JUN 26, 2026

Seedance 2.0 Mini on the Dreamina API

Dreamina API v1 now supports Seedance 2.0 Mini (seedance-2.0-mini) — the cheapest, fastest member of the Seedance 2.0 family. It runs on POST /videos at 720p, 4–15s, and supports the full Seedance 2.0 input set: text-to-video, first/last frame, and Omni Reference.

Omni Reference is the interesting one. You upload several images, drop them straight into the prompt as @image1, @image2, …, and the model composes them into a single scene. A quoted line in the prompt becomes spoken dialogue in the clip, with audio generated automatically.

This post walks one Omni demo end to end: three unrelated reference images become a 15-second 720p clip with motion and a spoken line, all through the public Dreamina API v1.

The demo: three references → one 15s clip

Reference images

Three images, each uploaded via POST /assets and referenced by placeholder in the prompt:

Anime-character rider reference
@image1 — rider
Motorcycle reference
@image2 — motorcycle
Golden Gate Bridge reference
@image3 — bridge background

Omni does not require the references to share an aspect ratio, so the portrait, the landscape, and the background shot mix freely.

Generate the clip

One call to POST /videos. model: seedance-2.0-mini, duration: 15, ratio: 16:9, and three omni_N_imageRef assets matched to the @imageN placeholders. The prompt does the directing — a tracking shot, the riding action, and a line for the always-on audio.

curl --location 'https://api.useapi.net/v1/dreamina/videos' \
--header 'Authorization: Bearer user:12345-…' \
--form 'account="CA:[email protected]"' \
--form 'model="seedance-2.0-mini"' \
--form 'duration="15"' \
--form 'ratio="16:9"' \
--form 'omni_1_imageRef="CA:[email protected]:…-uri:…"' \
--form 'omni_2_imageRef="CA:[email protected]:…-uri:…"' \
--form 'omni_3_imageRef="CA:[email protected]:…-uri:…"' \
--form 'prompt="Tracking shot following @image1 riding @image2 at high speed along the wet beach, front wheel slicing through the shallow surf and kicking up a spray of water, sand and sea-spray flying, cinematic motion blur, @image3 in the background. She grins and shouts excitedly: Woo-hoo! Fastest ride on the whole coast!"'

Result

From three stills, the rider, the motorcycle, and the bridge land in one beach scene — with the motion and the spoken line driven by the prompt alone, on the most cost-effective Seedance 2.0 model.

Mini, Fast, or full 2.0?

Seedance 2.0 Mini is the cheapest and quickest option, and it is 720p-only. For 1080p or 4K output use seedance-2.0 (full), which also gained a 4k tier in this same update. seedance-2.0-fast sits in between. All three share Omni Reference, first/last frame, and text-to-video.

For the current, authoritative capabilities check the live Model Matrix and the POST /videos docs.

Get started

Dreamina API v1 docs · Setup Dreamina · POST /videos · POST /assets