Create your own custom AI Avatar

Start by creating a Video Canvas project. You will want to have already created either a 16:9 or a 9:16 image of your actor or if you are using an image of yourself, have cropped that image to these dimensions. Once you have your image ready, look on the right side of the video canvas for the purple Create Actor Videos button. Click this to open up the Actor Video Assistant, an agentic workflow panel that lets you upload or select your actor image from your brand kit. Add your actor image(s) then run a vision analysis (optional) and enter in the Actor Dialogue and Action Sequence you want. Chat with the Actor Assistant about your concept, then click the Make Actor Videos button when you are ready. It takes around 2 to 3 minutes to generate up to 6 videos. Once generated, you can save the videos to your brand kit, download them or place them directly on a scene in your video canvas.

Yes, but keep in mind that if you have the actor moving, or things in the scene with motion, you will be limited to 8 seconds or less of dialog when you lip-sync a new version of the actor. Lip-sync will loop the video, meaning it will start over at the beginning of the video when the recorded scene ends. These videos are 8 seconds or less in ln length, so if you have an actor walking towards the camera, for example, and the script you give is 20 seconds long, the actor will walk towards the camera then at 8 seconds, will continue with the script but be shown back at the starting point of the video walking towards the camera again. This is why stationary actors, sitting, standing with limited hand gestures are the recommended avatar format you create. The same holds true with Camera Motion settings, if you have the camera pan in or zoom out, you will see this loop on longer lip-sync scripts if you use the avatar for other scenes in the future.

The initial generative AI video is limited to 8 seconds of dialog. This is just long enough to say two medium length sentences. Once your AI Avatar is created, you can lip-sync a much longer script (minutes long). The key to making your own AI Avatar video is to ensure the action sequence has minimal background movement (i.e. don't have a busy office scene in the background or you may see people moving/looping on longer lip-sync videos) and minimal hand gestures (i.e. don't start the video with the actor showing their hands raised in the air or it will start and end that way and the longer lip-sync will show them raising and lowering their hands again and again). If you are creating the AI Avatar for short statements (8 seconds or less), you can use any action sequence, hand gesture or scenes you want.

We recommend if you want to use the same actor in different scenes of your video project, sitting, standing, or with different camera angles, you first start with the Create Actor Permutations panel. This is another AI assistant that generates scenes with your actor in them and is built to make images with different expressions, camera angles and gestures. It will make between 5 to 10 variations per scene you request. So for example, you can upload an actor facing forward, and ask for 4 scenes with 10 variations each, showing the actor with different camera angles and expressions and the AI assistant will return 40 actor images in around 2 minutes or less, and you can pick the angles and poses you like. Each can be saved to your brand kit, downloaded or placed on a scene in your video canvas project. They are static images and are great for supporting visuals or to use as the basis for making new AI Avatar videos speaking the script you want.

When you have your custom Full Scene AI Avatars made, they each will speak the initial dialog you entered in but you may notice the voices vary. This is ok as when you finally lip-sync your script to the voice actor you want, all will have the exact same voice, tone and infliction you create on your script. Think of making your Full Scene avatar as a reusable asset - if you like the initial short script from the generation, great. No need to lip-sync a new script. But if you want to change the script later (A/B testing, variation, used in different scenes) you will want to lip-sync the generation.

Green Screen AI Avatars are created from a single photo and have a special masking process applied to them (called green screening) so that when they are used on a video canvas project, they can be composited over other videos, background elements (like text or shapes or images) and they have no background. We have created a really easy process to add the greens creen, generate the AI Avatar and return it masked so you can place this on your video canvas and see how the final generation will look. To create a Green Screen AI Avatar, simply add a high-quality photo of yourself or an actor and then open up the Create Green Screen Avatar Videos panel (click on the green head icon on the right side of the video project), add your photo, enter the initial dialog you want (keep it short as these are 8 second vidoes) and click the Make Actor Videos button. It takes around 3 minutes to make your video. You can create up to 6 custom Green Screen AI Avatars at a time. Once you make them, you can save these to your brand kit for use across any project for that brand, download them, or add them to a scene in your current video canvas project.