이미지: X — 미디어·생성AI
Summary
- Krea unveiled its video production tool "Seedance Studio" on X on August 19, 2026
- The tool combines prompt presets, camera paths, and 3D scene controls in a single screen for working with Seedance 2.5
- Seedance is a video generation model built by ByteDance Seed, the AI research organization of TikTok parent ByteDance
- 도구 이름
- Seedance Studio
- 발표 주체·시점
- 크레아(Krea) 공식 X 계정 · 2026년 8월 19일 19:14(UTC)
- 대상 모델
- Seedance 2.5
- 공개된 구성 요소
- 프롬프트 프리셋, 카메라 패스(camera paths), 3D 씬 컨트롤
- Seedance 개발
- 바이트댄스 시드(ByteDance Seed) — 틱톡 모회사 바이트댄스의 AI 연구 조직
- 크레아의 성격
- 자체 이미지·영상 모델을 훈련하면서 타사 모델 수십 개도 함께 유통하는 플랫폼
- 이용자 참여
- 크레아가 게시물에서 추가로 원하는 기능을 댓글로 받는 중
For anyone who has used video generation AI, the most common obstacle isn't image quality — it's control. To get the output to reflect where the camera starts and moves to, or where a character and background are positioned, you ultimately have to describe it in words. Seedance Studio, unveiled by Krea on August 19, moves that point of control onto the screen itself.
What the announcement confirmed
In a post on its official X account, Krea introduced Seedance Studio as "a tool that brings together everything you need to create with Seedance 2.5 in one place." Three specific components were named: prompt presets, camera paths, and 3D scene controls.
The post came with an introduction video, and in the same post Krea asked users what features they'd like to see added. Comments began appearing within hours of the announcement, with one user citing model training features while noting they use Krea as their primary platform.
| Component | Name as stated in the announcement | Area covered |
|---|---|---|
| Prompt presets | prompt presets | Loading saved instruction formats instead of writing prompts from scratch each time |
| Camera paths | camera paths | The path along which the camera moves |
| 3D scene controls | 3D scene controls | Controls for handling a scene in three dimensions |
The post did not specify how each control appears on screen or how many detailed parameters exist. What's confirmed is that all three axes have been brought together in a single tool.
What Seedance is, and where Krea fits in
It's worth clarifying a point that's easy to confuse. Seedance is a video generation model developed by ByteDance Seed, the AI research organization of TikTok's parent company ByteDance. This model typically appears not as a standalone app, but as a "selectable model" within other production tools.
Krea is a company that does two things simultaneously. Founded in San Francisco in 2022, it built its reputation on tools that generate images and video in real time, and it trains its own proprietary image and video models. At the same time, it aggregates dozens of models built by others — such as Google's Veo, OpenAI's Sora, and Kuaishou's Kling — into a single screen where users can choose among them. Seedance Studio falls into the latter category. The model originates from ByteDance, while Krea built the workspace for handling it.
Work produced with Seedance has appeared repeatedly in domestic coverage recently. The 1990s Hong Kong-cinema-style AI short film "Red Flag," released by Higgsfield on YouTube on August 19, was also made using Seedance run on the Higgsfield platform.
Why camera paths get their own category
What determines the output of a video generation model can be broken down into two broad factors: what is shown, and how it is viewed. The first is relatively easy to describe in words — a character's clothing, the time of day, the texture of a background. The second is harder. A camera move that circles from behind a character to the front, an angle that grazes past a wall, a change in distance from the subject — the moment these are put into a sentence, room for interpretation widens. This is why entering the same prompt twice can still produce different camera movements.
The fact that camera paths and 3D scene controls appear as separate items in the announcement signals an intent to handle this second axis through on-screen manipulation rather than text. Prompt presets, by contrast, are closer to the first axis — saving and recalling frequently used style instructions.
How to try it
Where to start — Seedance Studio is a newly added tool within the Krea service. A link to the tool was included in the announcement post, though the exact menu location within the app was not specified in this announcement. The starting point is logging into a Krea account and finding the Seedance Studio entry in the list of video tools.
Workflow — Based on the three components confirmed in the announcement, the flow appears to be as follows:
- Open Seedance Studio and confirm that the model in use is Seedance 2.5.
- Choose a prompt preset close to the desired tone or style — a step that removes the need to write a sentence from scratch every time.
- Use 3D scene controls to set up the composition within the scene.
- Set the path the camera will follow using camera paths.
- Run the generation, check the result, and go back to adjust only the specific control tied to whatever doesn't match.
What you can try — For a product demo video, you could fix lighting and texture tone with a preset and vary only the camera path to produce the same product from multiple angles. For a short narrative video, instead of writing a new prompt for every scene, you could share a single preset across scenes and distinguish cuts through camera movement alone. This is generally where repetitive work saves the most time.
This is the extent of what's been disclosed so far. Krea is taking requests for additional features, and anyone with a specific request can leave a comment on the relevant post on Krea's official X account.
Editor's take
The competitive axis in video generation tools is shifting from the model itself to the control interface. A year or two ago, most news in this space concerned output quality — how many seconds of video could be generated, whether fingers rendered correctly. That's no longer the case. Quality has converged to a similar level across companies, and the remaining differentiator is how many attempts it takes to get the desired shot. Seedance Studio bringing together prompt presets, camera paths, and 3D scene controls in a single screen sits squarely within that trend.
Krea's position in this structure is interesting. Because it both builds its own models and distributes others', it doesn't lose out regardless of which model wins — as long as it holds the workspace. Model performance can flip within months, but working habits don't change that quickly. A user who has accumulated presets and grown accustomed to camera path controls will return to the same screen even after the underlying model changes. This division of labor — ByteDance Seed building the model, Krea building the tool — reflects an industry structure, familiar from ingredients companies splitting off from kitchen-equipment companies, now taking hold in video AI as well.
For domestic teams, two practical points stand out. First, teams evaluating adoption of video AI should look at control interface design before benchmark scores. In practice, the real cost leak isn't rendering fees but the number of re-runs needed because the desired shot didn't come out. Whether presets can be shared at the team level, and whether camera movements can be saved in a reproducible form, are what determine that number. Second, there's typically a gap between the announcement of such tools and the point at which they actually stabilize. Early versions often have inconsistent accuracy even when controls exist. Rather than deploying directly on commercial projects, it's safer to run several internal tests first, confirm reproducibility, and only then integrate.
Similar announcements are likely to follow in the coming weeks. The very fact that Krea asked users for additional feature requests suggests this category still lacks a standard control interface. Timeline editing, character consistency locking, and reference-image-based scene composition are likely candidates near the top of the next wish list. Where video AI tools ultimately converge already has an answer: the shape of editing software.
