Google Flow AI Tutorial for Beginners | Easy Step-by-Step Guide to Create Stunning Images & Videos in 2026
Last Updated: April 2026 | 12 min read | By WhiskAILabs Editorial Team
Opening Google Flow AI for the first time feels like walking into a room that’s bigger than you expected.
There’s a lot there. Image generation, video clips, editing tools, a timeline, an ingredients system, audio controls — and no obvious “start here” sign anywhere. If you came from Whisk AI, the simplicity you were used to is gone. If you’re brand new to AI creative tools, the options can feel overwhelming before you’ve made a single thing.
This tutorial fixes that. We’re going to walk through everything step by step — how to get in, what each section does, how to generate your first image, how to turn it into a video, and the specific prompt habits that separate good results from great ones.
By the end of this page, you’ll have made something. That’s the whole point.
How to Access Google Flow AI {#access}
Before anything else — let’s make sure you can actually get in.
Step 1 — Open Google Flow AI
Go to labs.google/flow in your browser. Chrome works best. If you’re using Firefox, Edge, or Safari and something looks broken, switching to Chrome usually fixes it.
Step 2 — Sign in with your Google account
Click “Sign in” and use any Google account. Gmail accounts work. Google Workspace accounts work. The account needs to belong to someone who is 18 or older — Flow applies the same age requirement that Whisk did.
Step 3 — Accept the terms
If this is your first time using Google Labs tools, you’ll see a terms of service page. Read through it and accept to continue.
Step 4 — You’re in
The main Flow workspace will load. It looks more complex than Whisk did, and that’s because it is. Don’t let that slow you down — we’ll walk through it section by section.
If you see “not available in your country”: Flow is available in 149+ countries, but some regions — particularly EU countries and the UK — face access restrictions. A VPN connected to a US server resolves this in most cases. Our access guide covers the exact method.
Understanding the Interface {#interface}
When you first open Google Flow AI, you’ll see a few main areas. Here’s what each one does.
The Creation Panel (Left Side)
This is where you tell Flow what you want to make. You’ll see options for:
Generate Image — Describe an image in words and Flow produces it using Imagen.
Generate Video — Describe a video clip and Flow produces it using Veo 3.1.
Ingredients to Video — Upload reference images for your subject, scene, and style, then generate a video incorporating them. This is the closest equivalent to Whisk’s three-image input system.
Frames to Video — Give Flow a starting image and an ending image and it generates the video transition between them.
The Canvas / Preview Area (Center)
This is where your generated content appears. After each generation, the result shows up here. You can view it, download it, or send it to the timeline for editing.
The Library (Right Side or Top Menu)
Your Library stores everything you’ve generated. Images, video clips, ingredients, full projects — it all lives here. Think of it as your personal project folder inside Flow.
The Timeline (Bottom)
When you’re building a longer video project, the Timeline is where you arrange your clips in sequence. Drag clips from your Library into the timeline, reorder them, and preview the full sequence.
Credits Display
Somewhere in the interface — usually top right — you’ll see your remaining daily credits. Free accounts get 50 daily credits for video generation (180 in some countries). Image generation has separate limits. Credits reset at midnight.
Your First Image Generation with Google Flow AI {#first-image}
Let’s make your first image. This is the simplest thing you can do in Flow and a good way to get comfortable with how it responds to descriptions.
Step 1 — Click “Generate Image”
In the creation panel, select the image generation option.
Step 2 — Write your description
A text box appears. Type a description of what you want. Start simple.
Good first prompt example:
A golden retriever puppy sitting in a sunlit garden, soft morning light,
shallow depth of field, warm colours, photorealistic
Step 3 — Click Generate
Flow sends your description to Imagen and produces the image. This typically takes 10–20 seconds.
Step 4 — Review the result
Look at what came back. If you like it — download it or save it to your Library. If something’s off — maybe the lighting feels wrong, or the background isn’t what you wanted — you have two options:
Option A: Adjust your description and regenerate. Add or change specific details.
Option B: Use the edit tools (covered below) to modify specific parts of the existing image without regenerating the whole thing.
Step 5 — Try a variation
Hit generate again with the same description. Flow produces a different interpretation each time. Running 3–4 variations of the same prompt quickly shows you the range of what’s possible.
Your First Video Clip {#first-video}
This is where Google Flow AI goes beyond what Whisk ever offered.
Step 1 — Click “Generate Video”
Select the video generation option in the creation panel.
Step 2 — Write your video description
Video prompts work best when they describe motion, lighting, and scene clearly. Think like a film director — you’re describing a shot, not just a picture.
Good first video prompt:
A tabby cat sitting on a wooden windowsill, rain falling outside,
warm amber indoor light, the cat looks out the window calmly,
soft cinematic focus, gentle ambient sound of rain
Step 3 — Choose your settings
You’ll see options for aspect ratio (landscape, portrait, square) and generation quality. For a first attempt, leave these at their defaults.
Step 4 — Click Generate
Video generation takes longer than image generation — typically 30–90 seconds depending on server load and your connection. This is normal.
Step 5 — Watch the result
Your clip will be up to 8 seconds long. It will include audio synchronized to what’s happening in the video — in the cat example above, you’d hear rain ambient sound and possibly soft movement.
Step 6 — Download or add to timeline
If you’re happy with the clip, download it directly. If you want to use it as part of a longer sequence, drag it into the timeline.
The Ingredients System {#ingredients}
This is one of the most powerful features in Google Flow AI — and the one that most resembles what Whisk did with its three-image input.
The Ingredients system lets you define visual assets — a specific character, a specific object, a specific style — and use them consistently across multiple generations.
How to create an Ingredient:
Upload any image to your Library. Then designate it as an Ingredient — you’ll see this option when you click on any image in your Library. Give it a name (e.g., “Main Character,” “Product Shot,” “Background Style”).
How to use Ingredients:
When generating a new image or video, you’ll see an option to “Add Ingredient.” Select the ones you want to include. Flow will use those images as visual references — keeping your character’s appearance, your product’s design, or your style reference consistent across multiple generations.
Why this matters:
Whisk generated a different interpretation every time. If you wanted consistent results — same character design across ten images, or a product shown in multiple settings while looking like the same product — Whisk couldn’t reliably do it. Flow’s Ingredients system solves this directly.
Prompt Writing for Google Flow AI tutorial {#prompts}
The biggest difference between good results and mediocre ones in Google Flow AI comes down to how you describe what you want. These habits make a real difference.
For Images — The Four Elements
Strong image prompts in Flow contain four things:
Subject — What is the main thing in the image? A ceramic coffee mug
Setting — Where is it? What’s around it? on a dark wood table in a moody coffee shop
Light — How is it lit? warm low overhead light, soft shadows
Quality signal — What standard should the output match? editorial product photography, shallow depth of field
Put together:
A ceramic coffee mug on a dark wood table in a moody coffee shop,
warm low overhead light, soft shadows, editorial product photography,
shallow depth of field
That’s the complete formula. One subject, one setting, one lighting note, one quality signal.
For Videos — Think in Shots
Video prompts work best when you describe a camera shot, not just a scene.
Instead of: a forest
Try: A slow push-in through a misty forest, morning light filtering through the trees, no people visible, atmospheric and quiet, cinematic wide shot
The difference is the camera instruction — “slow push-in” tells Flow how the frame should move. Without it, Flow picks a camera movement on its own, which is often less interesting than a deliberate choice.
Camera movements that work well:
slow push-in— camera moves gradually toward the subjecttracking shot— camera follows alongside the subjectaerial descent— camera moves downward from abovestatic wide shot— no camera movement, full scene visiblehandheld close-up— slightly shaky, intimate feel
For Audio
Veo 3.1 generates audio automatically. If you want specific sounds, mention them:
ambient rain sounds, busy coffee shop background noise, character speaking softly, wind through leaves, upbeat music
If you don’t mention audio, Flow makes its own choice — which is usually appropriate but not always exactly right.
Editing Tools — Insert, Remove, Extend {#editing}
Once you have an image or video you mostly like, these tools let you refine specific parts without regenerating everything.
Insert
Adds an element to an existing image or video frame.
How to use: Select an image in your Library. Choose “Insert.” Describe what you want to add and roughly where.
Example: You generated a coffee shop interior but want to add a person sitting at a table in the background. Click Insert, describe “a person with short dark hair, sitting at a table in the background, reading, slightly out of focus.” Flow adds them while maintaining the existing lighting and style.
When it works best: Adding single elements with clear descriptions. Simple objects, people in background positions, environmental details.
When it struggles: Very complex additions, exact positioning requirements, adding text that needs to be legible.
Remove
Removes an element from an image or video frame and fills in the background.
How to use: Select an image. Choose “Remove.” Describe or indicate what you want removed.
Example: Your generated outdoor scene has a power line cutting through it. Click Remove, describe “the power line in the upper portion of the image.” Flow removes it and fills in sky behind it.
When it works best: Removing relatively simple elements against consistent backgrounds. Objects that aren’t the main subject.
Extend
Makes a video clip longer than its original duration.
How to use: Select a video clip. Choose “Extend.” Flow generates additional footage that continues naturally from the final second of your clip.
This is how you build sequences longer than 8 seconds. A 60-second finished video requires chaining multiple 8-second clips with Extend — approximately 8 generations total.
Important: Extend works from the final second of your clip. If your clip ends in an awkward moment or mid-motion, the extension may look inconsistent. Try to end clips at natural pause points for the smoothest extensions.
Common Beginner Mistakes {#mistakes}
These are the patterns that consistently produce disappointing results — and how to avoid each one.
Mistake 1 — Prompts That Are Too Vague
"a nice landscape" gives Flow almost nothing to work with. The result will be competent but generic.
Fix: Add specifics. Time of day, weather, a dominant color, a camera angle. "Rolling green hills at golden hour, long shadows, no people, wide establishing shot, cinematic" gives Flow ten times more direction.
Mistake 2 — Too Many Instructions at Once
"a happy smiling woman with long brown hair wearing a red dress standing in a field of sunflowers at sunset with mountains behind her and birds flying overhead and a vintage filter" — that’s six competing instructions. Something will get lost.
Fix: Pick your three most important elements and describe those clearly. Leave the rest out. You can add details in a second generation if you need them.
Mistake 3 — Generating Once and Giving Up
The first generation is never the final one. Flow’s outputs vary between generations even with identical prompts. If your first result isn’t quite right, generate 2–3 more variations before changing anything.
Fix: Run 3–4 generations of every prompt before deciding to change the description.
Mistake 4 — Ignoring the Credit System
Free daily credits don’t roll over. Generating the same prompt fifteen times in a row burns through your daily allowance fast.
Fix: Plan your generations. Describe clearly the first time. Run 3–4 variations. Move on. Come back tomorrow with fresh credits if you need more.
Mistake 5 — Not Saving Work You Might Want Later
Flow’s Library keeps your generations, but the interface can feel temporary. People frequently close a tab without saving something they actually wanted.
Fix: Download anything you like immediately. Don’t rely on Flow’s Library as your only copy.
Frequently Asked Questions {#faq}
Do I need any creative skills to use Google Flow AI?
No. The whole point of tools like Flow is that you communicate in plain language and the AI handles the technical execution. If you can describe what you want in a sentence, you can use Flow. Better results come with practice, but you don’t need to be a designer, photographer, or filmmaker to get started.
How is Google Flow AI different from just using Gemini?
Gemini’s image generation produces still images from text descriptions. Google Flow AI is a full creative studio — it includes video generation with audio, a timeline for sequencing clips, editing tools, and a project management system. Flow is what you use when you want to build something more complex than a single image.
Can I use Google Flow AI outputs for commercial work?
Check Google’s current terms at labs.google/flow/about before using outputs commercially. The terms for Labs tools can change, and commercial use policies vary. For work where commercial licensing certainty is essential, Adobe Firefly remains the safest choice.
What’s the best way to get the Whisk AI sticker or plushie style in Google Flow?
Since Flow doesn’t have those preset buttons, describe the style directly in your image prompt. For sticker style: die-cut vinyl sticker design, bold black outline, white background, flat saturated colors, kawaii aesthetic. For plushie style: soft felt plushie, hand-stitched seam detail, button eyes, even studio lighting, white product background. Our Whisk AI prompts guide has the full list of style prompts that work in text-based tools.
Why does my video generation keep failing or spinning without finishing?
Usually either a slow internet connection or high server load. Try at a different time of day — US evenings and weekends are peak traffic hours. Also check your remaining credits — if you’ve run out of daily credits, generations will fail silently. Check Google’s status page if the problem persists across different times.
Can I make videos longer than 8 seconds?
Yes — using the Extend feature. Each extension adds footage that continues from the final second of your previous clip. A one-minute video requires chaining approximately 8 clips with Extend. Plan your shot list accordingly, and try to end each clip at a natural pause rather than mid-motion for smoother continuity.
That’s everything you need to get started with Google Flow AI from zero.
The tool is more capable than Whisk was. It’s also more complex. The learning curve is real — but the free tier means you can work through that curve without spending anything, and every generation teaches you something about how the tool responds.
For the prompt techniques that carry over from Whisk into Flow, our full AI prompt guide covers the language that works with Imagen-based generation specifically. And for a complete overview of everything in the Google Labs ecosystem that’s active in 2026, the Google Labs AI tools guide has the full picture.
