AI Video Generation
Flowmo's AI Video Generator turns text descriptions into short video clips -- perfect for hero backgrounds, motion graphics, product demos, and social media content. You describe a scene, pick your settings, and the AI produces a ready-to-use video that you can drop directly onto your canvas or timeline.
Like image generation, video generation is available in two places: as a standalone chat assistant ("Video Gen") in the AI sidebar, and as a Video Generator node on the Media Gen flow canvas. This article covers both, plus the Video Scene Picker for importing scenes from existing video assets.
Prerequisites
Before you start generating videos, make sure you have:
- An active Flowmo project open in the designer. Generated videos are saved to your project's media library.
- A signed-in account with a verified email. Video generation requires authentication and email verification.
- A clear description of the scene you want. Video prompts benefit from describing motion, camera angle, mood, and subject.
How It Works
Under the hood, Flowmo uses Veo 3 Fast -- Google's video generation model. The chat assistant uses Veo 3 Fast by default, while the Media Gen canvas gives you a choice between two model tiers:
| Model | Label in UI | Best for |
|---|---|---|
| Veo 3.1 Fast | Veo 3.1 Fast | Faster generation, standard quality |
| Veo 3.1 | Veo 3.1 | Higher fidelity, more detail |
Video generation is asynchronous. After you submit a prompt, the AI starts processing in the background. You can continue working in Flowmo while the video renders. The assistant polls for completion and notifies you when the video is ready.
Method 1: Video Gen Chat Assistant
The Video Gen assistant provides a straightforward chat-based workflow for generating videos from text prompts.
Step 1: Open the AI Sidebar and Select Video Gen
- Open your project in the Flowmo designer.
- Open the AI sidebar on the right side of the screen.
- Click the assistant type dropdown at the top of the chat panel.
- In the popover menu, click Video Generator (shown with a video camera icon).
The assistant header updates to show "Video Gen" as the active mode. The description reads: "Generate professional videos with Veo3 Fast!"
Step 2: Configure Video Settings
Before submitting your prompt, you can adjust the following settings:
- Aspect Ratio: Choose between 16:9 (landscape, default) or 9:16 (portrait/vertical). Use 16:9 for website hero sections and desktop content; use 9:16 for mobile-first or social media content.
- Duration: Choose 4 seconds, 6 seconds, or 8 seconds (default). Shorter durations generate faster.
- Resolution: Choose 720p (default, faster) or 1080p (higher quality).
Step 3: Write Your Prompt
Type a description of the video scene you want. Good video prompts describe:
- Subject and action: What is happening in the scene (a drone flying over a city, water splashing in slow motion, etc.)
- Camera movement: Dolly in, pan left, orbit, static shot, etc.
- Style and mood: Cinematic, documentary, abstract, dreamy, high-contrast, etc.
- Lighting: Golden hour, neon, overcast, dramatic shadows, etc.
Example prompts:
"Slow-motion aerial drone shot flying over a misty mountain range at sunrise, cinematic color grading, golden light"
"Close-up of coffee being poured into a white ceramic mug, steam rising, warm ambient lighting, shallow depth of field"
"Abstract fluid simulation in deep purple and electric blue, smooth camera orbit, dark background"
Press Enter or click the send button to submit.
Step 4: Attach a Starting Image (Optional)
You can attach a single image to use as the first frame of the generated video. This is useful when you want the video to start from a specific composition or match an existing design.
- Click the attachment button in the chat input.
- Upload your starting image.
- Write your prompt describing how the scene should animate from that image.
Note: Only one starting image is supported per generation. If you attach multiple images, the assistant will ask you to remove the extras.
Step 5: Wait for Generation
After submitting, the assistant displays a confirmation message:
"Video Generation Started! Your video is being generated using Veo3 Fast with the following settings..."
The message lists your chosen aspect ratio, duration, resolution, and whether a starting image was included. The assistant polls for completion every 15 seconds (after an initial 10-second delay) and updates you when the video is ready.
Generation typically takes 1 to 5 minutes depending on duration, resolution, and server load.
Step 6: Use Your Generated Video
Once the video is ready, the assistant displays a success message with the video file size and a playable preview. You have three options:
- Add -- Inserts the video as a new element on your canvas.
- Replace -- Replaces the currently selected element's video source with the generated video.
- Download -- Saves the MP4 file to your computer.
The generated video is also automatically uploaded to your project's media library.
Method 2: Media Gen Flow Canvas
The Media Gen canvas provides a node-based workflow for video generation with more granular controls, including video extension and start/end frame inputs.
Opening Media Gen
- In the AI sidebar, click the assistant type dropdown.
- Select Media Gen (Beta) at the bottom of the list.
Video Generator Node
The Video Generator node is the core video creation node on the Media Gen canvas.
Input handles (left side):
- Prompt -- Connect a Text Prompt node with your scene description.
- Start Frame -- Connect an image to use as the video's opening frame.
- End Frame -- Connect an image to use as the video's closing frame.
- Extend -- Connect a previously generated video to extend it with additional footage.
Output handle (right side):
- Video Output -- The generated video, which can be chained to other nodes.
Settings (expand the node to configure):
- Model: Choose between Veo 3.1 Fast or Veo 3.1.
- Duration: 4 sec, 6 sec, or 8 sec.
- Ratio: 16:9 or 9:16.
- Res: 720p or 1080p.
- Generate Audio toggle -- When enabled (shown with a speaker icon), the model generates synchronized audio/sound effects for the video.
Advanced Options (click to expand):
- Person Generation -- Controls whether the model can generate images of people:
- Allow All (default) -- No restrictions.
- Allow Adults Only -- Prevents generation of children.
- Disallow -- No people in the generated video.
- Seed -- Enter a specific seed number (0 to 4,294,967,295) for reproducible results. Leave empty for random generation. Click the dice icon to generate a random seed.
Click Generate Video to start. The button shows a loading spinner while processing.
Video Extension Mode
When you connect an existing video to the Extend input handle, the node automatically switches to Video Extender mode. In this mode:
- The header changes to "Video Extender" with a green accent.
- Aspect ratio and resolution controls are disabled (they inherit from the source video).
- The generate button reads Extend Video instead of "Generate Video."
- A notice appears: "Extension Mode Active. Inheriting [resolution] ([aspect ratio])."
This lets you chain multiple Video Generator nodes together to create longer sequences from a single starting clip.
Using Generated Videos
When a video finishes generating, it appears in the node's output area. Hover over the video preview to see quick-action buttons:
- Add -- Adds the video to your canvas.
- Apply -- Replaces the currently selected element (appears only when you have a canvas element selected).
Click the three-dot menu for additional options: Add to canvas, Paste to Replace, and Download.
Videos play on hover and loop automatically in the node preview.
Video Scene Picker
The Video Scene Picker is a dedicated tool for working with existing video assets. It uses AI analysis to break a video into individual scenes, then lets you select and insert specific scenes onto your timeline.
How to Access the Scene Picker
The Scene Picker opens from the video editor's left panel when you have a video asset with a completed AI analysis. It appears as a modal titled "Scene Picker -- [video name]".
Scene Picker Interface
The Scene Picker has four main areas:
- Video Preview (top-left): A 320x180px video player. When you hover over a scene block in the timeline, the video automatically seeks to that scene and plays it in a loop. A mute/unmute toggle sits in the bottom-right corner.
- Scene Info (top-right): Displays details about the currently hovered scene:
- Scene type badge (color-coded): intro, outro, b-roll, talking-head, transition, content, text-card, or other.
- Time range in mm:ss format.
- Visual description of what happens in the scene.
- Audio description (italic) of dialogue or sound.
- Text on screen if any text appears in the frame.
- Tags for the scene content.
- Toolbar (middle):
- Zoom controls: buttons and a slider to zoom the timeline from 0.5x to 10x. You can also hold Ctrl/Cmd and scroll to zoom.
- Select All / Deselect buttons for batch selection.
- Insert button (activates when scenes are selected) showing the count of selected scenes.
- Scene Timeline (bottom): A horizontally scrollable timeline with a time ruler (showing mm:ss markers) and color-coded scene blocks. Each block is labeled with its scene type and start time. Click a block to select it; Shift-click to select a range. Selected blocks show a checkmark and a highlighted border.
Inserting Scenes
- Select the scenes you want by clicking their blocks on the timeline. Use Shift-click for range selection.
- Click the Insert button in the toolbar. The button shows how many scenes are selected (for example, "Insert (3)").
- The Scene Picker groups adjacent selected scenes into clips (with a 50ms tolerance gap).
- Each clip is created as a full-size video element on your canvas, positioned at the current playhead location on the timeline.
- The timeline duration automatically extends to accommodate all inserted clips.
The footer bar shows a summary: total scene count, video duration, mood, and how many scenes are currently selected.
Tips for Better Results
- Describe motion, not just appearance. Video prompts should explain what happens over time: "camera slowly dollies forward through a foggy forest" works better than "a foggy forest."
- Keep it focused. Short clips (4-8 seconds) work best with a single, clear action or camera movement. Trying to cram multiple scenes into one generation reduces quality.
- Use starting frames for consistency. If you are generating multiple clips for the same project, use a frame from the previous clip as the starting image for the next one to maintain visual continuity.
- Chain with video extension. In Media Gen, connect the output of one Video Generator node to the Extend input of another to build longer sequences while keeping style consistent.
- Use seeds for iteration. When you get a result you almost like, note the seed (or set one manually) and re-generate with a refined prompt. The same seed with a different prompt produces a similar but adjusted result.
- Enable audio when appropriate. The "Generate Audio" toggle in Media Gen adds synchronized sound effects. Enable it for hero sections or product videos where ambient sound adds impact.
Common Issues
| Issue | Solution |
|---|---|
| "Please authenticate to use the Video Generation Assistant." | Sign in to your Flowmo account. |
| "User email must be verified to use video generation." | Check your inbox for a verification email from Flowmo and confirm your email address. |
| "Video generation supports only one starting image." | Remove extra attached images. Only one starting frame is supported per generation. |
| "Duration must be 4, 6, or 8 seconds." | Select a valid duration from the available options. |
| "Aspect ratio must be either 16:9 or 9:16." | Only landscape (16:9) and portrait (9:16) ratios are supported for video. |
| Generation takes a very long time | Video generation can take 1-5 minutes. The assistant polls automatically. If it takes longer than 10 minutes, try generating again with a simpler prompt or lower resolution. |
| Video quality is lower than expected | Try using Veo 3.1 (non-Fast) in Media Gen for higher fidelity. Also try 1080p resolution and a more detailed prompt. |
| Scene Picker shows "Preview unavailable" | The video file may not be accessible. Check that the video URL is valid and the file has not been deleted from your media library. |
Related Articles
- Getting Started with Vibe Mode -- Overview of the AI-powered design workspace.
- AI Image Generation -- Generate and edit images with AI.
- AI Smart Router -- How the Main Agent automatically routes your requests to the right assistant.