Gemini Omni Flash is Google's multimodal video generator that turns text, images, audio, and video into short clips with physics-aware motion. It is the replacement for Veo — specializing in video editing but also capable of generating videos from text and/or references.
Omni is first-and-foremost a video editor — it excels at video-to-video generations.
Variants at a Glance
- Google Omni Flash: Video generation model. Accepts text, image, and video reference inputs. Max resolution: 720p. Talking point: Accepts video and image as inputs.
Key Features
-
Audio & Output
- Native audio: Omni natively creates audio (such as perfectly synced lip movement and environmental footsteps) alongside the video.
-
Inputs
- Two input modes: Text-to-video; image-to-video (note: audio-to-video and reference-to-video coming soon).
- Multimodal references: Combine up to 5 images and 1 video clip in one generation; reference them in-prompt as @Image1 / @Video1 or conversationally as "use the first image as..."
-
Video Edit
- Director-level camera control: Switch camera style to handheld movement, change camera positions and angles, apply cinematic moves like dolly zooms and tracking shots.
- Motion and transitions: Smooth cuts between clips, complex multi-shot scenes with consistent subjects and lighting.
-
Advanced Controls
- Negative prompts: Specify what you don't want in the output to refine quality.
- Seed control: Reproduce specific outputs for consistency across variations.
Technical Capabilities
- Modalities: Text-to-video, Reference-to-video (image/video references)
- Maximum Resolution: 720p
- Aspect Ratios: 16:9, 9:16
- Duration: 3–10 seconds per generation
- Audio: Native audio generation included
- Max Reference Images: Up to 5 images
- Max Reference Videos: 1 video clip
- Credit cost: 150 credits per second (audio included)
Best Use Cases
Professional Video Production: Create cinematic video content with advanced camera control, multi-shot storytelling, and precise motion handling for commercial, advertising, and narrative projects.
Video Editing & Transformation: Edit existing footage to change style, lighting, or narrative while maintaining original motion and structure. Perfect for rapid iterations and creative variations.
Multi-Language Content: Generate videos with consistent characters and styling across multiple language variations using reference images and multimodal inputs.
Strengths
- Physics-aware motion that feels natural and believable
- Director-level camera control for cinematic results
- Native audio sync with lip-sync and environmental sounds
- Strong multi-shot generation with consistent subjects
- Flexible multimodal reference system (up to 5 images + 1 video)
- Excellent for video editing and transformation workflows
Limitations
- Resolution capped at 720p — no 1080p or 4K output.
- Coming soon features: Audio-to-video and reference-to-video inputs are in development.
- Duration: Limited to 3–10 seconds per generation.
Prompting Tips
- Use references: Upload reference images to define character appearance and style consistency.
- Describe camera movement: Specify camera style (handheld, steady, tracking shot) for cinematic control.
- Layer instructions: Combine text prompts with image references for maximum control and consistency.
- Negative prompts: Use negative prompts to exclude unwanted elements and refine output quality.
- Seed control: Lock the seed value to reproduce specific variations or maintain consistency across related clips.