Seedance 2.0 API Guide: Cinematic Text-to-Video

If you are wiring AI video into a product in 2026 — ad creative, product demos, social loops, animated stills — the first integration question is rarely "which model looks best." It is "how do I call it, what does it cost per clip, and how do I keep a long-running render from blocking my request." This guide walks through the Seedance 2.0 API on hiapi end to end: the task lifecycle, every input parameter, two real generated clips with their exact prompts and task IDs, the per-second pricing math, and the image-to-video path.

Seedance 2.0 is ByteDance's video model. The thing that makes it a sensible default rather than a niche pick is that one model handles both text-to-video and image-to-video, generates native audio, and renders 4–15 second clips at 480p, 720p, or 1080p. You do not maintain two integrations for two modalities — same endpoint, same request shape, you just add a first-frame image when you have one.

The task lifecycle: create, poll, download

Video generation is slow by nature — a few seconds of 1080p footage takes real compute time. hiapi exposes every model, Seedance included, through one async task API so a render never holds a connection open. The pattern is always the same three steps:

1. Create a task with POST /v1/tasks → you get a taskId back immediately.
2. Poll GET /v1/tasks/{taskId} until status is success (or fail).
3. Download the clip from output[0].url — and download it promptly, because the URL is a time-limited link.

# 1. Create the task
curl -X POST https://api.hiapi.ai/v1/tasks \
  -H "Authorization: Bearer $HIAPI_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "seedance-2.0",
    "input": {
      "prompt": "A slow cinematic dolly-in on a sleek matte-black smartwatch on a concrete pedestal, soft studio light",
      "aspect_ratio": "16:9",
      "duration": 5,
      "resolution": "720p",
      "generate_audio": false
    }
  }'

# 2. Poll until status = success
curl https://api.hiapi.ai/v1/tasks/$TASK_ID \
  -H "Authorization: Bearer $HIAPI_TOKEN"

For anything past a prototype, do not sit in a poll loop. Pass a callback object and let hiapi push you the terminal state instead:

{
  "model": "seedance-2.0",
  "input": { "prompt": "...", "aspect_ratio": "16:9", "duration": 5, "resolution": "720p" },
  "callback": { "url": "https://your-domain.com/hiapi/callback", "when": "final" }
}

Input parameters

The two parameters that matter most for your bill are resolution and duration — pricing is per second and scales hard with resolution (see below). Keep both low while you are still iterating on the prompt.

Worked example 1: a product demo (text-to-video)

This is a 5-second 720p clip, no audio, generated from a single prompt — the kind of shot you would drop into a landing page or an ad.

VIDEO

A slow cinematic dolly-in on a sleek matte-black smartwatch resting on a
brushed-concrete pedestal, soft diffused studio light, the screen wakes to
show a glowing minimalist watch face, faint reflections move across the metal
bezel, premium tech commercial look, calm and clean.

• Model: seedance-2.0 · aspect_ratio: 16:9 · duration: 5 · resolution: 720p · generate_audio: false
• Task ID: tk-hiapi-01KVQFE96NADP6XSQA3PS14Z5N
• Current estimated cost: 5 s × $0.293/s (720p) = $1.47

The camera move and the soft "wake" of the watch face came straight from the prompt — no keyframing, no post. For product shots, lead with the camera verb ("dolly-in", "slow pan", "push-in") and the lighting; Seedance follows motion direction well.

Worked example 2: an ambiance scene (text-to-video)

Same shape, different intent — a mood/establishing shot for a hero header or a background loop.

VIDEO

A gentle slow pan across a cozy specialty coffee shop interior at golden hour,
warm light through the window, steam rising from a cup on the counter, a few
patrons softly out of focus, cinematic, smooth camera motion.

• Model: seedance-2.0 · aspect_ratio: 16:9 · duration: 5 · resolution: 720p
• Task ID: tk-hiapi-01KVQFPHHSCV4P7DM585VCDP36
• Current estimated cost: 5 s × $0.293/s (720p) = $1.47

Image-to-video: pin the first frame

When the opening frame has to match an asset you already own — a product render, an approved key visual, a character sheet — switch from text-to-video to image-to-video by adding first_frame_url. Everything else about the call is identical; you are just giving the model a starting frame to animate instead of generating one from scratch.

{
  "model": "seedance-2.0",
  "input": {
    "prompt": "Animate the product with a slow cinematic push-in, subtle light sweep across the surface",
    "first_frame_url": "https://static.hiapi.ai/your-asset/first.jpg",
    "aspect_ratio": "16:9",
    "duration": 5,
    "resolution": "720p"
  }
}

Use text-to-video when you are exploring an idea and the exact subject is flexible. Use image-to-video when brand consistency is the whole point and the first frame is non-negotiable.

Pricing: per second, by resolution

Seedance is billed per second of output, and the per-second rate is set by resolution. These are the live rates on hiapi:

Two practical consequences:

• Iterate at 480p. At $0.136/s, a 4-second test clip is $0.54. Lock the prompt and camera move at 480p, then re-render the winner at the resolution you actually ship. Burning 1080p renders on prompt exploration is the most common way to overspend.
• Divide cost by your acceptance rate. If you keep 1 clip in 4, a $3.645 1080p render really costs ~$14.58 per usable clip. For high-volume work, that gap is the whole budget conversation — and it is why the iterate-at-480p habit pays for itself.

Where Seedance fits

Seedance 2.0 is the premium, do-everything pick: cinematic motion, native audio, and both modalities behind one integration. It is not the cheapest path to HD video on the platform — if you are rendering hundreds of variants, a lower-cost model is the right call for the variant pass. For how Seedance compares to the other video models on cost and capability, see Best AI Video Generation API in 2026.

When you are ready to ship: run your own prompts on the Seedance 2.0 model page, then move to the API with the task lifecycle above. Start at 480p, confirm the prompt, scale the resolution, and wire callbacks before you go to production.