All guides
Tutorials

AI Character Video Maker: Turn Yourself Into an Original Character

Film yourself, describe a character that does not exist yet, and let the app build it onto your movement. What to type, what carries over and what it costs.

EEge AlpDeveloper, ActLike AIUpdated 21 October 20266 min read

How do you turn yourself into a character with AI?

To turn yourself into a character with ActLike AI, film a clip of yourself up to 20 seconds long, then describe an original character in words instead of uploading a photo. The app creates the character and applies it to your movement and expressions. Text mode adds one credit, shared with Scene Match: a 10-second video costs 11 credits at 480p or 21 at 720p with Scene Match.

AI-generated example. Left: a 5-second source clip of an AI-generated performer, not a real person. Right: the same clip after character replacement with the models the ActLike AI app uses, from a vampire description (pale skin, red eyes, fangs, high-collared cape), at 720p Best Quality with Scene Match: 11 credits in the app. Skin, hair and costume carried over; eye colour and fangs did not.
On this page

What an AI character video maker does with your clip

An AI character video maker keeps the performance and replaces the performer. In ActLike AI you supply two things: a short clip of one person, which provides the movement, timing, expressions and framing, and a character, which provides the new appearance. The character can be a photo you have permission to use or, as this guide covers, a description of someone who does not exist.

Describing a character is the simplest way to turn yourself into a character: you are the actor, and the words are the casting call. Because the character is invented, there is no question of whose face it is and no permission to ask for beyond your own.

Make a character video in four steps

  1. 1Film yourselfRecord one person, face visible from the first frame, in steady, even light. The limit is 20 seconds; five to ten is enough for a first try and keeps the cost down.
  2. 2Choose to describe a characterInstead of adding a photo, type a short visual description. The app creates an original character from it. This adds one credit to the render.
  3. 3Pick quality and Scene MatchCredit Saver 480p for a draft, Best Quality 720p for the version you share. Scene Match fits the character to your lighting and uses the same single credit as the description, so turning it on with a described character costs nothing extra.
  4. 4Render, review and saveRenders take a few minutes and finish in the background on both iPhone and Android. Watch the result in motion, then save it or share it from History.

The iPhone walkthrough and the Android walkthrough cover installation and each screen in more detail.

Writing a character description that works

  1. 1Name the type first"A sea captain", "a space pilot", "a 1920s jazz singer". One clear type is easier to render than a mixture of several.
  2. 2Add two or three visible detailsSkin, hair, headwear and clothing work best. "Weathered skin, a grey beard and a navy wool cap" gives the model something concrete to draw.
  3. 3Leave out the storyBackstory, personality and plot do not show on a face. Describe only what the camera would see.
  4. 4Keep it originalDescribe a type, never a named film, game or cartoon character, and never a real person. Named characters belong to someone else, and real people need to give consent.

Eight original characters to try

Character to describeClip to film480p draft720p + Scene Match
A weathered sea captain with a grey beard and a navy wool cap8 s: squint into the distance, then turn to camera917
A space pilot with short silver hair and a white flight suit collar6 s: look up, then nod713
A 1920s jazz singer with finger-waved black hair and pearl earrings8 s: sway slightly and smile917
An elderly wizard with a long white beard and a dark blue hood6 s: raise your eyebrows and lean in713
A Victorian detective with a tweed cap and sideburns5 s: a slow suspicious glance611
A medieval blacksmith with soot on his cheeks and a leather apron8 s: wipe your brow and look up917
A desert explorer with sun-tanned skin and a canvas wide-brimmed hat6 s: shade your eyes, then lower your hand713
A 1980s fitness instructor with big curly hair and a teal headband5 s: a big grin and a thumbs-up611

Credits are calculated from the clip length: 1 per second at 480p or 2 per second at 720p, plus the one shared credit for the description. Keep hands, hats and hair from covering your eyes in the source clip; the description can add a hat, but your own face needs to stay visible. For monsters, the Halloween guide has twelve more descriptions.

What carries over, and what does not

The recorded example above is a text-described character: a five-second clip at 720p with Scene Match, 11 credits in the app. Skin, hair and costume changed to match the vampire description. The requested red eyes and fangs did not appear, because the result follows the performer’s own eyes and mouth movement.

  • Usually reliable: skin tone, hair, headwear, collars and costume, overall face shape and the lighting mood.
  • Less reliable: eye colour, teeth and small details such as scars or jewellery. Describe the look through larger features instead.
  • Always yours: the performance. A smile, a glance or a head turn comes from your clip, so act the moment you want to see.

Can it make a cartoon character video?

ActLike AI is built to keep real movement and expressions from your clip, and the only recorded example on this site is a realistic character. We have not published a tested cartoon-style result, so we do not promise that a description such as "a cartoon fox" will produce an animated look.

If you want to try a stylised character, treat it as an experiment: describe a human-shaped character with exaggerated features, keep the clip to five seconds and render at 480p first, which costs 6 credits with the description. Compare it with the examples page, which separates recorded output from suggested recipes.

Trying several characters on the same clip

One good clip can carry many characters. Each render is a separate job with its own credits, so a draft of three different descriptions on a 5-second clip at 480p costs 18 credits in total. Pick the character you like, then remake only that one at 720p.

The Google Play version updated on 20 October 2026 makes this faster: it can reuse recent clips and faces with one tap, try another face on the same clip and render several videos at once. Check the App Store release notes before assuming the same on iPhone.

Credit packs are one-time purchases with no subscription; see the pricing page. The app licence covers personal or internal creative use, so read the terms before using a result anywhere else, and follow the acceptable use policy. The official App Store listing describes the describe-a-character option in the app’s own words.

Questions about this guide

Do I need a photo to make an AI character video?

No. In ActLike AI you can describe an original character in words instead of uploading a photo, and the app creates the character for you. Text mode adds one credit to the render, the same single credit as Scene Match.

How much does an AI character video cost in ActLike AI?

A described character costs the clip length in credits plus one: a 5-second video uses 6 credits at 480p, and a 10-second video uses 21 credits at 720p with Scene Match. Generation is paid with one-time credit packs, and credits for a failed render are returned automatically.

Can I turn myself into a famous movie or cartoon character?

No. Named film, game and cartoon characters belong to their owners, and real people need to consent. Describe an original character type with your own visible details instead.

More guides

Ready to make one yourself?

Reading gets you halfway. Download ActLike AI, follow the steps with your own clip, and you will have a finished video in a few minutes.

Download on the App StoreGet it on Google Play

iPhone requires iOS 17.0 or later. Android version on Google Play.

Written by Ege Alp · Developer, ActLike AI · Updated 21 October 2026