API Hop

Seedance 2.5 multimodal reference video API

Create Seedance 2.5 video from image or video references, with optional audio references, on API Hop. At least one image or video is required.

Seedance 2.5 online multimodal reference tool

Model settings

Sign in to run with your account balance.

Pricing and billing

Seedance 2.5 requests and examples

Recorded with Seedance 2.5 on 2026-10-22. One sample, not a quality or latency guarantee.

View prompt

One continuous four-second studio shot of a clear glass with orange citrus drink and ice on a gray stone pedestal, lemon wedge and green leaf beside it, turquoise background. Slow subtle clockwise camera orbit; keep the entire glass visible and preserve its shape and drink color. Tiny bubbles rise. No cuts, new objects, hands, text, logos, music or speech. Silent video.

{
  "model": "seedance-2.5",
  "prompt": "One continuous four-second studio shot of a clear glass with orange citrus drink and ice on a gray stone pedestal, lemon wedge and green leaf beside it, turquoise background. Slow subtle clockwise camera orbit; keep the entire glass visible and preserve its shape and drink color. Tiny bubbles rise. No cuts, new objects, hands, text, logos, music or speech. Silent video.",
  "duration": 4,
  "resolution": "480p",
  "aspectRatio": "16:9",
  "generateAudio": false,
  "metadata": {
    "content": []
  }
}
Public evidence record

Recorded with Seedance 2.5 on 2026-10-22. One sample, not a quality or latency guarantee.

View prompt

One continuous four-second product shot between the supplied frames. Keep the glass, ice, pedestal, lemon wedge, green leaf, turquoise background and camera position stable. Pour orange drink from above, raise its level to the final frame, and stop pouring. No cuts, new objects, text, logos, music or speech. Silent video.

{
  "model": "seedance-2.5",
  "prompt": "One continuous four-second product shot between the supplied frames. Keep the glass, ice, pedestal, lemon wedge, green leaf, turquoise background and camera position stable. Pour orange drink from above, raise its level to the final frame, and stop pouring. No cuts, new objects, text, logos, music or speech. Silent video.",
  "duration": 4,
  "resolution": "480p",
  "aspectRatio": "adaptive",
  "generateAudio": false,
  "metadata": {
    "content": [
      {
        "type": "image_url",
        "role": "first_frame",
        "image_url": {
          "url": "https://api.apihop.ai/generated/364459846248173568/0.png"
        }
      },
      {
        "type": "image_url",
        "role": "last_frame",
        "image_url": {
          "url": "https://api.apihop.ai/generated/364460494205222912/0.png"
        }
      }
    ]
  }
}
Public evidence record

Recorded with Seedance 2.5 on 2026-10-22. One sample, not a quality or latency guarantee.

View prompt

Use the image for product identity and the video only for slow clockwise camera motion. One continuous four-second studio shot of a clear glass with orange citrus drink and ice on a gray stone pedestal, lemon wedge and green leaf beside it, turquoise background. Slow subtle clockwise camera orbit; keep the entire glass visible and preserve its shape and drink color. Tiny bubbles rise. No cuts, new objects, hands, text, logos, music or speech. Silent video.

{
  "model": "seedance-2.5",
  "prompt": "Use the image for product identity and the video only for slow clockwise camera motion. One continuous four-second studio shot of a clear glass with orange citrus drink and ice on a gray stone pedestal, lemon wedge and green leaf beside it, turquoise background. Slow subtle clockwise camera orbit; keep the entire glass visible and preserve its shape and drink color. Tiny bubbles rise. No cuts, new objects, hands, text, logos, music or speech. Silent video.",
  "duration": 4,
  "resolution": "480p",
  "aspectRatio": "16:9",
  "generateAudio": false,
  "metadata": {
    "content": [
      {
        "type": "image_url",
        "role": "reference_image",
        "image_url": {
          "url": "https://api.apihop.ai/generated/364460494205222912/0.png"
        }
      },
      {
        "type": "video_url",
        "role": "reference_video",
        "video_url": {
          "url": "https://api.apihop.ai/examples/seedance-2-0/2026-10-03/364460440891424768-text-to-video.mp4"
        }
      }
    ]
  }
}
Public evidence record

Recorded with Seedance 2.5 on 2026-10-22. One sample, not a quality or latency guarantee.

View prompt

Animate the supplied first image. One continuous four-second studio shot of a clear glass with orange citrus drink and ice on a gray stone pedestal, lemon wedge and green leaf beside it, turquoise background. Slow subtle clockwise camera orbit; keep the entire glass visible and preserve its shape and drink color. Tiny bubbles rise. No cuts, new objects, hands, text, logos, music or speech. Silent video.

{
  "model": "seedance-2.5",
  "prompt": "Animate the supplied first image. One continuous four-second studio shot of a clear glass with orange citrus drink and ice on a gray stone pedestal, lemon wedge and green leaf beside it, turquoise background. Slow subtle clockwise camera orbit; keep the entire glass visible and preserve its shape and drink color. Tiny bubbles rise. No cuts, new objects, hands, text, logos, music or speech. Silent video.",
  "duration": 4,
  "resolution": "480p",
  "aspectRatio": "adaptive",
  "generateAudio": false,
  "metadata": {
    "content": [
      {
        "type": "image_url",
        "role": "first_frame",
        "image_url": {
          "url": "https://api.apihop.ai/generated/364459846248173568/0.png"
        }
      }
    ]
  }
}
Public evidence record

Seedance 2.5 API pricing and billing

480p · With video reference

Official price

Completion tokens
$6.0000 /M

API Hop price Save 2%

Completion tokens
$5.8800 /M

480p · Without video reference

Official price

Completion tokens
$10.0000 /M

API Hop price Save 2%

Completion tokens
$9.8000 /M

720p · With video reference

Official price

Completion tokens
$6.0000 /M

API Hop price Save 2%

Completion tokens
$5.8800 /M

720p · Without video reference

Official price

Completion tokens
$10.0000 /M

API Hop price Save 2%

Completion tokens
$9.8000 /M

1080p · With video reference

Official price

Completion tokens
$6.5714 /M

API Hop price Save 2%

Completion tokens
$6.4400 /M

1080p · Without video reference

Official price

Completion tokens
$11.0000 /M

API Hop price Save 2%

Completion tokens
$10.7800 /M

Reference price - not independently verified. Check the publisher's current rates, billing units and applicable terms before comparing costs.

Seedance 2.5 multimodal references

Create Seedance 2.5 video from image or video references, with optional audio references, on API Hop. At least one image or video is required.

Multimodal references: inputs and settings

Arrange references around the decisions they should guide: subject appearance, setting, camera language or sound. A larger allowance is useful for distinct contributions, not a reason to submit conflicting examples.

Images for appearance, video for movement

Use reference_image inputs to establish appearance and reference_video inputs to demonstrate movement or framing. Explain which subject features should transfer and which background or action details should be ignored, then describe the new sequence in the top-level prompt.

Audio references and generated audio are different controls

Use reference_audio to guide sound while keeping generated audio as a separate output choice. Audio references require at least one image or video in the same request. Specify where sound belongs in the scene rather than assuming it determines the entire timeline.

Keep within the 30 / 10 / 10 media limits

The request accepts at most 30 images, 10 videos and 10 audio files, with 50 media items in total. Shorthand and typed content share these limits. Use public HTTP(S) URLs or existing asset:// IDs; this form does not upload local files.

Choose 4-30 integer seconds or -1 automatic duration, with 5 seconds as the default. Output supports 480p, 720p or 1080p and the listed fixed ratios or adaptive. Reference-video length is not itself the output-duration setting.

Joint audiovisual generation and output review

Seedance 2.5 can generate sound with the image sequence. Describe ambience, effects and dialogue in event order, then inspect audiovisual transitions across the clip. Generated audio is an output switch; reference audio is a media input whose availability depends on the mode.

Multimodal references: reviewing the output

Evaluate with the duration, input mode and sound settings intended for your actual clip. Pay particular attention to event order and continuity in longer sequences; demonstrations do not establish guaranteed quality, latency or success rates.

Reference contribution and subject consistency

Check each reference against its assigned contribution. Confirm that a camera example has not replaced the intended subject or background, and that appearance remains consistent through the longer sequence.

Conflicting inputs and sound alignment

Compare action timing, camera direction and sound across the full output. Remove conflicting assets or clarify their responsibilities when references pull the scene in different directions; adding more inputs is not always the remedy.

Explain how the references work together

Review the reference roles and event order together. A useful result borrows the requested characteristics without copying unwanted content or drifting away from the written brief.

Use text-to-video for an idea without media. Use first and last frames when an image must define an actual boundary composition, rather than merely influence the appearance of the new clip.

Record total duration, rework and actual cost

Use a consistent task set to record submission-to-result time, revision count and account charges. Distinguish waiting, task completion and human acceptance. Establish an acceptable quality and budget before increasing volume, rather than treating one wait as a stable service-speed estimate.

API request and job tracking

Submit the current mode's request to POST /v1/video/seedance, then query GET /v1/jobs/{jobId} with the returned job ID. Wait for COMPLETED with output before using the result; checking an existing task does not require submitting another generation.

Seedance 2.5 multimodal references FAQ

Does the reference video determine the output duration?

No. Set output duration to an integer from 4 to 30 seconds or choose -1 for automatic duration. A reference video provides media guidance; its length is not the output-duration setting or a promise to reproduce the complete source clip.

Is any existing asset:// ID usable?

It must refer to media already available to the provider account used for the request. This page does not create or upload those assets. A syntactically valid ID alone does not prove that the selected channel can access the media.

Can Seedance 2.5 use an audio reference by itself?

No. This mode requires at least one image or video reference. Audio is an optional additional input, while generateAudio separately controls generated sound in the output. Supplying audio does not replace the visual-input requirement.

How do I tell Seedance 2.5 what each reference should contribute?

Specify the subject, desired motion and the role of each input in one brief. Use the ordered image, video and audio references to distinguish assets, and describe what to preserve or borrow from each. An asset ID is a media address, not an instruction about how to use it.

Do the four Seedance 2.5 pages use different model parameters?

No. All four input modes use model: seedance-2.5, submit an asynchronous video task to POST /v1/video/seedance, and query GET /v1/jobs/{jobId}. Do not substitute a page mode, display name or another model ID for the API model value.

Which durations and output settings does Seedance 2.5 support?

The current interface accepts integer durations from 4 to 30 seconds or -1 for automatic duration, defaulting to 5. Resolutions are 480p, 720p and 1080p, not 4k. Ratios are 16:9, 9:16, 1:1, 4:3, 3:4, 21:9 and adaptive; first/last-frame inputs require adaptive. Generated audio is separate and output is MP4 only.

How should I troubleshoot authentication, balance or generation errors?

Check Bearer authentication, balance, seedance-2.5 model access and the required inputs for the selected mode, then verify media accessibility. Retain the full error and job ID, and distinguish submission failure, a pending job and task failure instead of inferring the cause from one numeric code.

How can I confirm the cost of Seedance 2.5 generation?

Check pricing for the resolution, duration and presence of video input in this page's pricing section, then confirm actual charges against account records. An unavailable price does not mean free generation; do not apply another version's rates or text-model pricing to this video task.

Other video modes

Related video APIs