Animate a trading or tarot card image: pick the model by aspect ratio

The short answer

Card art is tall, and the model choice depends on the ratio. In the Sume catalog, Wan 3.0 and MiniMax H3 accept 3:4 and 9:16, so a card keeps its shape. Kling 3 accepts only 16:9, 9:16 and 1:1, and Gemini Omni Flash only 16:9 and 9:16. A 5-second 3:4 card costs $0.3125 on Wan at 480p and $0.375 on H3 at 768p.

What ratio your art really has

A standard trading card is 2.5 by 3.5 inches, which is 5:7, and tarot decks vary, so neither is guaranteed to match a catalog ratio. Wan lists auto and adaptive, and H3 lists adaptive, which lets the model follow the picture. With an exact 3:4 you can also pass 3:4 directly. Anything else is cropped by a fixed ratio, so keep the card's border and title bar inside a safe margin.

Which models take which ratio

Ratios come from the catalog rows.

Request

The card art is the first_frame. The prompt should move the art, not the frame: a foil shimmer, drifting mist, a character blinking. Ask for a locked camera so the border stays put.

curl -X POST https://api.sume.com/v1/videos \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: card-001" \
  -d '{
    "model": "wan-3.0",
    "prompt": "Foil shimmer across the art, mist drifting, camera locked, card border static",
    "duration": 5,
    "resolution": "480p",
    "aspect_ratio": "3:4",
    "frame_images": [{
      "type": "image_url",
      "image_url": {"url": "https://example.com/card.png"},
      "frame_type": "first_frame"
    }]
  }'

Cost of a five second card

Wan at 480p is $0.0625 a second, so 5 seconds is $0.3125. H3 at 768p is $0.075 a second, which is $0.375. H3 always generates audio. If you need a silent loop for a store page, strip the track in your player.

When 9:16 is fine

If your card only has to be 9:16 for a phone, Kling 3 at $0.70 for 5 silent seconds is also open to you, and it adds a last frame so the card can end on the revealed face.