luma-ray-2/image-to-video
Lifelike characters, realistic physics, and stunning effects.
Seedance 2.0 Pro is a multimodal text-to-video model from ByteDance Seed that turns scene descriptions and optional references into short cinematic clips. On RunComfy you drive generation with a prompt plus optional images (up to 9), videos (up to 3), and audio (up to 3) for multimodal reference mode, and you can set aspect ratio, duration, resolution, generate audio, seed, and optional tools (e.g. [{ "type": "web_search" }] to allow online search when the model chooses).
Output (this playground): selectable 480p / 720p / 1080p · 4–15 seconds · aspect ratios including adaptive (default), 16:9, 9:16, 4:3, 3:4, 1:1, 21:9 · optional native audio when Generate audio is enabled
type: web_search so the model can search the web when needed; check usage.tool_usage.web_search on the task query response for how many searches ran.Inputs match the RunComfy OpenAPI Input schema for this template.
| Parameter | Required | Type | Default | Range / Options | Description |
|---|---|---|---|---|---|
| prompt* | Yes (*) | string | — | CN ~≤500 chars, EN ~≤1000 words rec. | Text prompt for the video |
| image_url (Images) | No | array (image URIs) | [] | 0–9 | jpeg, png, webp, bmp, tiff, gif |
| video_url (Videos) | No | array (video URIs) | [] | 0–3 | mp4, mov; duration 2–15 s per clip |
| audio_url (Audio URLs) | No | array (audio URIs) | [] | 0–3 | wav, mp3; 2–15 s, < 15 MB |
| aspect_ratio | No | string | adaptive | adaptive, 16:9, 9:16, 4:3, 3:4, 1:1, 21:9 | Adaptive: model picks closest ratio; check task result for actual ratio |
| duration | No | integer | 5 | 4–15 (seconds) | Clip length in whole seconds |
| resolution | No | string | 720p | 480p, 720p, 1080p | Output resolution preset |
| generate_audio | No | boolean | true | true / false | When true, outputs video with synchronized audio (speech, SFX, music) |
| seed | No | integer | — | — | Random seed for video generation |
| tools | No | array of objects | [] | type: web_search only | Declares allowed tools; with web_search, model may search per prompt; see usage.tool_usage.web_search on task query |
1) Describe your scene — Cover subject, action, setting, mood, lighting, and camera. For prompt length, Chinese ~≤500 characters or English ~≤1000 words is recommended.
2) Add references (optional) — Upload images for look or identity; add short reference videos or audio. Respect duration and file-size limits for each modality.
3) Choose aspect ratio — Use adaptive for general exploration, or a fixed ratio (e.g. 9:16 or 16:9) for a known deliverable.
4) Set duration — Any integer from 4 to 15 seconds.
5) Pick resolution — 480p for fast drafts, 720p as the default balance, 1080p for higher detail.
6) Generate audio — Leave enabled for dialogue, SFX, or music; disable if you only need silent video.
7) Optional seed — Fix the seed while iterating so changes come from prompt or media, not randomness.
8) Optional web search — Add tools [{ "type": "web_search" }] when you want the model to be able to look up timely facts (adds latency when used); after the job completes, read usage.tool_usage.web_search on the task query.
9) Generate and iterate — Refine wording, swap references, or adjust ratio, duration, and resolution.
In short, Seedance 2.0 Pro on RunComfy supports text plus up to nine images, three videos, and three audio references, 4–15 s clips, 480p–1080p presets, flexible aspect ratios including adaptive, and optional native audio aligned with the live playground fields.
Lifelike characters, realistic physics, and stunning effects.
Consistent characters, objects, and scenes in any setting or angle.
Transform stills into narrative clips with synced audio and fluid camera motion.
Generate lifelike motion visuals fast with Dreamina 3.0 for designers.
Create photo-based, speech-aligned videos with natural motion
Browser tool for quick, detailed creative clips from images or text
On this playground, resolution is one of 480p, 720p (default), or 1080p. Aspect ratio can be adaptive (default—the model picks the closest ratio; the task result reflects the actual output) or fixed: 16:9, 9:16, 4:3, 3:4, 1:1, or 21:9.
Prompt: Chinese ~≤500 characters or English ~≤1000 words is recommended. Images: up to 9 files (jpeg, png, webp, bmp, tiff, gif). Reference videos: up to 3 (mp4, mov), each 2–15 seconds. Reference audio: up to 3 (wav, mp3), each 2–15 seconds and under 15 MB. Only prompt is required; all reference fields are optional.
Use the RunComfy API with the same Input fields as the playground (prompt, image_url, video_url, audio_url, aspect_ratio, duration, resolution, generate_audio, seed). Validate prompts and media limits in the UI, then use your account API key and credits for automated jobs.
Seedance 2.0 Pro targets cinematic short clips with multimodal references (many images plus optional video and audio), 4–15 second duration control, flexible aspect ratios including adaptive, and native audio when generate_audio is on—useful for lip-sync and synced SFX or music. Exact benchmarks depend on your content; compare outputs on identical prompts and references.
Choice depends on your workflow. On RunComfy, this Seedance 2.0 Pro template offers up to nine images, three reference videos, three audio references, 480p–1080p presets, and toggled generated audio. Wan 2.5 and Kling 2.6 differ in pricing, limits, and strengths—run parallel tests on your typical prompts and reference sets.
Yes, in practice reference images (and optional reference video or audio) plus a clear prompt help anchor identity, wardrobe, and tone. There is no special “@” syntax documented in the API schema; consistency comes from aligned text and reference media within the supported limits.
When Generate audio (generate_audio) is true (the default), the model is described as outputting video with synchronized audio (speech, SFX, music). Set it to false if you only want silent video. Lip-sync quality still depends on prompt clarity and the scene you describe.
Duration is an integer from 4 to 15 seconds (default 5). Pick any whole-second value in that range per generation.
Commercial use depends on ByteDance’s licensing terms for the model and RunComfy’s terms of service. Review the official model license and RunComfy documentation, or contact hi@runcomfy.com before using generated footage in paid campaigns or public distribution.
Teams and creators who need short cinematic clips with optional image, video, and audio references, platform-specific aspect ratios, up to 1080p exports on this playground, and optional built-in audio—for example ads, social video, previsualization, and branded narrative tests.
RunComfy is the premier ComfyUI platform, offering ComfyUI online environment and services, along with ComfyUI workflows featuring stunning visuals. RunComfy also provides AI Models, enabling artists to harness the latest AI tools to create incredible art.
