How do you keep the same character in every AI video?
Make the character once on the Characters page in Faceless: describe it in one sentence or upload a photo, pick one of four looks, and it draws three reference pictures (front, waist up and full body). Pick it when you start a series, and every episode is drawn from those pictures. A character costs 14 credits to create.
Updated
Starter characters are ready-made and free to copy into your team. Episodes cost each format's normal price.
Describe your character
Free until you draw looksStyle
Sign in free to continue. Four looks are 8 credits and the three reference pictures of the one you keep are 6, 14 in all.
How to make a consistent AI character in 5 steps
- 1
Describe it or upload a photo
Write one sentence (up to 280 characters) and pick a style, or start from a photo of a person, a pet, a toy or a drawing.
- 2
Pick one of four looks
Faceless draws four looks for 8 credits. Keep one, ask for more like it, or edit the sentence and draw again.
- 3
Get the reference pictures
The look you keep is drawn three ways, front, waist up and full body, for 6 credits. These are what every episode is drawn from.
- 4
Name it and give it a voice
Name the character, pick a voice and check the personality it wrote: one line, traits and how it talks. All free and editable later.
- 5
Start a series that stars it
Pick a format from the list it offers. Each row shows the episode price and whether the format keeps its exact look. Set the schedule and launch.
Why AI characters drift, and the fix
A video model given only words draws a new person every time. “A red-haired detective in a grey coat” is a hundred different people, so episode two never matches episode one, and a channel built around a recurring face falls apart.
The fix is reference pictures. Faceless draws your character three ways once, stores those pictures, and passes them to every clip in a format that takes references. Its name, description, voice and personality are saved with it, so you set them once instead of retyping a prompt for every video.
Where it stars, and how much of it shows
Every series format that can star a 3D animated character today, at each format's starting tier, read from the same table the app uses. A realistic face is kept only on Motion Plus and Motion Max in most of them.
| Format | The character | Per episode |
|---|---|---|
| Street Interview | Exact lookSame look, in a live-action world. | 46 credits |
| Time-Traveller POV | Exact lookSame look, in a live-action world. | 55 credits |
| Stick & Meme Explainer | Same traitsA stick figure with the same traits. | 20 credits |
| Character Vlog | Exact lookSame look, in a live-action world. | 35 credits |
| Talent Show | Exact lookSame look, in a live-action world. | 66 credits |
| Pixar Celebrity Explainer | Same design, restyledRedrawn once in Pixar-style 3D. | 120 credits |
| Nursery Rhymes | Same design, restyledRedrawn once in the video's style. | about 180 credits |
| Music Video | Same design, restyledRedrawn once in the video's style. | about 270 credits |
| Reddit post | AvatarNarrates, and is the poster's avatar. | 20 credits |
| Facts & stories | DescribedDetails can change between scenes. | 20 credits |
| Dream ASMR | DescribedCan look different each episode. | 104 credits |
On Storyboard, your character is described in words, so details can change between scenes. A consistent look needs Motion Plus or higher.
The formats that keep the exact look
Each format's own sample episode, untouched. These are the formats that draw a character from its reference pictures in every clip.
Where it falls short
Know these before you plan a channel around one face.
Cheaper tiers describe, not draw
On Storyboard, your character is described in words, so details can change between scenes. A consistent look needs Motion Plus or higher.
Realistic faces need a higher tier
A realistic face is kept on Motion Plus and Motion Max in most formats, and in Micro Drama. On Motion and Storyboard it is usually described in words, so it can change between episodes. Each format says which before you launch.
No famous people
Characters must be original. A celebrity, a public figure or a well-known character from a film or game is refused, and so is a voice described as a real person's.
No lip-sync for characters yet
A character can narrate and appear on screen, but its mouth is not matched to the words. For talking-head ads with lip-sync, use AI actors in the UGC studio.
Frequently asked questions
Everything about keeping one AI character across videos
Give the video model pictures, not just words. A description alone is redrawn from scratch every time, so the face, the clothes and the colours drift. Faceless saves your character as three reference pictures and hands them to every clip it makes, in formats and tiers that take reference pictures. Each format says before you launch whether it keeps the exact look or describes the character in words.
Related tools
One Character is just the start. Automate your whole faceless channel
Make a character once and it stars in every episode: a series that writes, draws, voices and posts a new video on your schedule.
Faceless.so is an AI faceless video generator that writes, voices and posts a new video in your niche every day.
- Same look from reference pictures
- Its own voice and personality
- Posted automatically
- Starting price
- $29/mo
- To create
- 14 credits
- Styles
- 8
- Formats that keep its design
- 7
Free to sign in. A character is 14 credits; an automated series needs a paid plan.
Ready to Create Your Own Videos?
Start creating amazing AI-powered faceless videos in minutes with Faceless