How do I create a custom avatar?

Learn more about creating your custom avatar

Updated this week

To create an even more unique experience for your video projects you are able to create your own avatar, which can be then used in any of your projects. 💁‍♀️

The creation of a custom avatar is a paid add-on for yearly plan users only.

The following article takes through the process of creating a custom avatar including requirements needs for the optimal result.

Please note that the process listed here does not include a voice clone. However you can use our new voice cloning feature to do this

Let's jump in! 🥳


Synthesia Requirements

It is advised to practice hand movements, resting position, and reading the script at least once before the first take.

Please note that the avatar's quality depends on Synthesia receiving footage that matches these guidelines. If the standard is not met then a reshoot might be needed.

The requirements for creating a custom avatar are as follow. It is recommend that you read through each section ot ensure you create optimal footage for your avatar:

You can download a pdf version of the requirements here.


1. Video Background

When creating custom avatar footage the background of the video is important. We need to be able cut out the actor from the video.

Here’s what works well:

  • A GREEN SCREEN BACKGROUND and ensure the screen is well lit with no shadows. This method is required to create the highest quality avatar.

  • Make sure that the background does not clash with the clothes or skin tone.

  • Avoid wearing any green with a green background. Your footage may be rejected if this is not followed. Please use a white or blue background if you are wearing green clothing.

  • Do not remove the background yourself. We will do this before avatar creation.

Below are examples of a well lit green screen (left) and a plain white background (right)

Q: What do I do if I can’t shoot in a studio?

A: We can use footage shot in an office but be extra careful to make sure the actor stands out from the background and you have good lighting and clear audio.

We recommend that you have access to reshoot footage based on our feedback, if required.


2. Camera

A good quality camera is critical when creating footage. Make sure to shoot with a dedicated video camera.

Here’s what we need for the camera setup:

  • UHD resolution at 3840x2160

  • 29.97 fps or 30 fps (frame rate per second)

  • Position the camera at the same height as the actor’s eyes

  • Make sure the focus on the face is sharp

  • Do NOT use any type of streaming software, webcams or applications that capture your recording over an internet connection. Standard video production technique that consists of a camera and external microphone is required.

Q: What if I can’t shoot at UHD?

A: If you can only shoot HD video footage, that’s fine, but make sure to shoot the exact framing you would like to see for the avatar.

Q: What if I am shooting video in Europe?

A: If you shoot video in Europe with conventional room lighting such as office fluorescents, then you will need to shoot at 25fps to avoid video flicker. This is OK, however if you have studio lighting then please record at NTSC 29.97fps in Europe.


3. Framing

Set the framing for the upper body of the actor.

  • We recommend filming your actor in UHD 3840x2160 as we are able to reposition the framing slightly, if required.

  • If you are only able to film in HD 1920x1080, then we ask that you frame the avatar in the exact framing that you want the avatar to be in, as we are NOT able to adjust framing during the Avatar creation process.

Good examples of framing

Mid torso up Avatar

Hands in shot (left) and Hands not in shot (right). Our production team may need to reframe your avatar in order to create an avatar to a quality standard.

Chest up Avatar

Q: What if I want my avatar to show the upper body?

A: Make sure the hands, fingers and arms always return to the exact same place after speaking in each take, resting at the middle of your torso (or out of shot if you do not want your hands in the frame when resting). We recommend practicing this before turning the camera on. Keeping your hands together can help greatly with this.

Q: Can the actor use hand gestures?

A: Yes but please don’t move your hands outside of your body’s silhouette or above your chest. Find a comfortable resting position, and come back to it every time you are pausing. It is important to be resting more than you should be gesturing. If your hands and fingers do not come back to the exact same resting position, it is likely that our production team will crop your hands out of the framing


4. Lighting

Good lighting sets the tone and feel for your avatar.

Here’s what we recommend:

  • Use a three-point lighting setup

  • Ensure fixed even illumination on the actor

  • Avoid shadows, particularly across the face

  • Use a main Key light at the front to illuminate the actor, this is positioned slightly to one side to add depth to the face.

  • Use a Fill light on the other side of the camera to avoid shadows and a Backlight if possible, to highlight the actor and separate them from the background.

Q: What if a three-point setup is not possible?

A: No problem, just make sure the actor is well lit from the front and clearly stands out from the background. Make sure to check there are no strong shadows across the face and body. Keep the light flat and even.


5. Audio

We use audio to train our AI algorithms to match the lip movement for the actor. We can use the on camera audio but make sure it is clear:

  • The audio must be perfectly in sync with the video

  • No echo

  • No background noise

  • Make sure only the actor speaks, do not feed lines to the actor

  • Audio from a lapel mic (lavalier) or a boom mic is ideal, but we can use on camera audio.

Please note that we cannot remove a lapel mic that is visible so please hide well and practice to make sure the lapel mic is not scratching on any moving clothes.

Q: Do I need perfect audio?

A: Yes, please avoid all background noise or people talking other than the actor, especially at the same time.

We’re looking for 3 clear takes of the actor talking to the camera and as little background noise as possible.


6. Color

In the Avatar creation process, we DO NOT adjust the color grade for the actor.

It it therefor important to provide a video with the look you would like for your avatar:

  • Avoid dark footage, make sure you have good lighting to avoid this

  • Shoot with the look you want on camera, or grade the footage afterwards

  • Please DO NOT provide raw ungraded footage


7. Wardrobe

It’s your Avatar, so dress how you want the Avatar to look!

Here’s some important tips though to get the best footage:

  • DO NOT wear any clothing that clashes with your background

  • Avoid green clothing

  • Makeup must look natural

  • Keep hair tight to the head, avoid seeing the background through the hair

  • No sunglasses, we need to be able to see the eyes

  • No low hats, we need to be able to see the forehead above the eyebrows

  • No long beards, but short beards are OK, however, please check with customer support before filming.

  • Tattoos are usually okay but please also check with customer support before filming.

💡 TOP TIP 💡

Some things to avoid

  • Strong shadows

  • Camera blur

  • Not speaking to camera

  • Background action

  • Hair across the face

  • Sunglasses

  • Utilsgestures.

If in doubt, follow the rule that the face should be clearly visible to the camera at all times.

Q: Can I wear eyeglasses?

A: Yes, however, make sure there is no glare visible on them from your studio lighting. Seeing the light reflection in your glasses may introduce artifacts.

  • Try to avoid wearing glasses with transparent or thin frames.

  • Avoid rotating your head.

  • If in doubt, avoid wearing them if possible.


8. Performance

The performance of the actor on camera defines how your avatar videos will look on Synthesia. It helps to make sure the actor is calm, relaxed and takes their time on camera.

We use a standard script to help us analyse how the actor moves their lips in speech. It is important to deliver the script with short pauses and lips closed.

What should the actor’s performance look like?

  • Happy, friendly and relaxed; you’re smiling, are warm and inspire confidence

  • Avoid rotating your head and do not use exaggerated head movements.

  • Keep hands below the chest and within the silhouette of the body at all times

  • During pauses, if hands and fingers are in shot, they should return to the exact same resting position after gesturing - This is key for our technology to work well. Keeping your hands together can help greatly with this.

  • Avoid specific gestures like pointing and counting. It is better to be resting for longer periods of time than gesticulating.

  • Try not to blink too often or take prolonged moments with your eyes closed.

  • No rocking, or adjusting of body weight. Keep feet planted throughout.

  • Do not move away from the camera to check the script, or hold the script.

Please read the script first to familiarize yourself with the content. The words in between <> brackets are instructions only and should not be read out loud.

💡 Remember to:

  • SMILE!, relax and take your time 😁

  • When you take a break, close your lips 💋

  • Keep hand gestures at waist level 👋

  • Bring hands and fingers back to the same resting position when pausing 🤝

  • Keep the eye-line to the camera lens 🎥

  • Avoid exaggerated head movements 👩👨

Q: What if the actor is not a native English speaker?

A: If the actor cannot deliver the script with natural speech then change it. The content does not matter, only the delivery.

Do however maintain the instructions within the <> brackets as we need the actor to close their lips during breaks and pauses with different lengths. Translate the script or create a new one with similar length and break


9. Script

Below is the script we use to record an avatar. We ask that you provide 3 takes of the script (eg. repeat the same script 3 times).

  • The Closed Lips Image (💋 ) is the part of the script where the actor should pause and close their lips.

  • Do not read out loud the instructions within the <> brackets, they are just there to remind you of our requirements and to prompt you as to what to do with your hands

TELEPROMPT

START

<💋Lips Closed, pause for 3 seconds before beginning>

Join me as I demonstrate the creation of my lifelike avatar.

<💋 Look into the camera for 1 second with lips closed and hands in resting position>

In a quiet studio, I stand in front of a green screen, wearing a color that is different to the background.

<💋 Look into the camera for 1 second with lips closed and hands in resting position>

I'm all set with a smile on my face. My hands move whilst I speak, before then returning to rest.

<💋 Look into the camera for 1 second with lips closed and hands in resting position>

Behind the scenes, clothing and makeup professionals have made me look great. After moving my hands in front of my stomach and within my silhouette, I return my hands and fingers to a consistent resting position.

<💋 Look into the camera for 1 second with lips closed and hands in resting position>

I feel comfortable and relaxed as this will show in my body language.

<💋 Look into the camera for 1 second with lips closed and hands in resting position>

In post-production magic, the top three well-recorded takes are then processed to create my avatar.

<💋 Look into the camera for 1 second with lips closed and hands in resting position>

Avatar processing kicks off, blending precision and algorithms. Technology acts as a dedicated learner, observing my facial expressions and motion patterns.

<💋 Look into the camera for 1 second with lips closed and hands in resting position>

My avatar will reflect what I have done today. A mix of realism and technology, shaping moments through innovation and care.

<💋 Look into the camera for 1 second with lips closed and hands in resting position>

Finally, still looking towards the camera, I smile again for a few seconds at the end of this take. I move my hands for the final time, before bringing them back to the resting position.

<💋 Look into the camera for 1 second with lips closed and hands in resting position>

END

Q: Should I cut the video clips to provide the exact script?

A: No, please do not cut the videos. We require full takes without any cuts. It does not matter if the actor gets the lines wrong or word skipped. Just ask the actor to continue delivering the script naturally even if they make a mistake.

Q: Do I need to use a teleprompter?

A: You will get the best result if the actor is familiar with the script and then delivers to camera with a teleprompt to assist. If you don’t have a teleprompt then use a tablet positioned as close to the camera as possible, but please do check the eye line.

Q: But I can’t get a teleprompter! The eye line is off camera with my teleprompt! The actor will not follow a script, what do I do?!

A: Our script is designed with our tech requirements in mind.. Using another, may impact the final quality.

If you cannot use our script, show the actor the example video with the script and ask them to speak in a similar style looking directly to the camera lens. They will need to speak for 90s and take pauses with their lips closed.


Consent Recording

As part of the submission we require you provide a recording of the actor reading the following script face camera.

This script needs to be read in the actor's native language, for which we provide translations below.

For this recording you don’t need to worry about the style of the delivery, as long as the script is clearly stated by the actor, while facing the camera.

Consent Script - English

My name is Full name>, and this video recording will be used by Client Name> to create a lifelike custom avatar on the Synthesia platform.

Consent Script - German

Mein Name ist Vollständiger Name>, und diese Videoaufnahme wird von Kundenname> verwendet, um einen lebensechten benutzerdefinierten Avatar auf der Synthesia-Plattform zu erstellen.

Consent Script - Spanish

Mi nombre es Nombre completo>, y esta grabación de video será utilizada por Nombre del cliente> para crear un avatar personalizado realista en la plataforma de Synthesia.

Consent Script - French

Je m'appelle Prénom Nom>, et cet enregistrement vidéo sera utilisé par Nom du client> pour créer un avatar personnalisé sur la plate-forme Synthesia.

Consent Script - Portuguese

Meu nome é Nome completo>, e esta gravação de vídeo será utilizada por Nome do cliente> para criar um avatar personalizado realista na plataforma da Synthesia.

Consent Script - Italian

Mi chiamo Nome completo>, e questa registrazione video sarà utilizzata da Nome Cliente> per creare un avatar realistico personalizzato sulla piattaforma Synthesia.

Consent Script - Dutch

Mijn naam is Volledige naam>, en deze video opname zal gebruikt worden door Naam van cliënt> voor het maken van een levensechte avatar op maat op het Synthesia platform.

Consent Script - Mandarin

我的名字是Full name>,Client Name> 将使⽤这段视频在Synthesia 平台上创建⼀个栩栩如 ⽣的⾃定义头像。

Consent Script - Japanese

私は <⽒名> といいます。この映像は、<顧客名> が、動画プラットフォー厶のSynthesia上で 本⼈そっくりのカスタムアバターを作成するために使⽤されます。

Consent Script - Russian

Мое имя Полное имя> и данная видеозапись будет использована Имя клиента> для создания реалистичного пользовательского аватара на платформе Synthesia.

Consent Script - Danish

Mit navn er og denne video optagelse vil blive brugt af til at lave en digital avatar af mig selv til brug på Synthesia Studio platformen.

Consent Script - Finnish

Nimeni on Koko nimi>, ja Asiakkaan Nimi> tulee käyttämään tätä videotallennetta luodakseen aidon oloisen Avatarin Synesthesia-alustalla.

Consent Script - Romanian

Mă numesc Nume complet>, și această înregistrare video va fi folosită de Nume Client> pentru a crea un avatar realistic personalizat pe platforma Synthesia.

Consent Script - Polish

Nazywam się Imię i Nazwisko>, a to nagranie wideo będzie używane przez Nazwa firmy> do kreowania realistycznego awatara na podstawie mojego wizerunku na platformie Synthesia.


Submitting Your Footage

Once you have recorded your footage then send your 3 video takes and your consent recording through to the Synthesia team using the form available in Synthesia.

The files must be:

  • .h264 codec

  • .mp4 format

  • Less than 900mb per file.

Go into the Avatar section and click on the + Create avatar button, then follow the instructions.

By submitting your footage you acknowledge that the actor fully understands that their visage will be used to create an AI Avatar, that you agree to the Synthesia Ethical Guidelines and accept the Synthesia Terms and Conditions of Service.

It is important to note that the footage that you provide must have no jump cuts, or edits that have removed parts of the performance from the middle of the takes, otherwise this footage cannot be used to create your custom avatar.

Did this answer your question?