Veo 3.1 Fast Ingredients is the video entry point in the Google Veo series for creating with blended reference images, suitable for projects that already have product images, character images, or scene assets. It uses 1-3 images together with text prompts to generate dynamic visuals, focusing on organizing reference content into video rather than specifying first and last frames. Images are required; text-only generation is not supported. It is suitable for ad concepts, concept shots, and iterative asset planning.
Submit requests to the public API at api.acedata.cloud using the documented parameters, then use the results in your application.
api.acedata.cloudveo31-fast-ingredientsInput 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 are the input and output settings for this entry point on this platform and do not represent the complete specifications of all native Veo series versions.
Use images to provide visual content, then use prompts to explain their relationships and actions within the scene. You can develop videos around product, character, or background assets without having to describe every aspect of their appearance using text alone. Multi-image blending is suitable for exploring asset combinations; when submitting images, clearly specify which elements are the main subjects and which are environmental references.
Images provide visual references, while text provides narrative and motion instructions. Describe how the subject moves, how the camera observes it, and the atmosphere of the image, rather than simply writing “make the image move.” Chinese-language creation can enable automatic translation with translation, making it easier to use existing Chinese scripts for video tasks while keeping the prompt structure clear.
Generation tasks can be submitted asynchronously: first save the task_id, then query the completed result; you can also set callback_url to receive result notifications. The completion response includes the video link, video ID, and status, making it easier to associate each generation with asset combinations and prompt versions, and organize subsequent downloading, review, and asset archiving.
Input product images and scene references, then use text to describe the product's position, presentation action, and camera movement to generate advertising visuals for team discussion. Suitable for comparing different environments and presentation approaches before formal shooting or detailed retouching, delivering reviewable dynamic assets rather than simply assembling multiple static images into a carousel.
When character designs and environment concept art are already available, use them as blend references and describe shots of the character entering the scene, observing the environment, or completing an action. The results are suitable for concept shorts and storyboard discussions, helping the team determine whether the visual combination works before deciding whether to continue producing more complete shots.
Plan landscape and portrait compositions separately around the same set of reference images, and generate asset options with different action or shot descriptions. Landscape can be used for scene presentation, while portrait can emphasize the subject. When delivering, record versions by aspect ratio, reference image, and prompt for easy selection of visuals suited to subsequent editing and publishing layouts.
If the goal is to organize content from several assets into a dynamic scene, choose veo31-fast-ingredients. If the opening or ending visual is already determined and you want images to serve as the first frame or the first and last frames, standard veo31-fast is better suited to the task. Both use reference images, but their control semantics differ: blending references does not fix the start and end points of the video.
If you already have images and want to explore combination options, prioritize the Ingredients entry point; if you only have a text script, choose veo31-fast or veo31, which support text-to-video. If the project explicitly requires 4K, consider veo31, which supports that setting, rather than adding 4k to an Ingredients request. First ensure the choice meets the input method and delivery requirements, then compare the actual visuals.
Prepare 1–3 publicly accessible images, clearly describing how the product, person, or scene should be combined; these images are material references, not start and end frames.
Specify model=veo31-fast-ingredients and action=ingredients2video for /veo/videos, and provide image_urls and prompt. Start with aspect_ratio=16:9 and resolution=720p, then increase the output level within this model's supported range.
After setting async=true, save the task_id and retrieve the completed video through /veo/tasks or a callback; verify reference material integration and subject consistency; if the result includes an audio track, listen to it, then save the video ID and downloaded file for subsequent production.
The first image is a teacup, the second image is a wooden table, and the third image provides a window-side environment; naturally place the teacup on the wooden table, with the camera slowly pushing in under warm light.
You must provide 1–3 image_urls and select ingredients2video; the images are used as integration references and do not mean the first image will necessarily become the first frame.
Yes, you can input 1–3 images, and one image also meets the requirements. However, creation here still follows the reference fusion approach and is not equivalent to a fixed first frame. It is recommended to specify in the prompt the main content you want to preserve, as well as the desired action or scene changes.
Send a POST request to /veo/videos, set model to veo31-fast-ingredients, use the ingredients2video operation, provide image links through image_urls, and use prompt to describe the fusion relationship and camera intent. Images must be provided.
Ingredients requires images and focuses on fusing reference content; veo31-fast supports text-only generation, as well as a single-image first frame and two-image first and last frames. Choose the former when you want to combine assets, and the latter when you want to start from a specified image or connect specified first and last images.
You can select 1080p and choose 9:16 vertical or 16:9 horizontal through aspect_ratio. When output settings are not specified, 720p is used by default. When planning vertical video assets, it is recommended to also specify the subject position and composition in the prompt, to avoid changing only the aspect ratio without adjusting the camera design.
After setting async=true, first obtain the task_id, then retrieve the status and result through task queries; you can also use callback_url to receive completion notifications. After completion, read video_url from data to get the video, and save the video ID to facilitate linking generation records and subsequent processing.
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