How to make an AI video in four steps
- 1
Describe your video
Type a topic, story idea or outline — or upload an image or paste a URL. Add the angle, audience and ending you want.
- 2
Choose style, voice, format and length
Pick an art style, one of up to 15 voices, the narration language, 9:16, 16:9 or 1:1, and a length from 15 seconds to 5 minutes.
- 3
Pick a video model
Choose from 15 models, from budget (Wan 2.2 Fast, HeyGen) to premium (Kling 3.0, Seedance, Veo 3.1). The credit estimate updates as you choose.
- 4
Generate and download
ShortsMachine writes the script, generates each scene, adds voiceover and captions and stitches the video in the background. Download the MP4 when it is done.
Four ways to start a video
Text
Type a topic, story or outline. The most flexible way to start and the one that makes the longest videos.
Image
Upload a photo or illustration and describe the motion. Best for animating a product shot or a character you already have.
URL
Paste an article link and get a video built from its key points. Ideal for repurposing blog posts (Pro and Premium).
Slideshow
Narrated AI images with slow zooms and captions. Fast, cheap and great for high-volume faceless channels.
Write an idea that becomes a good video
The script writer works from your idea, so the clearer the idea, the better the video. A topic like “octopus facts” works, but an angle works better: who it is for, what makes it surprising, and how it should end. “Three facts about octopuses that sound made up, for curious kids, ending with a question” gives the writer a hook, a tone and a payoff.
Mention the look too — art style, setting, palette and lighting — and list any facts that must appear. More examples in our guide to AI video prompts.
Characters that stay the same from scene to scene
AI video models have no memory between clips, which is why characters often change face or clothes between shots. ShortsMachine starts every video by writing a visual style line — the characters, their clothing, the setting, the palette and the lighting — and applies it to every scene. Each clip then starts from a key image drawn with that style, so the lighthouse keeper in scene one is still the lighthouse keeper in scene six. Read more in keeping AI characters consistent.
Voice, captions and sound
Every scene is narrated with a natural AI voice in one of 11 languages, and the edit is timed to the voice: each scene lasts as long as its line, so sentences are never cut off. Captions are split into short chunks and burned into the video in bold, high-contrast text. When the video model creates its own sound — ambience, waves, footsteps — it is mixed quietly under the narration.
How long should your video be?
| Use | Suggested length |
|---|---|
| Single fact, joke or teaser | 15–30 seconds |
| YouTube Shorts, TikTok, Reels | 30–60 seconds |
| Story or explainer Short | 1–3 minutes |
| YouTube story or documentary | 3–5 minutes (16:9) |
Videos longer than one clip are built scene by scene and stitched, so any model can produce up to 5 minutes. See how long AI videos are made.
Choose a video model
Cheap, fast models are ideal for testing ideas and daily faceless content; premium models give smoother, more realistic motion for flagship videos. You can switch per video.
| Model | Maker | Resolution | Clip length | Formats | Sound | Plan |
|---|---|---|---|---|---|---|
| Wan 2.2 Fast | Alibaba | 480p | 5s | 16:9, 9:16 | Voiceover only | Starter |
| HeyGen Video 1 | HeyGen | 768p | 5–15s | 16:9, 9:16 | Voiceover + model audio | Starter |
| Pruna P-Video | Pruna AI | 720p | 1–10s | 16:9, 9:16, 1:1 | Voiceover only | Creator |
| Grok Imagine 1.5 Lite | xAI | 720p | 1–15s | 16:9, 9:16, 1:1 | Voiceover only | Creator |
| MiniMax H3 | MiniMax | 768p | 5s | 16:9 | Voiceover + model audio | Pro |
| Wan 3.0 | Alibaba | 720p | 5s | 16:9, 9:16, 1:1 | Voiceover + model audio | Pro |
| Grok Video Pro | xAI | 720p | 1–15s | 16:9, 9:16 | Voiceover only | Pro |
| FLUX 3 Max | Black Forest Labs | 720p | 5–20s | 16:9, 9:16, 1:1 | Voiceover + model audio | Pro |
| Kling 3.0 | Kuaishou | 720p | 3–15s | 16:9, 9:16, 1:1 | Voiceover + model audio | Premium |
| Seedance 2.0 Mini | ByteDance | 720p | 4–10s | 16:9, 9:16, 1:1 | Voiceover + model audio | Premium |
| Wan 2.7 | Alibaba | 720p | 2–15s | 16:9, 9:16, 1:1 | Voiceover + model audio | Premium |
| Veo 3.1 Fast | 1080p | 4/6/8s | 16:9, 9:16 | Voiceover + model audio | Clip pack | |
| Gemini Omni Flash | 1080p | 3–10s | 16:9, 9:16 | Voiceover + model audio | Clip pack | |
| Seedance 2.0 | ByteDance | 720p | 4–15s | 16:9, 9:16, 1:1 | Voiceover + model audio | Clip pack |
| Seedance 2.5 | ByteDance | 720p | 4s | 16:9, 9:16 | Voiceover + model audio | Clip pack |
Specs as offered in ShortsMachine (October 2026). Clips are stitched, so any model can make videos up to 5 minutes.
What a video costs
Plans include monthly credits. One credit is about one second of the cheapest model; each scene also uses about two credits for its image and voiceover. You are only charged when a video finishes.
| Model | Credits per second of clip | Credits per finished minute | Minutes on Starter (200) | Minutes on Premium (2,000) |
|---|---|---|---|---|
| HeyGen Video 1 | 1 | ~72 | 2.8 | 27.8 |
| Wan 2.2 Fast | 1 | ~84 | 2.4 | 23.8 |
| Pruna P-Video | 2 | ~132 | 1.5 | 15.2 |
| Grok Imagine 1.5 Lite | 3 | ~192 | 1.0 | 10.4 |
| FLUX 3 Max | 4 | ~252 | 0.8 | 7.9 |
| Grok Video Pro | 5 | ~312 | 0.6 | 6.4 |
| MiniMax H3 | 6 | ~384 | 0.5 | 5.2 |
| Kling 3.0 | 8 | ~492 | 0.4 | 4.1 |
| Seedance 2.0 Mini | 9 | ~552 | 0.4 | 3.6 |
| Wan 2.7 | 10 | ~612 | 0.3 | 3.3 |
| Wan 3.0 | 14 | ~864 | 0.2 | 2.3 |
Per-minute figures include the scene images and voiceover (Flux images). Estimates; you are charged for what is actually generated.
| Plan | Price | Monthly credits | Slideshows | Video models |
|---|---|---|---|---|
| Starter | $3.99/mo | 200 | 20 | 2 |
| Creator | $9.99/mo | 500 | 50 | 4 |
| Pro | $19.99/mo | 1,000 | 150 | 8 |
| Premium | $39.99/mo | 2,000 | 300 | 11 |
Every plan starts with a 3-day free trial. Extra credits and Veo / Seedance clip packs can be bought any time.
Who uses an AI video generator
- Faceless channel creators producing daily Shorts in niches like history, horror, facts and motivation.
- Small businesses turning products and blog posts into social videos without a film crew.
- Teachers and trainers making short narrated explainers in several languages.
- Storytellers bringing fables, fiction and series to life with consistent characters.
Fixing the most common problems
| Problem | What to do |
|---|---|
| Characters look different in each scene | Describe the main character in detail (hair, clothes, one accessory) and keep the cast small. The shared style line does the rest. |
| The video feels slow | Ask for a hook in the first line and choose a shorter length. Fewer words per scene means faster cuts. |
| A scene shows a still image instead of motion | The model refused or failed that shot (often a content check). Rephrase anything graphic, or try another model. |
| Text in the video is garbled | AI video models cannot render reliable text. Keep words in the narration and captions, not in the visuals. |
| The facts are wrong | AI scripts can be confidently wrong. Put the key facts in your prompt and review the script before publishing. |
| Not enough credits | Use a cheaper model or a shorter length, buy a credit pack, or use slideshow mode, which has its own allowance. |
What ShortsMachine does well — and what it doesn’t
Does well
- Finished, narrated videos from one idea
- Consistent look across scenes
- Videos up to 5 minutes
- 15 models from budget to premium
- Captions and voiceover in 11 languages
Doesn’t do (yet)
- Editing footage you upload
- PDF to video (coming soon)
- Custom caption fonts or your own voice
- Readable text or logos inside AI visuals
- Realistic videos of real people
AI video glossary
- Scene
- One shot of the video: a key image, an AI clip, a line of narration and its captions.
- Start frame
- The image a video model animates from. Starting every scene from a generated image keeps the look consistent.
- Visual style line
- A one-sentence description of characters, setting, palette and lighting, applied to every scene.
- Stitching
- Joining all scenes into one video after normalising their size, frame rate and audio.
- Credits
- ShortsMachine's unit of usage. Roughly one credit per second of the cheapest model; premium models use more.
- Clip pack
- Prepaid clips for the most expensive models (Veo, Gemini Omni, Seedance 2.0/2.5). One clip per scene.
- Faceless video
- A video where no presenter appears on camera; narration, visuals and captions carry it.
- Aspect ratio
- The shape of the video: 9:16 vertical for Shorts, TikTok and Reels; 16:9 for YouTube; 1:1 for feeds.
More in our AI video guides, including Seedance vs Kling vs Veo and how to make faceless YouTube Shorts.