dreamina-3-0/text-to-video
Generate lifelike motion visuals fast with Dreamina 3.0 for designers.
| Parameter | Required | Type | Default | Range / Options | Description |
|---|---|---|---|---|---|
| prompt* | Yes (*) | string | — | — | Scene description using your references as cues |
| images | No | array (image URIs) | — | 0–9 | Reference images for look and identity |
| videos* | Yes (*) | array (video URIs) | — | 1–3 | Reference clips for motion and rhythm (at least one required) |
| audios | No | array (audio URIs) | — | 0–3 | Reference audio for mood and pacing |
| aspect_ratio | No | string | 16:9 | 16:9, 9:16, 1:1, 4:3, 3:4, 21:9, adaptive | Output frame ratio |
| duration | No | integer | 5 | 4–30 (seconds) | Clip length |
| generate_audio | No | boolean | true | true / false | Native synchronized audio |
Seedance 2.5 Reference to Video 4K bills on total counted video seconds (reference video duration + output duration):
Image and audio references are not billed as duration.
Generate lifelike motion visuals fast with Dreamina 3.0 for designers.
Cinematic 4K text-to-video at $0.47 per second of output.
Generate high quality videos from text prompts using Kling 1.6 Pro.
Generate high quality videos from text prompts with Wan 2.2 Plus.
Generate cinematic motion from text or images with efficient 3D VAE-based video synthesis for creatives.
LTX 2.5 Pro: still-to-video with synced audio, quality mode
Seedance 2.5 Reference to Video 4K is a ByteDance reference-guided endpoint that renders a 4K clip from reference images and clips. It fits identity-locked character finals, reference-matched product films, and campaign sets that must share one look across several shots.
Seedance 2.5 Reference to Video 4K takes up to 9 reference images, 1 to 3 reference videos, and up to 3 reference audio files. At least one reference video is required; images and audio are optional cues for look, identity, and mood.
Billing counts total video seconds, meaning your reference video duration plus the output duration, at $1.09 per second. Reference images and audio are not billed as duration, so trimming long reference clips is the main way to control what Seedance 2.5 Reference to Video 4K costs per run.
Beyond the references, Seedance 2.5 Reference to Video 4K accepts a prompt, aspect_ratio (16:9, 9:16, 1:1, 4:3, 3:4, 21:9, or adaptive), duration from 4 to 30 seconds with a default of 5, and generate_audio. Resolution is fixed at 4K.
Say what each reference is for rather than redescribing it. Phrases like "using the reference image for wardrobe and the reference clip for camera rhythm" tell Seedance 2.5 Reference to Video 4K which cue drives which part of the shot, which keeps identity and motion from fighting each other.
Yes. Build the shot in the RunComfy model UI, then call the same Seedance 2.5 Reference to Video 4K template through the RunComfy API, passing publicly reachable URLs for every reference. Both paths consume usd / credits from the same balance.
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.
