Seedance 1.0 Pro is ByteDance's video generation model, and doubao-seedance-1-0-pro-250528 corresponds to its specific invocation version. It converts text descriptions or static images into short videos, with a focus on balancing subject motion, visual structure, and camera expression. It is suitable for advertising clips, animated illustrations, and storyboards. You can build scenes from text or use first and last frames to constrain the starting and ending points of the image.
Submit requests to the public API at api.acedata.cloud using the documented parameters, then use the results in your application.
api.acedata.clouddoubao-seedance-1-0-pro-250528Input parameters and result formats vary by service. Use the public API for this model and follow its guide for generation, task retrieval and editing operations.
Multi-shot and motion performance are publicly available native capabilities of Seedance 1.0. Duration, input structure, and task operations are used according to this model's platform invocation scope.
A key feature of Seedance 1.0 is integrating multi-shot expression into video generation. Public examples demonstrate transitions between wide, medium, and close-up shots. When creating, you can describe in sequence a character entering a scene, performing actions, and emotional close-ups, allowing the short film to advance around events rather than simply keeping a single image in motion.
The model emphasizes natural subject motion and structural stability, making it suitable for dynamic content such as gliding, walking, and fabric movement. Specifying subject actions and camera movement separately in prompts helps clearly convey filming intent; the fixed-camera option is suitable for focusing attention on changes in the subject.
In addition to text generation, images can also serve as the starting point for videos, with first-and-last-frame combinations describing the direction of transition. It supports realistic, anime, film, advertising, and other stylistic expressions, combining existing compositions with actions, lighting, and camera movement in text for animated illustrations or visual concepts.
Use a static product image as the first frame, then describe camera movement, lighting changes, and motion in the surrounding environment to generate ad clips suitable for editing. First determine a landscape or portrait composition, then design the action around one main visual selling point; subtitles, voice-over, and brand information can be added later.
Break a short plot into consecutive shot descriptions, such as a character entering a room, examining an object, then cutting to a close-up of their expression, and use video to validate narrative pacing and shot-size relationships. The deliverable is a dynamic storyboard for discussion, suitable for comparing different shot approaches before formal filming or animation production.
Use an illustration or character scene image as the first frame, adding descriptions of actions such as a breeze, blinking, turning around, or slowly pulling back to create social content and showcase clips. If there is a clear ending composition, you can also provide a last frame so creation can focus on the changes between the two images.
1.0 Pro covers both text and image generation within the same model, making it suitable for maintaining both script creation and asset animation workflows. 1.0 Lite is divided into T2V and I2V variants; if the main goal is rapid previewing, consider 1.0 Pro Fast. 250528 and 251015 Fast are different invocation versions, and their model IDs should not be interchanged.
If the goal is a silent short film, first-to-last-frame transitions, or continuation of an existing 1.0 workflow, this version is more straightforward. When you need to generate video with sound, choose the audio-supported 1.5 Pro or 2.x; consider 2.0 when image, audio, and video joint references are needed, and 2.5 when you need to edit or extend existing video.
Provide text in content; images can use first_frame/last_frame to represent the starting and ending frames; the last frame should be consistent with the creation method and asset composition.
Specify model=doubao-seedance-1-0-pro-250528 for /seedance/videos; first test with duration=5, resolution=720p, and a clearly defined aspect ratio. The 1.0 series does not generate native audio.
First obtain the task_id asynchronously, then query /seedance/tasks or receive a callback; after completion, check the subject, action, and ending, then save the selected final video and task record. This model outputs no native audio, so add voice-over and music in post-production when sound is needed.
Start with a first frame showing an empty vase and end with a last frame showing it filled with flowers, with a hand naturally inserting flower stems in between while keeping the tabletop and camera position stable.
Prepare connectable first and last frames, and check the hand movement and final-frame composition; 1.0 does not use audio generation settings, and sound is produced in post-production.
No. doubao-seedance-1-0-pro-250528 is the fixed version introduced here, while Pro Fast uses doubao-seedance-1-0-pro-fast-251015. Both belong to the 1.0 series, but Fast is designed for faster generation; when maintaining existing projects, you should explicitly specify the required model.
Add a text prompt and two image_url items in content, setting their roles to first_frame and last_frame respectively. Each image address should be placed inside the image_url.url object. The text should describe the actions and camera changes between the two frames; do not write image_url directly as a string.
This version uses 480p, 720p, and 1080p resolutions and does not support generate_audio. During production, you can use it for high-definition visual clips and complete audio and editing separately; if the task itself requires 4k or native audio generation, choose another Seedance model with the corresponding capabilities.
First establish the shared subject and scene, then describe the framing, actions, and transitions of each shot in chronological order, such as entering in a wide shot, operating in a medium shot, and expressions in a close-up. Seedance 1.0 has demonstrated narratives with 2–3 shots within 10 seconds; in actual creation, keep events focused and avoid contradictory actions.
Submit model and content to /seedance/videos. Set async to true to obtain a task_id first, then query the task; you can also provide callback_url to receive completion notifications. The data.video_url in a successful result is used to retrieve the video, and your integration should handle both task status and error information.
Choose a development approach or application suited to this model and follow its setup guide.
Python SDK Integration Guide
SDK installation, authentication and examplesTypeScript SDK Integration Guide
SDK installation, authentication and examples
