happyhorse-1.1-r2v is the reference-image-to-video model for HappyHorse 1.1, suitable for tasks with existing character, clothing, or prop assets where you want to continue creating short videos in new scenes. It combines ordered reference images with textual shot descriptions, using visual assets to constrain subjects and prompts to arrange actions, environments, and camera work, with a focus on short-video creation requiring high character consistency.
Submit requests to the public API at api.acedata.cloud using the documented parameters, then use the results in your application.
api.acedata.cloudhappyhorse-1.1-r2vInput 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.
Native specifications include audio capability and 480P output; this platform entry provides reference-image generation calls at 720P and 1080P.
Unlike describing appearance with text alone, reference images provide clear visual grounding for characters, clothing, and props. You can change environments and actions in the prompt while continuing to use the same character assets, making it suitable for creating short shots of one character across different scenes; consistency is a creative goal, not a guarantee of complete frame-by-frame uniformity.
Images are not unordered attachments: use character1 and character2 to point to corresponding assets, and specify which image provides the subject and which provides clothing or visual style. For example, have character1 walk through grass while adopting the leather and gold-trim style of character2, establishing a clear relationship between assets and shot descriptions.
Submit a task through POST /happyhorse/videos, explicitly select reference_to_video and this model, then set duration, resolution, and aspect ratio. When long connections are inconvenient, use async or callback_url; after saving task_id, query the results or receive completion notifications to obtain the video link for downloading and further editing.
Input reference images of the same character, describe the scene, actions, and filming approach for each shot, and produce editable short clips. For example, create shots of a character walking on grass or pausing on a street, first verify appearance consistency, then select results that meet narrative requirements to assemble into a series.
Prepare character, costume, or prop assets, clearly specify the purpose of each reference image in the prompt, and generate dynamic previews of characters carrying props or presenting looks. Deliverables are suitable for discussing visual direction, action arrangements, and scene atmosphere; when product details are involved, check whether textures, logos, and structures match the original design.
Using the same set of character reference assets, create landscape, portrait, or square short clips for creative tests in different display placements. When providing input, adjust both the aspect ratio and composition description rather than changing only the ratio; after output, check subject placement and action space before passing it to the editing workflow to add subtitles and brand information.
If you want a character to continue appearing in a new environment without requiring the reference image to become the opening frame exactly as-is, choose happyhorse-1.1-r2v. If the task focuses on animating an approved image as the first frame, choose happyhorse-1.1-i2v; if there are no image assets and you mainly rely on script and shot text, use happyhorse-1.1-t2v.
Both 1.1-r2v and 1.0-r2v generate from reference images. The stated difference in public specifications is that 1.1 adds a native 480P option, while this entry still provides 720P and 1080P. Do not interpret version numbers as a fixed quality increase. If you already have a video that needs costume changes, style transfer, or partial replacement, choose happyhorse-1.0-video-edit.
Prepare 1–9 ordered image_urls, and use character1, character2, and so on in the prompt to reference characters or subjects by their corresponding sequence numbers.
Set model=happyhorse-1.1-r2v and action=reference_to_video for /happyhorse/videos. Choose an integer duration of 3–15 seconds and 720P or 1080P; text and reference-image generation can set ratio.
Use async or callback_url to integrate background tasks, save task_id and query /happyhorse/tasks; wait for succeeded before reading video_url, then check the characters, actions, and audio.
character1 and character2 are defined by the first two character reference images. They meet in the bookstore scene provided in the third image and exchange a book, with simple actions.
Ensure the reference image order matches the character references, and choose an appropriate aspect ratio; check whether the characters are swapped, as well as hand movements and the background. Complex continuous stories still require storyboards.
This model uses reference images to guide characters and visual elements, and should not be used as a fixed first-frame workflow. If you need an image with a specific composition as the opening frame, choose happyhorse-1.1-i2v. r2v is better suited for preserving the reference subject while rearranging the environment, actions, and camera through text.
Put 1–9 image URLs in image_urls, and reference them in order in the prompt using character1, character2, and so on. In addition to specifying the corresponding assets, clearly describe the subject's actions, scene, and the purpose of each asset. Avoid merely listing names without explaining their relationships.
Set action to reference_to_video, model to happyhorse-1.1-r2v, and submit prompt and image_urls. Do not rely on the default text-to-video action. Then choose resolution, ratio, and duration according to your needs to form a complete reference-image generation request.
This model has native audio capabilities, but reference-image generation does not mean preserving the original video's sound, nor does it mean specifying voice-over or lip sync. The origin usage for preserving original audio belongs to video editing tasks; if audio has specific delivery requirements, check the generated result and arrange any necessary audio post-production.
After submitting with async, save the task_id and query the task through /happyhorse/tasks; you can also provide callback_url to receive completion notifications. The result includes the task status and video_url. Statuses may be pending, succeeded, or error. Download the video for inspection after successful completion.
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