Generate 4, 6, 8, or 10-second videos from text, frames, images, or a short source clip at resolutions from 360P to 4K.
Version 1.1 is selected by default. Choose text, first and last frames, image references, or a source video, then set a supported duration, resolution, and aspect ratio.
Open the generator when you are ready to upload a reference or write a prompt.
Open generatorThese prompt patterns show how to structure frame animation and reference-led generation before you create.
Set the opening and closing composition, then describe the camera path and the action that should connect them.
Move from the clean opening frame into a neon-lit future city while keeping the orange robot recognizable. Use a slow push-in, natural reflections, and a stable final composition.
Use reference images to preserve the subject while the prompt defines expression, movement, setting, and camera behavior.
Keep the reference character consistent. Add expressive but controlled motion, soft cel shading, a clean silhouette, and a gentle cinematic camera move.
The 1.1 model combines several input paths, but each path has its own rules. Pick the workflow that matches the material you already have.
Describe the subject, action, setting, camera, and visual treatment when you want to create without source media.
Animate one opening frame or provide both endpoints when the beginning and ending composition matter.
Guide identity, style, composition, or motion with image references and at most one short source video in the reference workflow.
Choose 360P, 720P, 1080P, or 4K, with landscape and portrait aspect-ratio options.
The generator is configured around the current Kie model contract rather than the legacy Omni model.
The selected provider model is google/gemini-omni-flash-1-1.
Output durations are 4, 6, 8, or 10 seconds, with 360P, 720P, 1080P, and 4K options.
Supported aspect ratios are 16:9 and 9:16.
Version 1.1 is useful when input control matters as much as the prompt.
Connect approved opening and closing frames for pitches, shot planning, and visual preproduction.
Use image references to keep a recognizable subject while exploring new settings and motion.
Start from an existing short clip when you need the generated result to follow source motion or composition.
Choose one input path, describe the intended motion clearly, and confirm the output settings before generation.
Start with text, one or two frames, reference images, or a source video. Frame input is kept separate from the multimodal reference workflow.
Choose 4, 6, 8, or 10 seconds, then select 16:9 or 9:16 and a resolution from 360P to 4K.
The workspace shows the required vid2vid credits before the request is sent to the 1.1 model.
Answers about inputs, output settings, reference limits, and billing on the dedicated 1.1 page.
Open the dedicated generator with version 1.1 selected and start from text, frames, references, or a source clip.
