veo31-fast is the Fast video model in the Google Veo 3.1 series, suitable for turning text concepts or static images into dynamic shots and quickly comparing different creative directions. It supports text-to-video, single-image first-frame guidance, two-image first-and-last-frame transitions, and visual extension of existing videos. With landscape and portrait aspect ratios and asynchronous task processing, it can be used for advertising drafts, product presentations, and storyboard previews.
Submit requests to the public API at api.acedata.cloud using the documented parameters, then use the results in your application.
Input 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.
The above lists the actual creation and API specifications of veo31-fast on this platform and does not automatically treat the native capabilities of the series as all features available through this entry point.
Text-to-video is suitable for the creative stage when visual assets have not yet been prepared. Enter the subject, environment, action, and camera movement to generate videos for review. You can keep the scene fixed and adjust only camera movement or action descriptions to compare different presentation directions; the Fast positioning is better suited to this workflow of exploring first and selecting later.
Image-to-video is not simply about adding reference images: one image determines the starting point of a shot, while two images guide its start and end points. Combined with prompts describing the intermediate action, it can create product reveals, scene changes, or storyboard connections. It is suitable for projects with existing key visuals and also makes it more intuitive to compare different transition options.
With video extension, you can continue actions or camera changes after already generated footage and use new prompts to guide subsequent content. Extension results can also be extended again, making it suitable for building shots segment by segment. Both generation and extension support asynchronous processing, allowing applications to save task identifiers first and then receive completed results and video links.
Enter the product theme, usage environment, and opening action to create different shot drafts for the same advertising concept. Portrait format is used for mobile creative review, while landscape format is used for presentations or storyboard discussions. The deliverable is a playable video candidate, making it easier for teams to evaluate the visual direction rather than relying solely on written scripts to imagine the final result.
Use a product image as the first frame, describe the display action and camera movement, and generate a dynamic presentation draft; if a target ending image already exists, you can submit both the first and last images to explore the transition between the two states. Suitable for product displays, packaging reveals, and concept demonstrations; product form and details should still be checked before final use.
Turn storyboard keyframes into video to observe whether scene transitions match the narrative intent; if a shot needs to continue, use the generated video's ID to call the extension. The deliverable can be used for director discussions, rough editing rhythm tests, or concept presentations, helping clarify how shots connect before formal production.
When the task focuses on comparing multiple prompts, opening actions, or first-to-last-frame transitions, prioritize veo31-fast. If the final image's refined quality matters more, use the same assets to compare it with veo31's Quality mode. The tradeoff between the two should be based on the specific shot result; do not interpret Fast as a version with fixed processing time or one that is better suited to all tasks.
veo31-fast is suitable for text-only creation and for using one or two images to control a shot's beginning and end. If the task is to combine elements from multiple images, choose veo31-fast-ingredients, whose creation method requires image input. The two solve different problems: the former emphasizes temporal start-to-end changes, while the latter emphasizes combining content from multiple images; their image rules should not be interchanged.
Use text2video for text only; use image2video for image-driven generation, with one image in image_urls as the first frame and two images as the first and last frames, and describe the action that occurs in between.
Specify model=veo31-fast at /veo/videos, choose the action based on text or images, and fill in prompt. Start with aspect_ratio=16:9 and resolution=720p, then increase the output setting according to this model's supported range.
After setting async=true, save task_id and obtain the completed video through /veo/tasks or a callback; also save data[].id. Veo 3.1 extension uses the video ID and cannot use the task ID as a substitute.
Starting from the product's first frame, slowly raise the camera upward to show the desktop and window-side environment, keeping the subject unchanged and without switching scenes.
Compare movement range and composition, with at most two first and last frames; when multiple independent assets need to be merged into a new shot, choose Ingredients instead.
Submit model=veo31-fast, action=text2video, and prompt to /veo/videos. The prompt can be organized by subject, scene, action, and camera changes; set aspect_ratio=9:16 when you need portrait orientation. After completion, obtain the video from video_url in the result.
When using image2video, one image is used for first-frame guidance, while two images are used for first-and-last-frame guidance. Put the image links in image_urls, and use prompt to describe the actions or transitions you want to occur in between. This model is not a blending mode that freely combines multiple images into a scene.
You can select resolution=1080p during generation, or use the get1080p action on an already generated video and submit the corresponding video_id. Generation, obtaining a higher-resolution version, and extension are different actions; extension requests use the 720p or 1080p options of /veo/extend.
Submit model=veo31-fast and data[].id from the generation result as video_id to /veo/extend, and optionally add prompt to guide the subsequent footage. Extension results can also be extended further; save the task ID and video ID separately to avoid using the task-query identifier for extension.
You can enter Chinese prompts and enable automatic translation with translation=true. For application integration, you can set async=true to obtain task_id first and then query the result; you can also provide callback_url to receive completion notifications, then save the video ID and download link for subsequent production.
Choose a development approach or application suited to this model and follow its setup guide.
↗Python SDK Integration Guide
SDK installation, authentication and examples↗TypeScript SDK Integration Guide
SDK installation, authentication and examples
↗
↗
↗This is only a basic call example. See the full docs for more parameters and advanced usage.
View Docs