> ## Documentation Index
> Fetch the complete documentation index at: https://veniceai-mintlify-d2fddb8a.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Seedance 2.0

> Generate, edit, extend, and stitch videos with Seedance 2.0 on Venice — text, image, and reference-to-video workflows on one endpoint, with worked examples.

Seedance 2.0 is a flagship multimodal video model exposed on Venice as a family of three variants for text-, image-, and reference-driven video generation. The **reference-to-video** variant is unusually powerful: a single endpoint and a single model ID handle **four distinct workflows** (Reference, Edit, Extend, Stitch) — the workflow is inferred from the **shape of your prompt**.

This guide walks through the variants, the four workflows with their canonical prompts, the multimodal input limits, pricing, and complete `curl` examples.

## Variants

| Model ID                               | Variant  | Output resolutions  | Notes                                                                                                               |
| -------------------------------------- | -------- | ------------------- | ------------------------------------------------------------------------------------------------------------------- |
| `seedance-2-0-text-to-video`           | T2V      | 480p / 720p / 1080p | Text prompt only                                                                                                    |
| `seedance-2-0-image-to-video`          | I2V      | 480p / 720p / 1080p | First-frame (and optionally last-frame) image grounding                                                             |
| `seedance-2-0-reference-to-video`      | R2V      | 480p / 720p / 1080p | Up to 9 reference images + 3 reference videos + 3 reference audio donors. Powers Reference / Edit / Extend / Stitch |
| `seedance-2-0-fast-text-to-video`      | Fast T2V | 480p / 720p         | Faster, lower-fidelity tier                                                                                         |
| `seedance-2-0-fast-image-to-video`     | Fast I2V | 480p / 720p         | Faster, lower-fidelity tier                                                                                         |
| `seedance-2-0-fast-reference-to-video` | Fast R2V | 480p / 720p         | Faster, lower-fidelity tier; same workflow set                                                                      |

All variants are async. Submit via `POST /api/v1/video/queue`, then poll `POST /api/v1/video/retrieve` until the response body is `video/mp4`. See [Video Generation](/guides/media/video-generation) for the general queue flow.

## The "one model, four workflows" model

The reference-to-video variant (`seedance-2-0-reference-to-video` and its Fast sibling) is the same underlying model serving four different tasks. **The model infers the task from the prompt prefix and the shape of your inputs.** There is no `task` or `workflow` field — the prompt syntax is the routing.

| Workflow      | What it does                                                                                       | Prompt prefix                                                        | Inputs                                                                                            |
| ------------- | -------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------- |
| **Reference** | Generate a new video using uploaded reference files as donors for subject / motion / style / audio | `Refer to ... in <Image\|Video\|Audio N> to generate ...`            | Text + ≥1 image OR video reference (0-9 images, 0-3 videos), plus optionally up to 3 audio donors |
| **Edit**      | Modify a single input video while preserving the rest                                              | `Strictly edit <Video 1>, changing its ...`                          | 1 input video + text (images optional grounding)                                                  |
| **Extend**    | Forward / backward extension of one clip                                                           | `Extend <Video 1>, generate ...`                                     | 1 input video + text                                                                              |
| **Stitch**    | Stitch 2-3 clips with auto-generated transitions                                                   | `<Video 1> + <transition description> + followed by <Video 2> + ...` | 2-3 input videos + text                                                                           |

The **prompt syntax is canonical and case-sensitive**: angle brackets, capital first letter, single space before the number — `<Video 1>`, `<Image 1>`, `<Audio 1>`.

***

## Workflow patterns

### Reference workflow

Use the uploaded reference files as **donors** — subject, scene, motion, style, vocal timbre — to generate a brand-new video.

**Canonical prompt patterns**:

```
Refer to <Subject N> in <Image N> to generate ...
Refer to the [action | camera scene | style | sound effect] in <Video N> to generate ...
Refer to the [tone | timbre] in <Audio N> to generate ...
```

**Examples**:

* `Refer to <Subject 1> in <Image 1> to generate a 5-second clip of the same character riding a horse through snow.`
* `Refer to the camera scene in <Video 1> to generate a similar establishing shot of a futuristic city at dawn.`
* `Refer to <Subject 1> in <Image 1> and use the timbre in <Audio 1> for the narrator describing the scene.` (audio donors must be paired with at least one image or video reference — audio alone is rejected)

### Edit workflow

Modify a single input video. **Anything not explicitly named in the prompt is preserved.** Use this when you want a localized change (subject swap, weather/color change, element add/remove) rather than a wholly new video.

**Canonical prompt pattern**:

```
Strictly edit <Video 1>, changing its [original feature] to [new feature] ...
```

**Sub-patterns for finer control**:

```
Add Elements:
  At [timestamp / timing] and [spatial location] of <Video 1>, add [description of intended element].

Remove Elements:
  Remove [element to be deleted] from <Video 1>, keeping the rest of the video content unchanged.

Modify Elements:
  Replace [description of element to be changed] in <Video 1> with [description of intended element].
```

**Examples**:

* `Strictly edit <Video 1>, changing its weather from sunny to a heavy rainstorm.`
* `Add snacks such as fried chicken and pizza to the countertop in <Video 1>.`
* `Remove the red car from <Video 1>, keeping the rest of the video content unchanged.`
* `Replace the perfume featured in <Video 1> with the face cream from <Image 1>, with all original motions and camera work preserved.`

The last example combines Edit with an image reference — perfectly legal, the model uses `<Image 1>` as a visual donor for the replacement.

### Extend workflow

Continue a single clip forward or backward in time. **By default Seedance returns only the new content** — not the original input concatenated with the extension. This is by design, for transition continuity; if you want the input clip preserved alongside the extension, say so explicitly:

```
Extend <Video 1>, generate [description of extended content]
Extend <Video 1> backward, [description of extended content]
Extend <Video 1>, start with <Video 1>, then [description of extended content]      ← preserves input at start
Extend <Video 1> backward, [description], and then end with <Video 1>               ← preserves input at end
```

Transition handling: the model automatically extracts the transition frames for seamless blending, and the original segments of the input video are not re-generated.

**Examples**:

* `Extend <Video 1>, generate a dramatic chase scene through narrow alleys at dusk.`
* `Extend <Video 1> backward, the same character walking toward the camera before the original shot begins.`
* `Extend <Video 1>, start with <Video 1>, then the camera pulls back to reveal a vast landscape.`

### Stitch workflow (Track Completion)

Connect 2-3 input clips with AI-generated transitions. Total combined input duration must be ≤ 15 s.

**Canonical prompt pattern**:

```
<Video 1> + [transition description] + followed by <Video 2> [+ [transition description] + followed by <Video 3>]
```

**Examples**:

* `<Video 1> + a smooth seamless cut + followed by <Video 2>`
* `<Video 1>. The moment a leaf falls to the ground, it sets off a special effect of golden particles. A gust of wind blows by, leading into <Video 2>.`
* `<Video 1> + a wisp of smoke transforms into a flock of birds + followed by <Video 2> + a slow dolly-in + followed by <Video 3>`

The model auto-trims connecting segments at the join points for continuity.

***

## Universal prompt formula

Across all four workflows, the recommended authoring formula is:

```
Subject + Motion + Environment (Optional)
       + Camera Movement / Cut (Optional)
       + Aesthetic Description (Optional)
       + Audio (Optional)
```

* **Subject + Motion**: the logical foundation — define "Who" is performing "What action"
* **Environment + Aesthetics**: spatial background, lighting, visual style
* **Camera**: explicit shot type or movement
* **Audio**: ambient sound effects or vocal direction for immersive output

Layering this on top of a workflow prefix (e.g., `Strictly edit <Video 1>, changing its <subject + motion + environment + ...>`) produces the highest-quality outputs.

***

## Multimodal input limits

Values below are what the Venice API accepts. Requests outside these ranges are rejected at the schema layer with a 400 before reaching inference.

### Images

| Constraint                          | Value                                                               |
| ----------------------------------- | ------------------------------------------------------------------- |
| Input methods                       | URL (`http://`, `https://`) or Base64 data URL (`data:image/...`)   |
| Formats                             | `.jpeg`, `.png`, `.webp`, `.bmp`, `.tiff`, `.gif`, `.heic`, `.heif` |
| Aspect ratio (W / H)                | exclusive `(0.4, 2.5)`                                              |
| Minimum side                        | ≥ 300 px                                                            |
| Image count: I2V first-frame        | 1                                                                   |
| Image count: I2V first + last frame | 2                                                                   |
| **Image count: R2V (V2 / Fast)**    | **1 – 9**                                                           |

### Videos

| Constraint                  | Value                                                             |
| --------------------------- | ----------------------------------------------------------------- |
| Input methods               | URL (`http://`, `https://`) or Base64 data URL (`data:video/...`) |
| Formats                     | `.mp4`, `.mov`                                                    |
| Video codecs                | H.264 / AVC, H.265 / HEVC                                         |
| Audio codecs (in container) | AAC, MP3                                                          |
| **Duration per clip**       | `[2, 15]` s (inclusive)                                           |
| **Max clip count**          | 3 (R2V / Stitch / Extend)                                         |
| **Total combined duration** | ≤ 15 s across all clips                                           |
| **Per-clip size**           | ≤ 50 MB                                                           |

### Audio

| Constraint                  | Value                                                             |
| --------------------------- | ----------------------------------------------------------------- |
| Input methods               | URL (`http://`, `https://`) or Base64 data URL (`data:audio/...`) |
| Formats                     | `.wav`, `.mp3`                                                    |
| **Duration per clip**       | `[2, 15]` s                                                       |
| **Max clip count**          | 3                                                                 |
| **Total combined duration** | ≤ 15 s across all clips                                           |
| **Per-clip size**           | ≤ 15 MB                                                           |

Reference audio is supported on the R2V variants only. Each entry is forwarded to the model as a `role: "reference_audio"` content item that the prompt addresses as `<Audio 1>`, `<Audio 2>`, `<Audio 3>` — the model uses each clip for vocal timbre, sound effects, or background music depending on how the prompt frames it. The legacy singular `audio_url` field maps to the same content shape and is now equivalent to passing a one-element `reference_audio_urls`.

<Warning>
  **`reference_audio_urls` cannot be the only reference input.** The model requires at least one image or video reference alongside any audio donor. Pair `reference_audio_urls` with `reference_image_urls`, `reference_video_urls`, `image_url`, or `video_url` — audio-only submissions are rejected.
</Warning>

### Request size

The queue endpoint accepts JSON bodies up to **35 MB**. Inline data URLs for large videos can push past this — for multi-clip Stitch in particular, prefer URLs over inline base64.

***

## Pricing

Call `POST /api/v1/video/quote` to get a quote for a given request shape before submitting it to `/video/queue`. The quote endpoint is the only authoritative source; pricing details may change and shouldn't be cached or duplicated client-side.

When reference video(s) are part of the request, also pass `reference_video_total_duration` (the sum of all reference clip durations in seconds) so the quote matches what `/video/queue` will charge:

```bash theme={null}
curl -X POST https://api.venice.ai/api/v1/video/quote \
  -H "Authorization: Bearer $VENICE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "seedance-2-0-reference-to-video",
    "duration": "5s",
    "resolution": "1080p",
    "aspect_ratio": "16:9",
    "reference_video_total_duration": 5
  }'
```

***

## Complete examples

All examples assume `VENICE_API_KEY` is set in the environment.

### Text-to-video

```bash theme={null}
curl -X POST https://api.venice.ai/api/v1/video/queue \
  -H "Authorization: Bearer $VENICE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "seedance-2-0-text-to-video",
    "prompt": "A golden retriever frolicking through a sunlit meadow at sunset, slow camera dolly-in, shallow depth of field, warm cinematic lighting.",
    "duration": "5s",
    "aspect_ratio": "16:9",
    "resolution": "1080p"
  }'
```

### Image-to-video (first frame)

```bash theme={null}
curl -X POST https://api.venice.ai/api/v1/video/queue \
  -H "Authorization: Bearer $VENICE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "seedance-2-0-image-to-video",
    "prompt": "The lighthouse keeper turns toward the storm, lantern raised, waves crashing against the rocks.",
    "image_url": "https://example.com/lighthouse.jpg",
    "duration": "5s",
    "resolution": "720p"
  }'
```

<Note>
  `seedance-2-0-image-to-video` (and its Fast variant) **do not accept `aspect_ratio`** — the output aspect ratio is auto-derived from the input image's dimensions. Passing the field returns a 400 with *"This model does not support aspect\_ratio"*. Use the T2V or R2V variants if you need explicit aspect-ratio control.
</Note>

### Reference workflow — subject donor

```bash theme={null}
curl -X POST https://api.venice.ai/api/v1/video/queue \
  -H "Authorization: Bearer $VENICE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "seedance-2-0-reference-to-video",
    "prompt": "Refer to <Subject 1> in <Image 1> to generate a 5-second clip of the same character walking through a neon-lit Tokyo street at night.",
    "reference_image_urls": ["https://example.com/character.png"],
    "duration": "5s",
    "aspect_ratio": "9:16",
    "resolution": "1080p"
  }'
```

### Reference workflow — subject + audio donor

```bash theme={null}
curl -X POST https://api.venice.ai/api/v1/video/queue \
  -H "Authorization: Bearer $VENICE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "seedance-2-0-reference-to-video",
    "prompt": "Refer to <Subject 1> in <Image 1> to generate a 5-second clip of the same character walking through a neon-lit Tokyo street at night. Refer to the timbre in <Audio 1> for a soft female voiceover describing the scene.",
    "reference_image_urls": ["https://example.com/character.png"],
    "reference_audio_urls": ["https://example.com/voice-sample.mp3"],
    "duration": "5s",
    "aspect_ratio": "9:16",
    "resolution": "1080p"
  }'
```

### Edit workflow

```bash theme={null}
curl -X POST https://api.venice.ai/api/v1/video/queue \
  -H "Authorization: Bearer $VENICE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "seedance-2-0-reference-to-video",
    "prompt": "Strictly edit <Video 1>, changing its weather from sunny to a heavy rainstorm, with all original motions and camera work preserved.",
    "reference_video_urls": ["https://example.com/sunny-scene.mp4"],
    "reference_video_total_duration": 5,
    "duration": "5s",
    "aspect_ratio": "16:9",
    "resolution": "1080p"
  }'
```

### Edit workflow with image grounding

```bash theme={null}
curl -X POST https://api.venice.ai/api/v1/video/queue \
  -H "Authorization: Bearer $VENICE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "seedance-2-0-reference-to-video",
    "prompt": "Replace the perfume featured in <Video 1> with the face cream from <Image 1>, with all original motions and camera work preserved.",
    "reference_video_urls": ["https://example.com/perfume-ad.mp4"],
    "reference_image_urls": ["https://example.com/face-cream.png"],
    "reference_video_total_duration": 4,
    "duration": "5s",
    "aspect_ratio": "16:9",
    "resolution": "1080p"
  }'
```

### Extend forward

```bash theme={null}
curl -X POST https://api.venice.ai/api/v1/video/queue \
  -H "Authorization: Bearer $VENICE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "seedance-2-0-reference-to-video",
    "prompt": "Extend <Video 1>, generate a dramatic chase scene through narrow alleys at dusk, with neon signs flickering and rain on the pavement.",
    "reference_video_urls": ["https://example.com/alley-intro.mp4"],
    "reference_video_total_duration": 4,
    "duration": "5s",
    "aspect_ratio": "16:9",
    "resolution": "1080p"
  }'
```

### Stitch (3 clips)

```bash theme={null}
curl -X POST https://api.venice.ai/api/v1/video/queue \
  -H "Authorization: Bearer $VENICE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "seedance-2-0-reference-to-video",
    "prompt": "<Video 1> + a wisp of smoke transforms into a flock of birds + followed by <Video 2> + a slow dolly-in + followed by <Video 3>",
    "reference_video_urls": [
      "https://example.com/clip-1.mp4",
      "https://example.com/clip-2.mp4",
      "https://example.com/clip-3.mp4"
    ],
    "reference_video_total_duration": 12,
    "duration": "5s",
    "aspect_ratio": "16:9",
    "resolution": "1080p"
  }'
```

### Polling for completion

After every queue submission, save the returned `queue_id` and poll `/video/retrieve` until the response body is `video/mp4`:

```bash theme={null}
curl -X POST https://api.venice.ai/api/v1/video/retrieve \
  -H "Authorization: Bearer $VENICE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "seedance-2-0-reference-to-video",
    "queue_id": "123e4567-e89b-12d3-a456-426614174000"
  }' \
  -o output.mp4
```

The response is JSON (`{ "status": "queued" | "running" | "failed", ... }`) until the job completes, at which point the response body switches to `video/mp4` bytes. See [Video Generation](/guides/media/video-generation) for the full polling pattern.

***

## Troubleshooting

### `At least one reference is required for this model`

Reference-to-video submissions must include at least one of `reference_image_urls`, `reference_video_urls`, `image_references`, or `video_references`. Pure text-only generation isn't a valid R2V workflow — use `seedance-2-0-text-to-video` instead. `reference_audio_urls` alone is not sufficient (see the Audio section above).

### `reference_video_urls must have at most 3 videos`

The model caps reference videos at 3. If you need more clips, run a Stitch first (3 → 1), then use the output as a reference for a follow-up.

### `Per clip must be 2–15s` / aggregate `> 15s`

Per-clip duration is `[2, 15]` seconds **inclusive**; the sum across all reference videos is also capped at 15 seconds. Trim clips client-side before submission.

### Prompt routes to the wrong workflow

Workflow is inferred from prompt syntax. Common misroutings:

* Wanting to **Extend** but writing `Refer to ...` → model treats your video as a *donor*, not a canvas to continue
* Wanting to **Stitch** but writing `Refer to ...` → model picks one as the donor, ignores the others
* Wanting to **Edit** but writing `Generate a video based on <Video 1>` → ambiguous; model may default to Reference

Use the canonical prefixes exactly as written: `Strictly edit <Video 1>, ...`, `Extend <Video 1>, ...`, `<Video 1> + ... + followed by <Video 2>`.

### Quote doesn't match the queued amount

If you included a reference video but didn't pass `reference_video_total_duration` to `/video/quote`, the quote and the queued amount may differ. Always pass `reference_video_total_duration` (sum of all reference clip durations, in seconds) when reference videos are present.

***

## References

* Venice video queue endpoint: [`POST /api/v1/video/queue`](/api-reference/endpoint/video/queue)
* Venice quote endpoint: [`POST /api/v1/video/quote`](/api-reference/endpoint/video/quote)
* Companion guide: [Reference to Video](/guides/media/reference-to-video) (covers Kling O3 + Grok Imagine R2V)
* Companion guide: [Video Generation](/guides/media/video-generation) (queue / polling overview)