9:16 · native sound · captions · no watermark
AI video generator for YouTube Shorts — vertical 9:16 clips with sound, captions and voice
A Short is a vertical video of up to three minutes, and the first two seconds decide whether anyone stays. Here is how to make one with AI: write a hook, generate 9:16 clips with sound from a prompt or a start frame, add captions and a voice-over, and upload.
- Vertical 9:16 out of the box
- Clips up to 30 s with sound
- Captions in 10 styles
- No watermark
Updated by the Kenerate AI team
Made with Kenerate
On this page
Quick answer
What is the best AI video generator for YouTube Shorts?
An AI video generator for YouTube Shorts makes vertical 9:16 clips from a prompt or a still image. On Kenerate, Wan 3 is the all-round pick: 2–30 s per clip, 480p–1080p, native sound and spoken lines, from 15 credits a second at 480p. Use Seedance 2.5 for cinematic shots up to 4K and Veo 3.1 Fast for short punchy hooks. Shorts can run up to 3 minutes, so longer Shorts are several clips joined in an editor — then add captions with YouTube subtitles and narration with the Shorts voice generator. Video needs a paid plan.
Key facts
- Format
- 9:16 vertical on every model listed here
- Clip length
- 2–30 s per generation (Wan 3, Seedance 2.5)
- YouTube Shorts limit
- Up to 3 minutes (YouTube Help)
- Sound
- Native effects, music and spoken lines in quotes
- Captions
- Auto Subtitles, 10 styles, clips up to 4 min
- Plan needed
- Paid plan for video; no watermark
Real results
Vertical clips made for Shorts
Filter by style: talking clips with burned-in captions, faceless b-roll, product loops and an animated character. Real 9:16 outputs, five seconds each.
Every clip is an unedited output of the model named on it (the two captioned clips then went through Auto Subtitles), made from a fictional AI image or a text prompt; open one to see its prompt.
How to
How to make a YouTube Short with AI
- Time needed
- Steps
- 4 steps
Step 1
Write the hook and a short script
One idea per Short. Write the first line the viewer hears or reads — the hook — and split the rest into 5–10 second beats, one clip per beat.
Step 2
Make a start frame (optional)
For a recurring character, product or face that must stay the same, make a vertical still first with the AI image generator, or upload your own photo. Skip this for faceless b-roll and type the scene as a prompt.
Step 3
Generate 9:16 clips
Open the video studio, choose Wan 3, set 9:16 and 5 seconds, and describe the action and the sound. Put any spoken line in quotes. Draft at 480p, then re-render the keepers at 1080p.
Step 4
Add captions and voice
Run each clip through YouTube subtitles to burn in captions in one of 10 styles, or make narration with the Shorts voice generator and lay it over faceless clips.
See it in the app
Open Video Studio in Text to Video mode and write the shot — subject, action, camera move and light.
Watch the tour · 1:19
Text, image & reference to video, models, edit, extend & chat
Chapters
Checked 1 November 2026
Shorts formats and how to prompt them
Common Shorts formats, what to put in the start frame, and what to put in the motion prompt.
| Format | Start frame | Motion prompt |
|---|---|---|
| Talking tip | A fictional presenter facing the camera, vertical, clean background | One line in quotes + tone: "says: 'Stop doing this one thing.' confident voice" |
| Faceless b-roll | None — text prompt only | "close-up of hands kneading dough, flour dust, handheld camera, soft kitchen sounds" |
| Product loop | Your product photo on a plain backdrop | "spins slowly and bounces, studio light, upbeat music" — static background |
| Story series | The same character image for every episode | One beat per clip: "opens the door and gasps", then the next beat in the next clip |
| Before / after reveal | The 'before' state | "the camera pushes in as the room transforms, whoosh sound" |
| Animated explainer | A cartoon character or illustrated scene | "points to the side and explains, friendly narrator voice" |
Use fictional characters, your own photos and your own products — not real people's faces or copyrighted clips.
Sources: Kenerate video studio · YouTube Help — Get started with YouTube Shorts (up to 3 minutes) · YouTube Help — Disclosing use of altered or synthetic content · YouTube Help — YouTube channel monetisation policies
Engines
Which video model for which kind of Short
All of these render 9:16 from a prompt or a start frame. Credits are for a 5-second clip at the model's default resolution.
- Most examples
Alibaba
Wan 3
- Input
- Image or text (+ optional end frame)
- Output
- 2–30 s · 480p–1080p · sound · 150 credits / 5 s at 720p
Open Wan 3Best for
Talking clips, faceless b-roll and longer beats
ByteDance
Seedance 2.5
- Input
- Image, text or references
- Output
- 4–30 s · up to 4K · sound · 540 credits / 5 s at 720p
Open Seedance 2.5Best for
Cinematic action and multi-shot scenes
Google
Veo 3.1 Fast
- Input
- Image or text
- Output
- 4, 6 or 8 s · up to 4K · sound · 225 credits / 5 s
Open Veo 3.1 FastBest for
Short hooks with a spoken line
Pixverse
Pixverse v6
- Input
- Image, text or references
- Output
- 1–15 s · 360p–1080p · sound · 68 credits / 5 s
Open Pixverse v6Best for
Budget social clips and quick tests
Lightricks
LTX 2.3
- Input
- Image or text
- Output
- 5–20 s · 480p–1080p · silent · 45 credits / 5 s
Open LTX 2.3Best for
Cheap silent b-roll under a voice-over
Lengths, resolutions and credits are from the video studio's own model list (1 Nov 2026). The examples above name the model that made each clip.
Tips
What makes an AI Short hold attention
Shorts habits that also help the model.
Start mid-action: the first frame should already be moving or speaking, not a slow establishing shot.
Frame for vertical: one subject in the centre third, so captions at the top or bottom don't cover the face.
Keep spoken lines under ten words per 5-second clip — split longer lines across clips.
Name the sound in the prompt: "sizzling pan", "crowd cheering", "calm narrator voice".
Reuse one start frame for every clip of a series so the character stays the same.
Avoid near-identical uploads: vary scripts and visuals, because YouTube does not monetise mass-produced, repetitive content.
FAQ
AI videos for YouTube Shorts — questions
9 questions
How do I make a YouTube Short with AI?
Write a hook and split the idea into 5–10 second beats. In Kenerate's video studio pick Wan 3, set 9:16, describe each beat with its sound and generate the clips. Add captions with YouTube subtitles or a voice-over with the Shorts voice generator, join the clips in an editor and upload the MP4 to YouTube as a vertical video.
Link to this answerWhat size and length should an AI Short be?
Vertical 9:16, at 1080p for upload. YouTube Shorts can be up to 3 minutes long. Each AI generation is 2–30 seconds on Wan 3 and Seedance 2.5 (up to 15 on Pixverse v6, 8 on Veo 3.1), so a Short longer than one clip is several clips joined together.
Link to this answerIs there an AI video generator for YouTube for free?
Not on Kenerate: video needs a paid plan, and new accounts get free credits for images. The cheapest clip with sound is Wan 3 at 480p — 15 credits a second, so 75 credits for 5 seconds; LTX 2.3 makes silent clips for 45 credits per 5 seconds. Downloads have no watermark. Free tiers elsewhere usually add a watermark or cap length.
Link to this answerCan AI-generated Shorts be monetised?
Yes, if the channel meets the YouTube Partner Program rules. YouTube doesn't ban AI tools, but its monetisation policy excludes mass-produced or repetitive content — for example near-identical videos with only small changes. Original scripts, varied visuals and your own voice or commentary keep a channel on the right side of it.
Link to this answerDo I have to label AI videos on YouTube?
Only realistic ones. YouTube asks creators to turn on the altered or synthetic content setting when a video shows realistic people, places or events that were made or meaningfully changed with AI. Clearly unrealistic or animated content, and AI help with scripts, captions or thumbnails, don't need the label.
Link to this answerCan the people in the clip talk?
Yes, on sound models such as Wan 3, Seedance 2.5 and Veo 3.1. Put the line in quotes in the prompt and say how it's said — "calm", "excited", "deep voice". Keep it under ten words per 5-second clip. For an exact script in a chosen voice, make it with the Shorts voice generator and use lip sync.
Link to this answerCan I make faceless Shorts?
Yes. Generate b-roll from text prompts only — hands cooking, landscapes, products, city scenes — then add narration from the Shorts voice generator and burned-in captions. Silent models such as LTX 2.3 are a cheap fit when the voice-over carries the audio.
Link to this answerHow do I keep the same character across Shorts?
Make one vertical image of the character and use it as the start frame for every clip. Wan 3 and Seedance 2.5 also take reference images, which helps the face and outfit stay consistent across a series.
Link to this answerCan I make the thumbnail with AI too?
Yes — Shorts mostly show a frame from the video, but for the channel page and long-form videos you can make a thumbnail with the AI thumbnail generator.
Link to this answerIn depth
AI and YouTube Shorts, explained
On this page
One clip or many?
YouTube Shorts can be up to 3 minutes long [1], but an AI generation is a single shot of 2–30 seconds. Most good Shorts are 15–60 seconds built from three to eight clips: a hook, a few beats and a payoff. Generating beat by beat also means you only re-roll the clip that didn't work.
- Under 30 s: one Wan 3 or Seedance 2.5 clip can be the whole Short
- 30 s – 3 min: several clips joined, with captions and a voice-over tying them together
Disclosure and monetisation
YouTube asks for the altered or synthetic content label on realistic AI scenes — a real-looking person saying something they didn't, or an event that never happened — and not on clearly unrealistic or animated content [2]. Separately, the Partner Program does not pay for mass-produced, repetitive videos [3]. Neither rule bans AI: they ask for honesty about realism and for original work.
Keep it original
Use fictional characters, your own photos and your own products. Putting a real person or celebrity into a Short without permission can break YouTube's rules and likeness laws, and Kenerate refuses content of minors, sexual content of real people and illegal content.
YouTube Shorts AI video maker
Same job: generate vertical clips, add captions and a voice, join and upload. The model does the footage; your hook and script do the work.
AI video generator for YouTube free
Video on Kenerate needs a paid plan; the cheapest clip with sound is Wan 3 at 480p for 75 credits per 5 seconds, with no watermark.
YouTube Shorts AI generator from text
Type the scene as a prompt — no start image needed. Good for faceless b-roll; use a start frame when a character must stay the same.
AI Shorts with voice
Either a spoken line in quotes on a sound model, or narration from the Shorts voice generator laid over silent or ambient clips.
Explore
Every Kenerate video tool
Generators, tools, models and guides — all in one studio.
Try Kenerate AI Video Generator for YouTube Shorts
One studio for images and video.
Tried it? Tell us how it went
