Ijharul Islam
To make a video from text with AI, decide first whether you want short generated clips or a longer narrated video. For clips, write a prompt that describes one shot (subject, action, setting, camera, light and sound), pick a text-to-video model such as Veo 3.1 or Kling v3, set the length and shape, and generate. For a narrated video, paste a script into a script-to-video tool that adds stock footage, a voiceover and captions.
This guide walks through both, with a worked example of turning a few lines of text into a 30-second ad, the settings that matter, and what it costs.
The short answer
Split your text into shots, write one prompt per shot (subject, action, setting, camera, light, sound), generate each shot as a 4 to 15 second clip on a text-to-video model, then join the clips in an editor and add voiceover, music and captions. For narrated explainers longer than a minute, a script-to-video tool that uses stock footage is faster and cheaper.
Key takeaways
- Text-to-video models make one shot of 2 to 15 seconds per prompt; longer videos are several shots joined.
- Write prompts as shot descriptions, not as a summary of the whole video.
- For explainers of 1 minute or more, a script-to-video tool with stock footage and a voiceover usually fits better.
- Draft on a cheap model, then render the final on a premium one.
- A 30-second ad is about 4 to 6 shots.
Disclosure: we make QuickVeed and show how to do this in it, along with other ways. Details about other products come from their public pages, checked in October 2026 and linked at the end.
Two ways to make a video from text
“Text to video” covers two quite different jobs, and choosing the right one first saves time and money.
| Generated clips (text-to-video model) | Narrated video (script-to-video) | |
|---|---|---|
| Input | A prompt describing one shot | A script or a topic |
| Output | A new 2-15 s clip, made from scratch | A 1-30 minute video built from stock footage |
| Sound | Native audio on some models, or none | Voiceover, often with music and captions |
| Best for | Ads, hooks, product shots, B-roll, scenes you can't film | Explainers, faceless YouTube, training, listicles |
| Cost driver | Seconds of generated video | Minutes of finished video |
| Examples | Veo 3.1, Kling v3, Seedance 2.0, Runway Gen-4.5 | Pictory, InVideo AI, Fliki |
Many good videos use both: generated clips for the opening hook and the moments stock footage can't show, and a narrated structure for the rest. For the narration itself, see choosing a text-to-speech voice for YouTube. The rest of this guide focuses on generated clips.
Step 1: Turn your text into a shot list
A text-to-video model makes one shot at a time, so the first job is to break your text into shots. Read the text and, for each sentence or idea, ask: what would the viewer see?
| Your text | Shot | Length |
|---|---|---|
| “Mornings are rushed.” | Close-up of a hand slapping an alarm clock, then a blurry kitchen | 4 s |
| “Our cold brew is ready in the fridge.” | Fridge door opens, a glass bottle of cold brew in warm light | 4 s |
| “Pour, sip, go.” | Coffee poured over ice, slow motion | 4 s |
| “Order by Friday for free delivery.” | Bottle on a doorstep, a front door opens | 6 s |
Keep each shot to one action. If a sentence needs two actions, it needs two shots. Mark which shots are essential and which are nice to have; if credits or time run short, you can drop the extras and still have a complete video. Write the shot list in a simple document with three columns (text, shot, length) so you can paste prompts straight from it.
Step 2: Write a prompt for each shot
A good text-to-video prompt answers six questions in this order:
- Subject: who or what is in the shot? “A glass bottle of cold brew with a kraft label”
- Action: what happens, and how fast? “is lifted slowly out of the fridge”
- Setting: where and when? “in a bright kitchen in the morning”
- Camera: framing and movement. “close-up, slow push in, shallow depth of field”
- Light and style: “warm sunlight, soft shadows, photorealistic”
- Sound (audio models only): “the fridge hums, a bottle clinks”
Together: “Close-up of a glass bottle of cold brew with a kraft label being lifted slowly out of a fridge in a bright kitchen in the morning. Slow push in, shallow depth of field, warm sunlight, soft shadows, photorealistic. The fridge hums and the bottle clinks.”
To keep a product or character the same across shots, paste the same description into every prompt, or generate one good still and use image-to-video from it. More examples are in how to write prompts for AI video generators.
Step 3: Choose the model, length and shape
| Setting | How to choose |
|---|---|
| Model | Cheap model for drafts (Vidu Q3 Turbo, Luma Ray 3.2); Veo 3.1 or Kling v3 for finals; a native-audio model if someone speaks |
| Length | Match the action: 4-6 s for one movement, 8-10 s for a slow reveal, up to 15 s on Kling v3 or Seedance 2.0 for a short scene |
| Aspect ratio | 9:16 for Reels, TikTok and Shorts; 16:9 for YouTube and websites; 1:1 for feeds |
| Resolution | 720p for drafts and most social; 1080p for finals and larger screens |
Decide on sound at this step too. If one voiceover will run over the whole video, generate silent clips (or mute them later) and record or generate the voiceover separately, so the voice stays the same from start to finish. Use a native-audio model only for shots where someone on screen speaks, or where a specific sound, like a pour or a click, makes the moment.
Generate every shot in the same aspect ratio. Cropping a 16:9 clip to 9:16 later throws away most of the frame. Our free social media video size guide lists what each platform expects.
Step 4: Generate, review and assemble
- Generate a draft of each shot on a cheap model.
- Review each clip for the action, the framing, faces, hands and any text on screen.
- Fix one thing at a time and regenerate the shots that failed.
- Render the final versions on the model and resolution you want.
- Join the clips in any editor (CapCut, DaVinci Resolve, Clipchamp, iMovie), in order.
- Add a voiceover or music, an end card with your offer, and captions.
Worked example: a 30-second ad from four lines of text
| Shot | Model in QuickVeed | Length | Credits |
|---|---|---|---|
| 1-4 drafts | Luma Ray 3.2, 540p | 5 s each | 4 × 38 = 152 |
| Hook shot final | Veo 3.1 Fast, with audio | 8 s | 300 |
| Shots 2-4 finals | Kling v3, 720p | 5 s each | 3 × 210 = 630 |
| Total | 23 s of footage plus a 7 s end card | 1,082 |
That is a little over one month of the Starter plan's 1,000 credits. Doing the finals on Luma Ray 3.2 or Vidu Q3 instead would bring the total to roughly 300 credits, at a lower level of detail. Pick premium models for the shots people will look at hardest, usually the first one.
Common mistakes when making video from text
| Mistake | What happens | Do this instead |
|---|---|---|
| Pasting a whole paragraph as one prompt | The model picks one idea or mashes several together | One shot per prompt, one action per shot |
| Vague words like “cinematic” or “amazing” | Generic, unpredictable results | Name the framing, camera move, light and colours |
| Different style words in each prompt | Clips don't look like the same video | Write one style line and reuse it in every prompt |
| Asking for readable text in the scene | Letters warp and misspell | Add text later in your editor |
| Writing long dialogue for a short clip | Rushed or cut-off speech | One sentence per 6-8 seconds, or use a separate voiceover |
| Rendering finals before testing | Credits spent on shots that don't work | Draft every shot cheaply first |
| Mixing aspect ratios | Black bars or heavy crops in the edit | Pick the shape first and use it for every shot |
One more: don't expect the model to know your brand. It has never seen your product, logo or shop. For those, start from your own photo with image-to-video, or film the shot yourself and use AI for the scenes around it.
Doing it in QuickVeed
QuickVeed's AI video generator runs 15+ text-to-video models from one screen, so you can draft on one and finish on another.
- Open the AI video generator (in the app it is under Create video).
- Type a prompt, or upload an image to use as the starting frame.
- Choose a model. Clip lengths run from 2 to 15 seconds depending on the model: Veo 3.1 gives 4, 6 or 8 s, Kling v3 3 to 15 s, Seedance 2.0 4 to 15 s, Runway Gen-4.5 5 or 10 s.
- Pick the aspect ratio (16:9, 9:16, 1:1, 4:3, 3:4 or 21:9 where the model supports it) and the resolution (480p to 1080p).
- Check the price shown, then generate. Failed generations are refunded, and every result is saved in your asset library.
Nine of the models generate sound with the picture: Veo 3.1 and Veo 3.1 Fast, Kling v3 and Kling v3 Omni, Seedance 2.0, Seedance 2.0 Fast and Seedance 1.5 Pro, and Vidu Q3 Pro and Vidu Q3 Turbo. Pick one of these for shots where someone speaks or the sound matters.
For narration, QuickVeed's AI voice generator reads your script in one of 9 text-to-speech models. Once your video is edited, the captions and subtitles tool transcribes it and burns captions on in one of 11 styles.
Frequently asked questions
Can AI make a video from text?
Yes. Text-to-video models generate a short clip from a written description, and script-to-video tools turn a longer script into a narrated video with stock footage.
What is the best AI to turn text into video?
For short generated clips, Veo 3.1 and Kling v3 are strong choices; for narrated explainers, a script-to-video tool is better. Test two models on your own prompt.
How long can a text-to-video clip be?
Usually 2 to 15 seconds per generation, depending on the model. Longer videos are several clips joined.
How do I make a long video from text with AI?
Split the text into shots and generate each one, or use a script-to-video tool that builds a 1-30 minute video from stock footage and a voiceover.
Is text to video AI free?
Some tools have free credits or trials; QuickVeed's 14-day trial includes 75 credits, no card needed. Free tiers usually exclude premium models.
Try it with your own video
Turn one line of your text into a shot prompt and generate a draft. Write one shot prompt from your text, draft it on a cheap model, and judge the result for yourself. QuickVeed plans start at $18/month, and new accounts get a 14-day trial with 75 credits, no card needed. Sign up or see pricing.
Sources
Competitor details in this post were checked against these public pages in October 2026. Products and prices change, so confirm with each vendor before you decide.
