AIREITER
HappyHorse 1.0 AI Video Generation and Editing
HappyHorse 1.0, also written as Happy Horse 1.0, combines four distinct workflows in one AIReiter model page. Start from text, animate one image, guide a scene with ordered references, or edit an existing video with a direct instruction.
Four workflows on one page
Switch among text-to-video, image-to-video, reference-to-video, and video editing without changing the AIReiter model ID.
Flexible delivery formats
Generation workflows support 720p or 1080p output and whole-second durations from 3 to 15 seconds.
Prompt-controlled editing
Use one source video and a focused instruction to change a subject, wardrobe, style, object, or background while retaining the source composition.
Playground and API parity
The controls shown in the playground map to the same happyhorse_1_0 parameters accepted by the asynchronous API.
AIReiter-Tested HappyHorse 1.0 Results
These are real outputs generated through the AIReiter test environment on August 20, 2026. They document the prompt, request settings, delivered file, and the relationship between the original clip and the edited result.
Text-to-video test: paper airplane in a white studio
Request: 720p, 16:9, 5 seconds, random seed.
Delivered file: H.264 MP4, 1280x720, 5.163 seconds, with an AAC audio track.
Observed task time: about 87 seconds end to end in this test. Queue time can vary.
Video-edit test: change the airplane from red to blue
Instruction: Change the red paper airplane to blue. Keep the original background and camera movement.
Request: 720p with keep_original_sound disabled. No duration or aspect-ratio field was sent because video editing derives both from the source.
Delivered file: H.264 MP4, 1280x720, 5.163 seconds, with an AAC audio track.
Which HappyHorse 1.0 Workflow Should You Use?
Choose the workflow from the material you already have and the part of the result that must remain consistent.
Start with text
Use text-to-video when the scene does not need to match an existing subject. Describe action first, followed by camera behavior, lighting, environment, and style.
Animate one image
Use image-to-video when one still image defines the opening frame and subject. The output aspect ratio follows the uploaded image.
Coordinate several references
Use reference-to-video when people, props, clothing, products, or locations must come from one to nine ordered images.
Modify existing motion
Use video editing when the source timing and camera movement should remain recognizable while a specific visual element changes.
Practical HappyHorse 1.0 Use Cases
Product and advertising variations
Generate a product scene from text or reference images, then create color, background, or styling variations through video editing.
Social video in multiple formats
Create horizontal, vertical, square, standard, or cinematic clips for different publishing channels without rewriting the core concept.
Character and prop continuity
Provide ordered reference images when a person, outfit, object, or environment must remain recognizable across the generated clip.
Focused post-production changes
Replace one visual element or restyle a scene without rebuilding the source timing and camera movement from scratch.
HappyHorse 1.0 Parameters and Mode Limits
The form changes with the selected workflow so unsupported fields are not submitted or included in pre-generation billing.
Resolution
Choose 720p for standard output or 1080p when higher visual detail is required. The selected tier affects price.
Generated duration
Text-to-video, image-to-video, and reference-to-video accept any whole-second duration from 3 to 15 seconds.
Aspect ratio
Text-to-video and reference-to-video support nine ratios: 16:9, 9:16, 1:1, 4:3, 3:4, 4:5, 5:4, 9:21, and 21:9. Image-to-video follows the source image.
Video-edit duration and framing
Video editing does not accept video_length or aspect_ratio. Output duration is determined from the source and capped at 15 seconds; framing follows the source video.
Source audio control
For video editing, enable keep_original_sound to force retention of the source track. When disabled, the model decides based on context.
Seed
Use a positive integer from 1 to 2147483647 for more repeatable iterations, or omit the field for a random seed.
How to Get Better HappyHorse 1.0 Results
A compact prompt and clean source media usually provide more control than a long list of conflicting instructions.
Lead with the main action
State who or what moves and what should happen before adding atmosphere, visual style, or secondary details.
Describe the camera deliberately
Use one compatible camera instruction such as fixed shot, slow push-in, tracking shot, or gentle orbit instead of stacking contradictory movements.
Use clear source media
Choose sharp, publicly accessible images and videos with an obvious subject. Similar reference-image proportions reduce composition conflicts.
Change one variable per iteration
Keep the prompt and seed stable while testing resolution, framing, or one visual change so differences are easier to evaluate.
HappyHorse 1.0 vs HappyHorse 1.1
The two AIReiter model IDs solve different production problems rather than acting as simple quality tiers.
Choose HappyHorse 1.0 for workflow range
Use happyhorse_1_0 when you need text generation, one-image animation, multi-reference generation, or source-video editing from one model page.
Choose HappyHorse 1.1 for reference-first work
Use happyhorse_1_1 when the job centers on one to nine ordered images and explicit [Image N] references in the prompt.
Need to edit an existing video?
Choose HappyHorse 1.0. The current HappyHorse 1.1 integration is reference-to-video only and does not accept a source video.
Need stricter reference organization?
Choose HappyHorse 1.1 when each uploaded image must be named directly in the prompt as a character, object, wardrobe item, prop, or scene.
Use HappyHorse 1.0 Through the AIReiter API
AIReiter exposes all four HappyHorse 1.0 workflows through one asynchronous task contract and validates mode-specific parameters before task creation.
Model ID
Use happyhorse_1_0 as the model value and select the workflow with params.type.
Submit endpoint
POST /api/openapi/submit accepts the model, workflow parameters, and a unique out_task_id.
Query endpoint
POST /api/openapi/query accepts the same out_task_id. Poll until the task is completed or failed.
Strict mode parameters
For video_edit, do not send video_length or aspect_ratio. AIReiter rejects these unsupported fields before task creation and billing.
Video-edit billing basis
Video editing is billed from source-video duration plus generated output duration at the selected resolution tier. The source determines output length, capped at 15 seconds.
Related AI Video Models
HappyHorse 1.1
Use up to nine reference images for identity, object, and scene consistency in a dedicated reference-to-video workflow.
Gemini Omni Flash
A flexible video model with text, image, reference, and editing workflows through one model page.
Veo 3.1
A general-purpose video generation model for cinematic text-to-video and image-to-video output.
HappyHorse 1.0 FAQ
Answers about choosing a workflow, source media, duration, aspect ratio, video editing, billing, and the difference between HappyHorse 1.0 and 1.1.
/ 01Which video workflows does HappyHorse 1.0 support?
AIReiter exposes text-to-video, image-to-video, reference-to-video, and video-edit workflows through the happyhorse_1_0 model.
/ 02How many reference images can I use?
Reference-to-video supports one to nine ordered images. Video editing supports up to five optional reference images in addition to one source video.
/ 03Can I select video-edit duration?
No. Video editing does not support video_length. The model sets output duration from the source duration, capped at 15 seconds.
/ 04Can I set a video-edit aspect ratio?
No. Video editing follows the source-video aspect ratio. Requests that explicitly include aspect_ratio for video_edit are rejected.
/ 05How is video editing billed?
The billing basis is input video duration plus output video duration, multiplied by the selected resolution rate.
/ 06What is the difference between HappyHorse 1.0 and 1.1?
HappyHorse 1.0 combines four generation and editing workflows. The current HappyHorse 1.1 integration specializes in reference-to-video with one to nine ordered images and explicit [Image N] prompt markers.
/ 07How should I write a video-edit instruction?
Name the element to change, the desired replacement, and what must remain unchanged. Avoid unrelated scene directions when preserving source motion is important.
/ 08Which API model ID should I use?
Use happyhorse_1_0 with POST /api/openapi/submit, then poll POST /api/openapi/query using the same out_task_id.