How to Make an AI Avatar Video in 3 Steps

An AI avatar video takes three steps: one good photo, one clean audio clip, and a generation. The whole thing fits inside a coffee break. Here is the full walkthrough, including the details that decide whether the lip sync lands cleanly or comes out mushy.

Step 1: Take a photo the model can animate

The model builds the speaking avatar from your photo, so the photo sets the ceiling. One person, facing the camera, with even light across the face. A waist-up shot works best: the avatar can move head and shoulders naturally instead of freezing below the neck. Skip sunglasses, hats that shade the eyes, mirror selfies, and anything saved as a screenshot. Compression kills the detail the lip-sync model needs.

Plain backgrounds help the model find the edges of the face. A portrait taken by a friend beats a selfie; arm's length distorts less than you would think.

Step 2: Record the audio and pick a scene

The audio drives both the speech and the final video length, so treat it like the script it is. Record close to the mic in a quiet room. Phone voice memos are fine; echoey stairwells are not. A 5-second clip, about two sentences, is enough for a first test, and the 10 free sign-up credits cover exactly that.

Scene presets frame the render: Clean Portrait for the classic talking head, Podcast Guest for interview energy, News Desk for briefings, Street Interview for vlog texture. There is also a performance mode (Dance Clip, Cinematic Scene) that renders a prompt-driven clip with an original AI instrumental when you want entertainment rather than speech. The step-by-step guide covers every option.

Step 3: Generate, check the sync, download

Hit generate and the render usually finishes within minutes. Watch it once with sound on: the mouth should track the words through the whole clip, not just the first sentence. If the sync drifts, the cause is nearly always the audio, so re-record cleaner and run it again. Failed renders refund their credits automatically.

Download the MP4 (no watermark, portrait or landscape) and post it. For Reels, Shorts and TikTok use the 9:16 portrait; for YouTube, 16:9. Want a deeper version of any step? The generator explainer has the pricing tables and model details.