How to make AI videos from a sentence or a photo
Turn an idea, a photo or your own footage into short videos with captions, sized for social media. You'll also learn when an AI video needs a label.
Who it’s for: Creators, marketers, teachers and small businesses.

AI can now turn a single sentence into a short video clip, or bring a still photo to life. The results can be impressive, but they're also easy to get wrong. Faces change between shots, and movement can look strange. This course shows you how to describe a video like a director, so you get the clip you planned.
You'll start by making an eight-second clip from one sentence, and you'll learn the usual limits on length, quality and credits. Then you'll structure a prompt as subject, action, setting, camera move and sound. You'll animate a photo while keeping the subject steady, and plan a 30-second ad as four separate shots. You'll also use AI to edit your own footage, removing an object, extending a shot or changing the background.
For social media, you'll add captions, fix the names they get wrong and reframe wide footage for vertical Reels and Shorts. You'll cut three short clips from one long video, keeping only moments that make sense on their own. You'll also learn when platforms require an AI label, and why realistic fakes of real people are off-limits. The course is for creators, marketers, teachers and small businesses.
By the end you’ll be able to…
- Write a video prompt that controls subject, motion, camera and sound
- Animate a still image into a short clip with a stable subject
- Edit, caption and reframe footage for vertical platforms with AI
- Label synthetic content correctly under platform disclosure rules
Short on time? Start with the essentials
These 4 lessons cover the most useful ideas in this course. Each one takes about a minute.
Make an eight-second clip from one sentence
Generate a short text-to-video clip and understand the credit, length and resolution limits.
Describe a shot like a director, not a novelist
Structure a video prompt as subject, action, setting, camera move and sound.
Auto-caption and reframe for Reels and Shorts
Generate captions, correct names and jargon, and reframe horizontal footage to 9:16 with the subject centred.
Label synthetic video the way platforms require
Decide when platform rules require an AI label and avoid realistic fakes of real people or events.
Every lesson in this course
This course has 10 video lessons of about a minute each, from the basics to expert habits. Each one ends with a question, three takeaways and its sources.
Foundations 2 lessons
Make an eight-second clip from one sentenceEssential
Generate a short text-to-video clip and understand the credit, length and resolution limits.
“One sentence in, eight seconds of video out. Here's what it costs you.”
Describe a shot like a director, not a novelistEssential
Structure a video prompt as subject, action, setting, camera move and sound.
“'Cool drone shot of a city' vs a real shot description. Only one works.”
Building skills 4 lessons
Bring a still photo to life
Animate an image into a clip and keep the subject stable by limiting motion and naming what must stay still.
“Your product photo, now slowly turning with steam rising. No camera needed.”
Plan a 30-second ad as four shots
Storyboard a sequence, reuse reference images for continuity, and generate shot by shot instead of in one prompt.
“Nobody generates a 30-second ad in one go. Build it from four shots.”
Remove, extend or restyle your own footage
Use AI editing to remove an object, extend a shot or change a background in footage you filmed.
“A stranger walks through your best take. Remove them instead of reshooting.”
Auto-caption and reframe for Reels and ShortsEssential
Generate captions, correct names and jargon, and reframe horizontal footage to 9:16 with the subject centred.
“Lots of people scroll with the sound off. Captions aren't optional.”
Advanced 3 lessons
Make an avatar explainer without faking a person
Create a talking-avatar or AI-voiced explainer using your own likeness or a stock avatar, with consent and a clear label.
“A presenter who never needs filming, and must never pretend to be real.”
Cut three shorts from one long video
Use an AI clipping tool to find candidate moments in a long video, keep only those that make sense out of context, and fix their captions.
“A 20-minute talk holds three good shorts. AI finds candidates; you pick.”
Label synthetic video the way platforms requireEssential
Decide when platform rules require an AI label and avoid realistic fakes of real people or events.
“YouTube asks one question when you upload. Get it wrong and it may label it for you.”
Expert 1 lessons
Keep characters and style steady across a series
Build a reusable pack of reference stills, prompt templates and colour grade so an episodic series stays on-model.
“Episode 1 and episode 9 should look like the same show. Most AI series don't.”
Apps we recommend
These are the tools this course uses, and what each one is best for. We have no partnership with any of them.
Google Flow
Best for: Cinematic clips with native soundGoogle's creative studio for Veo video and Nano Banana images: build scenes from ingredients, extend shots and edit objects.
FreemiumRunway
Best for: Editing control and image-to-videoGen-4.5 generation plus tools to restyle, extend and edit existing footage; a long-standing pick for creative teams.
FreemiumCapCut
Best for: Fast short-form edits and auto-captionsPhone-first editor with auto-captions, templates, background removal and AI tools built for Reels, Shorts and TikTok.
FreemiumMake a 20 to 30 second video for a real purpose
Pick a real use: a product teaser, class explainer or club announcement. Storyboard 3 or 4 shots, generate or film each one, then edit, caption and reframe for the platform and apply its AI-label rule. Use your own footage, products or stock avatars, and never a realistic likeness of a real person without consent.
What you’ll make: A captioned 20 to 30 second video in the right aspect ratio, its shot list, and a note of the AI label you applied and why.
Check your own work
- Each shot matches a line in the shot list naming subject, action and camera move
- The main subject or product looks the same in every shot
- Captions are correct, including names, numbers and jargon
- No real person is shown or voiced without documented consent
- The AI-label decision follows the platform's current rule, with a link to that rule
Try a ready-made prompt
Copy a better prompt for a real task, and see why it works better than the version most people type.
Further reading and viewing
We’ve chosen these articles, research papers, videos and guides from other trusted sources.
Video generation prompt guide
Google's breakdown of subject, action, style, camera and audio cues for Veo prompts.
Learn about Google Flow models & supported features
Which Veo and image models Flow offers, what each costs and what each can do.
Runway Academy
Free tutorials on generating, extending and editing video with Runway's tools.
Disclosing use of GenAI content
When YouTube requires you to label realistic AI content, and what happens if you don't.
Questions about this course
What is the best AI video generator?
It depends on what you're making. Google's Flow uses its Veo video models, Runway is popular with creative teams, and CapCut is a phone-first editor with AI captions and templates. Tools change quickly, so the course teaches skills that carry across them. Those include writing a clear shot description, keeping characters steady and checking each tool's terms.
Can AI turn a photo into a video?
Yes. Many AI video tools can animate a still image into a short clip. The trick is to limit the motion and name what must stay still, such as a face or a product label. Otherwise the subject can warp or drift. Only animate photos you have the right to use, and never a real person's likeness without their consent.
Do I have to label AI-generated videos?
Often, yes. Major platforms such as YouTube, TikTok and Instagram have rules about labelling realistic content that AI made or changed. The details differ and change over time, so check each platform's current policy before you post. As a rule, label anything that could be mistaken for real footage, and never fake real people or events.
How do I add captions to a video automatically?
Most editing apps, including CapCut, can generate captions from the speech in your video. They're quick but not perfect. Names, jargon and numbers often come out wrong, so read every caption before you post. The course also shows you how to reframe wide footage to a vertical shape for Reels and Shorts, keeping the speaker in the centre.
How long does each lesson take?
Each lesson takes about a minute. You watch a short captioned video, answer one question and get three takeaways with sources. The course goes from a first clip all the way to keeping characters and style steady across a whole series. Your first five lessons in Tau are free, and Tau Pro opens every lesson in every course.
Get early access to this course
Tau opens soon. Join the waitlist and we’ll email you as soon as you can start this course.
We’ll only email you about Tau, and you can unsubscribe any time. Privacy policy.
We’ll email you as soon as Tau opens.


