In this guide
Chapter 1
What you need to make a movie with AI
An AI movie isn't one button that hands you a finished film. It's layers you stack one on top of the other: a written story, scenes described in words, shots generated by a video model, a voice that tells the story or characters who speak, then music, subtitles and an edit that joins it all into one file.
The good news is that you need no camera, no actors and no experience with complex editing software to get to a 30 to 60-second short film. You need a clear idea, and an honest sense of what today's AI tools do well and what they don't.
An idea and a short story
One clear event with one or two characters. Short films work because of the idea, not the number of scenes.
A tool that generates shots
It creates clips from a written description or animates a still, like Vocla's AI video generator.
A voice to tell the story
One narrator or characters in conversation, in Modern Standard Arabic, an Arabic dialect or another language, with text to speech.
Music, subtitles and an edit
An instrumental music bed, readable subtitles for people watching on mute, and a way to join the shots and export them in the right shape.
A realistic size for a first film: 30 to 60 seconds, 6 to 10 scenes, each 3 to 8 seconds long. That's what current video models handle well, and the length that suits TikTok, Reels and YouTube Shorts.
Chapter 2
Write the story and script before any picture
The most common mistake in AI films is starting from the pictures: you generate beautiful shots, then try to invent a story for them. Reverse the order. Write what the viewer will hear first, because the words decide how long each scene lasts and when it changes.
Use a simple three-part structure: a beginning that introduces the character and their problem in one sentence, a middle with a turn or a surprise, and an end that solves the problem or leaves a question. A narrator speaks roughly two to two and a half words a second at a natural pace, so a 45-second film needs about 80 to 110 spoken words; more than that means a rushed voice.
- Open with a line that grabs attention in the first two seconds: it decides whether people keep watching or scroll.
- One sentence per idea, and short sentences that can be said in one breath.
- Write numbers the way they're spoken, and add diacritics to Arabic names and words with more than one reading.
- Read the script out loud and time it, or paste it into the free TikTok script planner, which estimates the spoken length and splits it into scenes in your browser.
Chapter 3
Split the film into scenes (the storyboard)
A storyboard is a simple table: one row per scene with what is said, what's on screen and how the camera moves. That table is what you'll give the video tool, and it's what stops you wasting time and credits on shots you don't need.
Write each shot description like a note to a director: who or what is on screen, what happens, where, how the camera moves, how it's lit and what style it has. Write it in English, which video models follow most reliably, even if your film is in Arabic, and repeat the same style words at the end of every description so the film feels like one piece rather than separate clips.
- One scene means one subject and one action.
- Don't ask the model for titles or signs inside the shot: lettering, Arabic above all, comes out garbled. Put words in the subtitles.
- Avoid logos, real people and celebrities. Use generic characters, figures seen from behind, or hands.
- Models render clips of set lengths, so a 5-second scene may be generated as a 6-second clip and trimmed to the voice.
Our in-depth guide on how to generate videos with AI explains the six parts of a scene prompt and includes a library of prompts you can copy.
Chapter 4
Generating the shots: text to video and image to video
There are two ways to generate the shots of your film, and you can mix them in the same film: scenes painted from a written description, and scenes that start from an image you own.
Text to video
You write the scene description and the model generates the clip from scratch, which is the way to go when you have no image or footage at all. In Vocla Video you paste your script: the voice-over is generated first, then the speech is split into 3 to 8-second scenes timed to it, and each scene gets an English visual description you can edit before rendering.
Image to video
You give the model a still as the first frame and it animates from there while keeping its look: a drawing of your character, a landscape or a product. It's the best way to keep a character looking the same across scenes. In Vocla, from the Starter plan up, you can open any scene on your own photo instead of a generated still, and for product photos specifically the product video ad maker turns one to four product photos into a finished ad with a voice and subtitles.
How to make a cartoon with AI
Cartoons are among the easiest styles for video models, because the eye forgives small mistakes in a drawing more than in a realistic image. Add a fixed style at the end of every description, such as 2D hand-drawn cartoon style, soft pastel colors or 3D animated family film style, soft lighting, and describe the character with the same details in every scene: age, clothes and hair colour.
Keeping characters consistent
The biggest challenge in AI films is a character whose face changes from shot to shot. Three habits reduce it:
- Copy the character description word for word into every scene they appear in.
- Prefer wide shots, shots from behind or hands over face close-ups.
- Review a still of every scene before generating video. In Vocla this step is the storyboard: one image per scene that you approve or regenerate, and the approved image becomes the first frame of the clip.
Chapter 5
The voice-over: how to make your film talk
Video models only paint pictures, and Arabic speech is not their strength, so the voice is added as its own layer. In short films the voice matters more than how pretty the pictures are: viewers forgive a simple shot, but they leave if the voice sounds robotic or the dialect feels foreign to them.
One narrator tells the story
The easiest style, and the one that works best when you're starting out. In Vocla Video your script is read by one voice you choose from the library, in Modern Standard Arabic or one of 18 Arabic dialects, and the scenes are timed to the speech automatically.
Characters in conversation
For a conversation between two characters, write one line per turn as "Name: line" in multi-voice text to dialogue, pick a voice for each character and a delivery for each line, and you get one audio file to lay over the shots in your editor. Keep in mind that characters in generated shots don't move their lips with the words, so use wide, side-on or from-behind shots during dialogue, or let the narrator report what the characters said.
Getting the pronunciation right
Add diacritics to words with more than one reading and to unusual names. The tashkeel and pronunciation tools add diacritics in one click and show every change for review, and the same button is in the script editor of Vocla Video.
Chapter 6
Music and sound effects
Music makes half the feeling of a short film. The rule: instrumental only, a rhythm that fits the mood, and clearly quieter than the voice-over. In Vocla Video you can switch on a music bed that is ducked under the speech automatically, or generate a separate track from a text description with AI music generation and place it in your edit.
For sound effects such as footsteps, rain or a door, take them from libraries licensed for your use and keep them low: one effect at the right moment beats ten.
- Pick one mood for the whole film: mysterious, warm or upbeat.
- Lower the music under speech and lift it slightly in the transitions.
- Check the rights to any music you didn't generate yourself before posting.
Chapter 7
Arabic subtitles on the video
Many people watch short videos with the sound off, and burned-in subtitles keep the story clear for them. Don't ask the video model to write the words into the shot; keep subtitles on a separate track that renders Arabic correctly, right to left.
In Vocla Video, Arabic subtitles are burned in from your script, timed to the voice, in a Clean, Bold or Karaoke style (each word lights up as it is spoken), at the bottom or the centre of the frame. If you already have a video from somewhere else, upload it to AI captions to transcribe it and add Arabic subtitles in five styles, or translate an English video into Arabic, then export an MP4 or SRT and VTT files.
- Two lines on screen at most at any time.
- Keep subtitles away from the bottom of vertical video, where platform buttons cover it.
- Check names and numbers before exporting.
Chapter 8
Editing and export: the right shapes and formats
If you make the whole film in Vocla Video, the basic edit happens automatically: clips are trimmed to the scenes and joined, the voice, subtitles and music are laid on top, and you download a finished MP4. If a scene doesn't work, you regenerate that scene alone instead of the whole film.
If you bring material from several tools together, such as a separate dialogue track or footage you shot yourself, any free editing app on your phone or computer is enough to order the clips and lay the audio over them. Either way, generate the film in its final shape from the start: cropping a widescreen video to vertical throws away most of the picture.
| Shape | Where it's used | Notes |
|---|---|---|
| 9:16 vertical | TikTok, Instagram Reels, YouTube Shorts, Snapchat | The best fit for short films on phones. Leave margins at the top and bottom for platform buttons. |
| 16:9 widescreen | YouTube, screens, presentations | Suits longer films and more cinematic stories. |
| 1:1 square | Facebook and Instagram feed posts | A middle ground that looks good on phones and computers. |
MP4 is the format every platform accepts. In Vocla you get vertical 9:16 from Starter, and all three shapes and 1080p from Professional up.
Chapter 9
Posting tips for TikTok, YouTube and Reels
TikTok
Make the first two seconds a question or an unexpected shot, and keep subtitles and faces away from the screen edges. The TikTok video maker page covers hooks and safe zones in detail.
YouTube and Shorts
The title and description decide whether the film shows up in search, so write the title the way people search. The vertical version suits Shorts; a 16:9 version suits the main channel.
Instagram and Facebook Reels
Burned-in subtitles are essential here because playback usually starts muted. Pick a clear cover image from the strongest shot of the film.
Disclosure and rights
Many platforms ask you to disclose realistic AI-generated content, so check each platform's policy. Never use real people's faces or voices without their permission.
Chapter 10
Step by step: make a movie with AI in Vocla
The whole path, from a blank page to an MP4 ready to post.
- 1
Write the script and time it
Write a story of 80 to 110 words for a film of about 45 seconds, and time it with the free script planner.
- 2
Open Vocla Video and paste the script
Choose the voice and dialect, the shape (9:16, 16:9 or 1:1), the quality tier and caption style, and switch on music if you want it. You see the credit cost before you start.
- 3
Let Vocla voice it and plan the scenes
The voice-over is generated first, then the speech is split into 3 to 8-second scenes with an English visual description for each.
- 4
Edit the scene descriptions
Add your fixed style, cartoon or cinematic, and the same character description in every scene, or upload your own photo to open a scene on plans that include it.
- 5
Approve the storyboard
Generate a still for each scene and regenerate the ones you don't like before you pay for video.
- 6
Render the film and review it on your phone
Watch the whole film once it's rendered, regenerate any single scene if needed, then download the MP4 with voice, subtitles and music.
- 7
Post it in the right shape
Upload the film with a title that matches what your audience searches for, a clear cover and a few precise tags.
Chapter 11
Worked example: a 45-second cartoon short
A short cartoon called "The Last Lantern", told by one narrator, in 9:16, in six scenes. Copy the prompts and change the details to fit your story, keeping the style words at the end of each prompt as they are.
- Scene 10–7 s· An old alley at dusk and a shop with one lantern left
NarratorIn an old alley, Uncle Salim sold lanterns… and he had only one left.
Scene prompt for the video modelA narrow old Arab alley at dusk, a small wooden shop with one glowing lantern left on its shelf, slow push-in, warm orange light, 2D hand-drawn cartoon style, soft pastel colors
- Scene 27–14 s· The old shopkeeper kindly shakes his head at passers-by
NarratorEveryone who passed asked its price, but Salim always said: this one is not for sale.
Scene prompt for the video modelAn old shopkeeper with a white beard and a brown vest kindly shaking his head at passers-by, medium shot, warm dusk light, 2D hand-drawn cartoon style, soft pastel colors
- Scene 314–22 s· A girl in a yellow raincoat in the rain, holding a coin
NarratorThen, one rainy night, a little girl stopped at his door, holding a single coin.
Scene prompt for the video modelA little girl in a yellow raincoat standing in the rain in front of the small shop, holding one coin in her open hand, low angle, falling rain, blue night with warm glow from the shop, 2D hand-drawn cartoon style, soft pastel colors
- Scene 422–30 s· The shopkeeper leans down to hear her
NarratorHe asked why she wanted it. "Because my grandmother is afraid of the dark," she said.
Scene prompt for the video modelThe old shopkeeper leaning down toward the little girl at the shop door, seen from the side, lantern light between them, rain in the background, 2D hand-drawn cartoon style, soft pastel colors
- Scene 530–38 s· Old hands place the lantern in the girl's hands
NarratorSalim smiled, placed the lantern in her hands… and gave her back the coin.
Scene prompt for the video modelClose-up of an old man's hands placing a glowing lantern into a little girl's small hands, warm light on their fingers, raindrops glistening, 2D hand-drawn cartoon style, soft pastel colors
- Scene 638–45 s· A high shot of the dark alley with one lit window
NarratorThat night there was only one light in the alley… and it was enough.
Scene prompt for the video modelHigh wide shot of the dark old alley after the rain, a single window glowing warmly with lantern light, slow pull-back, calm and heartwarming, 2D hand-drawn cartoon style, soft pastel colors
The spoken script here is about 90 words, roughly 40 to 45 seconds at a calm pace. A film this long needs Professional, because the longest video on Starter is 15 seconds. At the default rates, the Lite tier costs 150 credits per rendered second and each storyboard still 200 credits, plus the voice-over; the exact cost is shown before you render.
Chapter 12
What's really free, and what needs credits or a paid plan?
Let's be straight: AI video generation is expensive to compute, and no serious tool gives you unlimited films for free. What you can do for free in Vocla is most of the work around the picture; generating the shots themselves starts on a paid plan.
If you want a completely free film, the realistic path is: write the story, generate the voice-over from your free credits, use your own photos or footage, then put it all together in a free editing app. When you want generated shots, music and burned-in subtitles in one place, compare the plans on the pricing page.
| Part | Free | Needs credits or a paid plan |
|---|---|---|
| Writing and timing the script | The script planner runs in your browser with no account | Nothing |
| Trying a voice | The demo on the text to speech page, no sign-up, up to 200 characters per try | Nothing |
| Voice-over | From the Free plan's credits: 500͏ credits a month, no card, for personal use | Commercial use and WAV downloads from Starter |
| Tashkeel and pronunciation | Free on every plan | Generating the audio uses credits |
| Dialogue, music, captions and dubbing | Not on the Free plan | On every paid plan; longer files on higher plans |
| Text-to-video generation | Not on the Free plan | From Starter: 30 seconds a month, videos up to 15 seconds, 720p, vertical, with a watermark |
| Image to video | Not on the Free plan | From Starter, with the same limits as text to video; every shape without a watermark from Creator, 1080p from Professional |
In some countries, currently including Egypt, audio generated on the Free plan ends with a short spoken tag saying it was made with Vocla. Paid plans never get the tag.
Chapter 13
Common mistakes when making a movie with AI
Starting from pictures, not a story
A string of pretty shots with no story is something nobody finishes. Script first, always.
Crowded scenes
Many characters and actions in one shot turn into visual noise. One idea per scene.
A style that changes every scene
Repeat the style words and the character description word for word in every prompt.
Asking for Arabic text in the shot
The letters come out garbled. Put words in the subtitles, not the picture.
A script longer than the film
A rushed voice kills the mood. Cut words instead of speeding up.
Rendering video before approving stills
Changing a still costs far less than regenerating a video clip.
Forgetting the screen edges
TikTok and Reels buttons cover the sides and bottom of vertical video.
Ignoring rights
No faces, voices or music of other people without their permission.
Chapter 14
Questions about making a movie with AI
How do I make an AI movie?
Write a short 30 to 60-second story, split it into scenes, generate a shot for each scene from a written description or from an image, then add a voice-over, music and subtitles and join everything into one file. In Vocla Video you paste the script and it handles the voice, scene planning, generation, subtitles and music, and hands you a finished MP4.
Can I make a movie with AI for free?
Partly. Writing and timing the script are free, and the voice-over runs on the Free plan's credits (500͏ credits a month, for personal use). Dialogue, music and captions come with every paid plan, and generating the video shots in Vocla starts with Starter. You can make a free film from your own photos or footage with a Vocla voice and a free editing app.
How do I make a movie from a photo with AI?
Upload the photo as the first frame of a scene and describe what happens in it (the camera move, what moves, the light), and the model animates it while keeping its look. In Vocla, from the Starter plan up, you can open any scene on your own photo, and the product video ad maker turns one to four product photos into a complete ad with a voice and subtitles.
How do I make a cartoon with AI?
Write a simple story, then add a fixed style at the end of every scene prompt, such as 2D cartoon style or 3D animated style, describe the character with the same details in every scene, and approve a still for each scene before generating video so the character stays consistent.
How long can my AI movie be?
Each clip a model generates is short, a few seconds, and a film is built from several scenes. In Vocla a single video can be up to 15 seconds on Starter, 60 seconds on Professional, 120 seconds on Business and 180 seconds on Enterprise. For longer films, make parts and join them in an editor.
Do I need editing experience?
No. If you make the whole film in Vocla Video, trimming and joining the clips and laying the voice, subtitles and music happen automatically. You only need a simple editing app if you combine material from different tools, such as a separate dialogue track or your own footage.
How do I make the characters talk?
The easiest way is one narrator telling the story. For dialogue, write the script as "Name: line" in text to dialogue, give each character a different voice and get one audio file to lay over the shots. Generated shots don't sync lips to speech, so use wide or side-on shots during dialogue.
Can I post the movie on YouTube and earn money from it?
A monetised channel is commercial use. On Vocla the Free plan is for personal use and evaluation, commercial use is included from Starter up, and every plan with video is a paid plan. The monetisation rules themselves are YouTube's, so check its policies on AI-generated content and disclosure, and publish original stories rather than repetitive clips.
Can AI write Arabic text inside the video?
Not reliably: Arabic letters inside generated shots usually come out garbled. That's why words are added as separate burned-in subtitles that render Arabic correctly, which Vocla Video does automatically.
What's the best video size for TikTok and YouTube?
Vertical 9:16 for TikTok, Reels and Shorts, widescreen 16:9 for YouTube, and square 1:1 for feed posts. Generate the film in its final shape from the start rather than cropping it later.