Prompts Built for Omni 1.1
Go beyond generic AI video prompts with examples for scene extension, keyframe transitions, image-to-video animation, reference-led motion, and conversational video edits.
Copy 12 practical Gemini Omni 1.1 Flash prompts for AI video generation and editing. Learn how to extend a scene, animate an image, connect first and last frames, direct cinematic camera movement, preserve characters, add audio, and create polished product videos. Choose an example, replace the details in brackets, and test your idea on Media.io.
Start with a template, change one variable at a time, and refine the result.
Go beyond generic AI video prompts with examples for scene extension, keyframe transitions, image-to-video animation, reference-led motion, and conversational video edits.
Direct push-ins, pull-outs, tracking shots, whip pans, orbit shots, dolly zooms, and seamless loops while keeping subject movement separate from camera movement.
Write from text, an image, a video clip, or first and last frames. Each template explains what to preserve, what to change, and how the shot should progress.
Reuse the prompt structure, swap the subject or setting, and vary one instruction at a time. This makes it easier to diagnose motion, continuity, framing, or audio problems.
Start with text, an image, a source clip, or two keyframes. Decide whether you want to generate a new shot, extend a scene, connect frames, reproduce reference motion, or edit an existing video.
Paste a Gemini Omni 1.1 Flash prompt, then specify what must stay unchanged, what should move, how the camera behaves, the timing, visual style, dialogue, sound, and any unwanted changes to avoid.
Generate a draft and check continuity, subject identity, framing, motion, and audio. If something misses, revise that instruction alone instead of adding more unrelated detail, then export the version that works.
A strong Gemini Omni 1.1 Flash video prompt separates the source context from the requested change. State what must remain consistent, what happens over time, and how the camera and audio behave. Use the formula below, but include only the details that affect the shot.
“Continue the video and make it cinematic.”
“Continue from the uploaded clip's final frame. Keep the same woman, red coat, snowy Tokyo street, nighttime lighting, and realistic style. She walks toward the camera as it tracks backward at her pace; storefront reflections move naturally across the wet pavement. Use one unbroken shot with no identity change or jump cut. Add soft footsteps, light wind, and distant traffic.”
Continue from the uploaded video's final frame. Keep the same woman, silver hair, white dress, fantasy temple, sunlight direction, and painted cinematic style. She turns slowly and steps forward as doves cross the foreground and loose feathers drift through the light. The camera makes a gentle push-in and slight upward tilt. One continuous shot, stable identity, no jump cuts; soft wind, wings, and restrained orchestral music.
Try This PromptUse the first uploaded image as the opening frame and the second as the final frame. Move smoothly from a close-up of the runner's shoes to the wide rooftop finish pose. Create one continuous low-angle tracking shot with natural acceleration, consistent clothing and facial identity, realistic city parallax, and no cuts, warping, or newly added objects.
Try This PromptAnimate the uploaded portrait without changing the person's face, hairstyle, outfit, body proportions, background composition, or photographic style. The subject looks toward the window, blinks once, then gives a subtle smile. Add slight breathing and natural hair movement. Use a slow push-in, soft golden-hour light, clean skin texture, and no exaggerated expression or lip movement.
Try This PromptUse the uploaded dance clip only as a reference for movement and timing. Apply the same choreography to the character in the uploaded image while preserving the character's face, outfit, proportions, and studio setting. Match the footwork, pauses, and arm movements in one full-body shot. Do not show the source dancer or copy the source background in the final video.
Try This PromptFreeze the character and all airborne debris at the current instant. Move the camera in a smooth, complete 360-degree orbit around the motionless character, revealing realistic depth and parallax across the street. Keep the character centered and identical from every angle. Maintain the existing light and color grade; no scene cuts, pose changes, duplicated objects, or camera shake.
Try This PromptContinue the scene in one unbroken shot. The camera dollies steadily toward the shocked character while zooming out so the face remains the same size in frame. Stretch the corridor's perspective naturally behind them without changing the architecture, lighting, expression, or body position. End on a stable medium close-up with a low cinematic rumble and no dialogue.
Try This PromptUse the uploaded product image as the exact visual reference. Preserve the product's shape, label, logo, colors, proportions, and surface details. Rotate it slowly on a glossy black plinth while a narrow highlight travels across the material and the camera makes a subtle orbit. Premium studio commercial, crisp reflections, controlled pacing, no label distortion, added text, or design changes.
Try This PromptA frosted perfume bottle stands on wet dark stone as a ribbon of mist passes behind it. Begin with a macro view of condensation, then glide left to reveal the complete bottle. Warm highlights move across the glass without changing the label. Add one soft spray at 0:02, a delicate glass chime at 0:06, and quiet ambient music. End on a clean hero frame.
Try This PromptEdit the uploaded video without changing the people, dialogue, actions, camera movement, set, duration, or framing. Gradually replace the flat midday light with warm sunset light from frame left. Add longer natural shadows and a subtle golden reflection on the windows. Keep skin tones realistic and preserve all original audio. Make no other visual or timing changes.
Try This PromptContinue the existing conversation from its last frame. Keep both speakers, voices, wardrobe, eyelines, room layout, camera height, and ambient sound consistent. The woman pauses, looks at the man, and says, “Then we take the other road.” He nods silently. Use a slow pullback in one continuous shot; natural lip sync, restrained expressions, no extra speakers, subtitles, or background music.
Try This PromptCreate a 9:16 social video from the uploaded full-body photo. Keep the person's identity, pose, body proportions, and room unchanged. Their casual outfit transforms smoothly into a silver futuristic look as light travels from shoes to shoulders. Push in during the change, then pull back for the final reveal. No body warping, extra limbs, logo changes, cuts, or on-screen text.
Try This PromptCreate a seamless loop of a paper boat floating around a small circular fountain at night. The camera remains locked while the boat completes exactly one orbit and returns to its opening position, orientation, and speed. Keep the water ripples, reflections, and lantern light continuous across the loop point. Gentle water ambience only; no cuts, new objects, zoom, or lighting shifts.
Try This PromptUse Input and Goal + Keep + Change + Camera + Timing + Style + Audio + Constraints. For an existing image or video, start with the details that must remain unchanged. Then describe the action and camera separately. Finish with audio cues and exclusions such as “one continuous shot” or “no identity change.”
Use enough detail to remove ambiguity, but avoid unrelated style keywords and conflicting actions. Two to five focused sentences can define continuity, action, camera, sound, and constraints. If a draft misses the target, revise the failed instruction rather than making the entire prompt longer.
Identify the source-image details to preserve: facial identity, outfit, product design, composition, background, and visual style. Then describe the subject's motion, camera path, scene duration, and sound. Add exclusions for common failures, such as unwanted cuts, extra objects, label distortion, or exaggerated movement.
Begin with “Continue from the uploaded video's final frame,” then list the characters, location, lighting, style, voices, and spatial relationships to retain. Describe the next action and camera movement as a natural continuation. Request one unbroken shot when you want to avoid a visible reset or jump cut.
Yes. Supply the opening and ending images, then explain the movement that should connect them. State which identity, objects, environment, and style must remain consistent. A clear camera path and “one continuous shot” constraint can help distinguish a smooth transition from an unrelated sequence of cuts.
Yes. Use an input image and tell the model to preserve the product's silhouette, proportions, label, logo, colors, and material. Then specify rotation, camera movement, studio lighting, reflections, audio, and the final hero frame. Explicitly exclude label distortion and unauthorized design changes.
Yes. Write dialogue in quotation marks, identify the speaker, and describe the delivery. You can also request ambience, music, and timed sound effects. Keep the sound plan simple and state what not to add—for example, “preserve the original dialogue” or “no music and no subtitles.”
Copy a template from this page, replace its subject and scene details, and test it in the available AI video workflow on Media.io. Match the prompt to your input type—text, image, source video, or keyframes—and review feature availability before relying on a model-specific control.