POV motion only)

When the motion is selfie-style, UGC, handheld, or POV — not a clean studio rotation or product packshot drift — read skills/references/shlabu-realism.md and apply its principles when composing the final motion_description written into the spec.

Key levers for Kling 3.0 (full reference for context):

• Anti-synthetic vocabulary repeated 4–6×: handheld, organic micro-shake, loose, unhurried, not performing, just messing around on a phone
• High-frequency detail callouts: visible pores, scattered freckles, baby hairs, individual strands catching the light
• Camera as choreography written as verb beats: "arm's length → drifts closer → tightens to cheek-macro → holds → pulls back" — not shot labels
• Background described as deliberately soft / blurred so Kling doesn't waste detail there

Skip these patterns for clean studio packshot or clinical product motion — they pull the render in the wrong direction.

Step 4 — Start frame and end frame

First time running this skill, explain before asking:

"Kling animates from a start frame to an end frame.

>

- Start frame only — Kling has full creative control over how the motion evolves. Good for natural, organic movement. - Start + end frame — you define exactly where the motion lands. Useful when you need a precise final state (e.g. a product landing in a specific position).

>

The source image you've selected will be the start frame. Do you also have an end frame you'd like to provide?"

If the user provides an end frame, ask where it's saved (same process as Step 2 — existing file or upload to motion/uploads/).

If no end frame, note it and continue.

Step 5 — Analyse image for products

Read the source image using the Read tool.

Determine whether the image contains a product (packshot, product-in-hand, product on a surface, product being held, etc.). If the image is clearly people-only with no product visible (e.g. someone applying cream with no product in frame), skip this step.

If a product is present:

1. Confirm with the user: "I can see what looks like [product description] in the image. Is this correct? And which product is it — I'll look it up in your brand's product images."

1. Once confirmed, locate the product in brands/[brand-name]/brand-identity/product-images/. If the file is there, use it. If not, check brands/[brand-name]/brand-identity/products.json for the product URL, then try to scrape additional images from the product page:

   # Try fetching the product page directly via WebFetch
   # Look for multiple product image URLs in the HTML/JSON
   # Download them to brands/[brand-name]/brand-identity/product-images/

1. Aim for at least 2 product images from different angles. If only one is available:

"I can only find one image of this product. Do you have another angle you can add to brands/[brand-name]/brand-identity/product-images/? More angles help Kling maintain accurate product rendering throughout the animation."

Wait for them to add more, or proceed with one if they can't provide another.

1. These product images become Element1, Element2, ... — they're passed to Kling so it can maintain product accuracy through the motion.

Step 6 — Duration and audio

Ask in one message:

Duration: "How long should the clip be? (3–12 seconds)"

Audio: "Would you like Kling to generate ambient audio for the clip? Note: this is environmental/atmospheric sound — not voiceover. (+$0.056/sec)"

Step 7 — Name the output

Derive a slug from the source image name and a short motion descriptor:

• hey-bud-tint-rotate
• ren-jacket-parallax
• mutimer-hero-drift

Ask the user to confirm or rename.

Step 8 — Write add-motion-spec.json

Create the output folder and write the spec:

brands/[brand-name]/motion/[output-name]/
  add-motion-spec.json

With end frame and product images:

{
  "output_name": "hey-bud-tint-rotate",
  "brand": "[brand-name]",
  "source_image": "brands/[brand-name]/product-shoot/hey-bud-tint-hand-hold/product-shot_v1.png",
  "end_image": "brands/[brand-name]/motion/uploads/hey-bud-end-frame.jpg",
  "motion_description": "The product rotates 90 degrees clockwise, slow and smooth, light catching the label.",
  "has_product": true,
  "product_name": "Hey Bud Acne Face Tint",
  "product_images": [
    "brands/[brand-name]/brand-identity/product-images/hey-bud-face-tint.jpg",
    "brands/[brand-name]/brand-identity/product-images/hey-bud-face-tint-side.jpg"
  ],
  "duration": 5,
  "generate_audio": false,
  "aspect_ratio": "9:16"
}

Without end frame or product:

{
  "output_name": "ren-jacket-parallax",
  "brand": "[brand-name]",
  "source_image": "brands/[brand-name]/clothing-shoot/ren-garage/shoot_v1.png",
  "end_image": null,
  "motion_description": "Slow camera pull-back with subtle parallax depth. Slight lens breathing.",
  "has_product": false,
  "product_name": null,
  "product_images": [],
  "duration": 6,
  "generate_audio": true,
  "aspect_ratio": "4:5"
}

Aspect ratio — detect from the source image dimensions and match automatically. Default to 9:16 if unsure. Ask the user if the image is ambiguous.

Step 9 — Generate

Run from the project root:

python3 skills/references/generate-motion.py brands/[brand-name]/motion/[output-name]

The script:

• Uploads the start frame (and end frame if provided) to FAL
• Uploads product images as Element1, Element2, ... and appends @Element references to the prompt
• Calls Kling 3.0 Pro with all inputs
• Downloads the output video

Cost: duration × $0.112 (audio off) or duration × $0.168 (audio on)

Step 10 — Present the result

Tell the user the exact path where the file was saved, e.g.: brands/[brand-name]/motion/[output-name]/[output-name]_v1.mp4

Ask: "Happy with this? Or would you like to regenerate, adjust the motion description, or add an end frame?"

To regenerate (same spec):

python3 skills/references/generate-motion.py brands/[brand-name]/motion/[output-name]

To adjust motion or add an end frame: update add-motion-spec.json and re-run.

Notes

• Kling 3.0 Pro accepts durations 3–15s — this skill caps the suggestion at 12s for cost and practical reasons
• The elements array gives Kling reference images for maintaining product accuracy. More angles = better fidelity through rotation and movement
• Locking a person's face across motion: if the source image contains a model generated via /model, pass brands/[brand-name]/models/[model-name]/character-sheet.png as an extra entry in product_images (rename the field mentally — Kling's element system accepts any subject reference). The 4-panel grid feeds Kling's "Bind elements to enhance consistency" feature so the face stays locked through the motion. Equivalent to what kling.ai's web UI does when you upload reference images to a bound element.
• End frames give you precise control over the motion arc. Without one, Kling interprets the motion description creatively
• Audio generated by Kling is ambient/environmental — it is not voiceover. For talking-head audio, use /ugc
• Each brand has its own motion/ subfolder — multiple brands can run independent motion workflows in the same project