SD 2 Omni Reference VIP (Pro) by ByteDance. Generate videos using up to 9 image references, up to 3 video clips, and up to 3 audio references with priority routing. Reference materials in your prompt with @image1…@image9, @video1…@video3, and @audio1…@audio3. Also supports @omni-character:<char_id> for trained characters.
About this model
SD 2 Omni Reference VIP (Pro) generates videos using up to 9 image references and up to 3 audio references with priority routing. Reference your uploaded materials in the prompt using @image1–@image9 and @audio1–@audio3 syntax. Supports character sheets via @character:<id> and trained characters via @omni-character:<char_id> for consistent identity across generations.
Cost analysis
| Provider | Cost | Notes |
|---|---|---|
| muapiapp | $0.30/sec (pro) | VIP priority routing with multi-reference support — ideal for character-consistent and audio-synced generation. |
| Fal.ai | Not available | SD 2 Omni Reference VIP not available. |
| Replicate | Not available | SD 2 Omni Reference VIP not available. |
VIP priority routing with multi-reference support — ideal for character-consistent and audio-synced generation.
SD 2 Omni Reference VIP not available.
SD 2 Omni Reference VIP not available.
** Competitor pricing is estimated based on similar model architectures and usage tiers.
Configuration schema
| Parameter | Type | Description | Default |
|---|---|---|---|
| Prompt | string | Video description. Use @image1…@image9 to reference images, @video1…@video3 for videos, and @audio1…@audio3 for audio. Use @character:<request_id> for a Seedance 2 character sheet or @omni-character:<char_id> for a trained Kinovi character. Multiple characters are supported. | @image1 is the main character. The person walks along a city street at sunset, cinematic lighting. |
| Image URLs | array | Up to 9 reference image URLs (JPEG/PNG/WebP). Each Nth image corresponds to @imageN in the prompt. | https://d3adwkbyhxyrtq.cloudfront.net/ai-images/186/712345784292/4a8c5c70-abcc-4920-873e-b0e219986453.jpg |
| Video Reference URLs | array | Up to 3 reference video clip URLs (MP4, max 15s each). Each Nth video corresponds to @videoN in the prompt. | undefined |
| Audio Reference URLs | array | Up to 3 reference audio files (MP3/WAV, total max 15s). Each Nth audio corresponds to @audioN in the prompt. | undefined |
| Aspect Ratio | Enum (6 options) | Output video aspect ratio. | 16:9 |
| Duration (seconds) | int | Video duration in seconds. | 5 |
| High Bitrate | boolean | Enable high bitrate mode for better visual fidelity. Produces larger files. | false |
Video description. Use @image1…@image9 to reference images, @video1…@video3 for videos, and @audio1…@audio3 for audio. Use @character:<request_id> for a Seedance 2 character sheet or @omni-character:<char_id> for a trained Kinovi character. Multiple characters are supported.
@image1 is the main character. The person walks along a city street at sunset, cinematic lighting.Up to 9 reference image URLs (JPEG/PNG/WebP). Each Nth image corresponds to @imageN in the prompt.
https://d3adwkbyhxyrtq.cloudfront.net/ai-images/186/712345784292/4a8c5c70-abcc-4920-873e-b0e219986453.jpgUp to 3 reference video clip URLs (MP4, max 15s each). Each Nth video corresponds to @videoN in the prompt.
undefinedUp to 3 reference audio files (MP3/WAV, total max 15s). Each Nth audio corresponds to @audioN in the prompt.
undefinedOutput video aspect ratio.
16:9Video duration in seconds.
5Enable high bitrate mode for better visual fidelity. Produces larger files.
falseDeveloper documentation
Upload references: Add up to 9 images in images_list and up to 3 audio files in audio_files. Files are referenced positionally — the 1st image is @image1, 2nd is @image2, etc.
Write your prompt: Reference your materials explicitly: "@image1 is the main character. The person walks along a city street at sunset." Include @audio1 for audio-synced scenes.
Use character references: Add @character:<request_id> to include a character sheet from a SD 2 Character generation. Use @omni-character:<char_id> for a trained omni character (supports native character IDs).
Set aspect ratio and duration: Choose from standard aspect ratios and 4–15 second durations.
Submit and retrieve: Poll with the request_id or provide a webhook_url.
Frequently asked
Use @image1, @image2, etc. to reference images by position in images_list. Use @audio1, @audio2, etc. for audio_files. Use @character:<request_id> for SD 2 character sheets, or @omni-character:<char_id> for trained omni characters.
Upload audio files (MP3/WAV, total max 15s) in audio_files and reference them in your prompt with @audio1. The model generates video that is visually synchronized to the audio rhythm and content.
@character:<request_id> uses the character sheet image from a completed SD 2 Character generation — the image is injected into images_list and referenced in the prompt. @omni-character:<char_id> uses a trained omni character ID directly, which enables deeper identity consistency via the native character path.
Up to 9 images (images_list), and up to 3 audio files (audio_files, total max 15 seconds). Audio files are trimmed automatically if needed.