ByteDance's all-in-one AI creation studio — one browser editor fronting a rotating lineup of frontier video and image models, plus talking avatars.
Last verified · 2026-10-29 · by Moe Ameen
Dreamina is ByteDance's all-in-one AI creation studio for images, video, and talking avatars, at dreamina.capcut.com (also reachable as dreamina.ai). It is the international counterpart to ByteDance's Jimeng app and sits inside the CapCut ecosystem; the common way to describe that stack is a pipeline where Dreamina generates the raw visuals, CapCut edits them, and TikTok or Douyin distributes.
What sets Dreamina apart from a single-model generator is breadth. One browser editor exposes a rotating lineup of frontier models rather than one engine: on the video side it has run Seedance 2.5, Seedance 2.0, a lighter Seedance 2.0 Mini, OpenAI's Sora 2, and Google's Veo 3.1; on the image side, Seedream and Nano Banana families. You pick the model per job instead of being locked to whatever one vendor ships. Model availability shifts as ByteDance adds and retires engines, so treat any specific model name as a snapshot.
Around the core generators, Dreamina bundles text-to-image and text-to-video, image-to-video, AI avatar and talking-avatar makers, face swap, logo and poster design, and a marketing studio aimed at product and ad visuals. Access runs on a free daily-credit tier with paid subscription plans above it, and credit cost varies by model, resolution, and clip length; rollout, free allowances, and pricing differ by region, so confirm current terms in your account.
Dreamina is a generation-and-light-editing environment. It does not burn in branded captions for sound-off feeds, hold one persona's identity across many posts, reframe per platform, or schedule and publish — the output is yours to finish and post. ByteDance also applies safeguards: generated output carries an invisible watermark alongside a visible one added on download, and Dreamina's own safety guide bars generating images of political figures.
Dreamina's whole pitch is model choice — swap to whichever engine renders your shot best this month. That flexibility is real, and it is also the reason your downstream workflow should be the opposite: a single, stable layer that does not care which model produced the file. Kompozy is that layer. Whatever Dreamina hands you — a Seedance clip, a Sora 2 scene, a Seedream poster, a talking-avatar take — Kompozy treats it as one input and turns it into finished, captioned, on-brand posts, without you rebuilding anything when Dreamina's lineup changes.
Concretely: drop a Dreamina export into Kompozy and it burns in on-style captions for muted feeds, reframes the master to 9:16, 1:1, or 16:9 per destination, and composites it into a Clipped Short or Marketing Short. A Dreamina image or poster becomes the hero of a Carousel Post or a face-locked Persona Photo, and the concept gets fanned into a quote card, a blog draft, and platform-native captions in your voice through the Persona Brief. Then Kompozy schedules and publishes the whole set across eight social platforms plus blog and email from one queue — the distribution Dreamina leaves entirely to you.
Dreamina is ByteDance's all-in-one AI creation studio for images, video, and talking avatars, at dreamina.capcut.com and part of the CapCut ecosystem. It runs a rotating lineup of frontier models — Seedance, Sora 2, and Veo 3.1 for video, Seedream and Nano Banana for images — behind one browser editor, on a free daily-credit tier with paid plans above it.
They are complementary parts of the same ByteDance ecosystem: Dreamina generates new images and video from prompts, while CapCut edits and assembles footage. A common way to describe the stack is Dreamina generates, CapCut edits, and TikTok distributes.
It has a free daily-credit tier so you can test generation, with paid subscription plans for more credits and higher-tier models. Credit cost varies by model, resolution, and clip length, and free allowances and pricing differ by region, so confirm current terms in your account.
It exposes several behind one editor and the lineup shifts over time; it has run Seedance 2.5, Seedance 2.0, Sora 2, and Veo 3.1 for video, plus Seedream and Nano Banana families for images. For the model-specific deep dive on its headline video engine, see our Dreamina Seedance 2.5 page.
No. It generates and lightly edits, then stops. To caption, reframe per platform, keep a consistent persona, and schedule and publish across TikTok, Reels, Shorts, X, LinkedIn, and more, bring the export into a tool like Kompozy, which also fans the asset into supporting posts in your voice.