Drone-style aerial shot with Seedance 2.5: 12 seconds, $3.22 at 480p
A 12-second aerial move on Sume's seedance-2.5 costs $3.22 at 480p. A prompt structure for altitude and heading, plus a reference-video option.

A 12-second drone-style aerial move on seedance-2.5 costs about $3.22 at 480p on Sume, and roughly $6.93 at 720p. Prompt the camera like a flight plan, one path per clip, and test at 480p before paying for a higher tier.
ByteDance lists professional camera movement among Seedance 2.5's strengths (read 2026-10-07, Seed). Sume's docs describe the id as a text, image and reference-to-video model with 4 to 30 second clips (video router docs).
A prompt structure for a flight path
Write the shot in the order a pilot would: starting altitude, the move, the heading, and the end point. One move per clip beats several, since a 12-second shot with three changes of direction tends to lose its subject.
- Start: "High angle, 120 m above a coastal road at sunrise."
- Move: "Slow descent while flying forward along the road."
- End: "Settle low over the lighthouse, horizon level."
- Keep: "No cuts, steady speed, no text."
Price
Seedance 2.5 is billed by output tokens on Sume (width x height x seconds x 24 / 1024), at $0.0214 per 1,000 tokens for 480p and 720p and $0.0234 for 1080p, times 1.25. That gives about $1.07 for 4 seconds at 480p, $8.06 for 30 seconds at 480p, and $42.65 for 30 seconds at 1080p. Twelve seconds at 480p works out to about $3.22.
| Clip | Resolution | Approx. price |
|---|---|---|
| 4 seconds | 480p | about $1.07 |
| 12 seconds | 480p | about $3.22 |
| 12 seconds | 720p | about $6.93 |
| 30 seconds | 480p | about $8.06 |
Copy a move from a real clip
If you have real aerial footage, send it as a video_url entry in input_references. Seedance accepts video references, and ByteDance says the model captures the framing and cinematic language of a reference, not only the motion (Seed, read 2026-10-07). Add a still of the location as an image reference to keep the look.
Caveats
The result is a generated scene, not a survey. Do not use it to represent a real property or location in a way that implies it was filmed there. If you need real aerial footage, hire a pilot.
Practical notes
Check the horizon and the scale of objects in every take. Generated aerial shots often get proportions wrong, such as cars too large for the road. If the clip is for an explainer or a mood piece this is fine; for a real-estate or construction use it is a reason to use real footage.
To lower the cost of iteration, test the path at 4 seconds and 480p for about $1.07, and only extend to 12 seconds once the first four look right.
Sources
Related posts
More in Models
- Anime-style ad with Seedance 2.5: use a character sheet as references
Send your mascot's character sheet as input_references to seedance-2.5 on Sume and test a 6-second anime-style ad at 480p before you pay for 720p.
- Can Gemini Omni Flash speak my script? It takes no audio input on Sume
On Sume, Gemini Omni Flash 1.1 takes image and video references but no audio reference. It makes its own sound. For your script in a face, use avatar video.
- Same character in three AI video shots: Omni reference image
Keep one character across separate Gemini Omni clips by sending the same reference image with every request on Sume. Request shape, prompt tags, 3-shot cost.
- ChatGPT Image 2 vs 2.5 at medium quality, 16:9: price on Sume
At 16:9 and medium quality, ChatGPT Image 2 bills $0.0525 per image on Sume and ChatGPT Image 2.5 bills $0.0105: five times less. Table, batch math and a swap.
Written by Sume