How to Use HeyGen: Your First AI Avatar Video, Step by Step

The quickest way to use HeyGen: sign up free, open Create > Create a Video, pick an avatar and a voice, write or paste a script in AI Studio, then press Generate and wait for the render. The Free plan gives you up to 3 watermarked videos a month, each up to 1 minute, for personal, non-commercial use only. Anything you publish for a business needs a paid plan, and on paid plans every minute of avatar video spends credits. The engine you choose sets that cost. Avatar III with a video look uses 4 credits a minute; Avatar V uses 48. Decide that first.

What to settle before your first video

Start on the Free plan. It needs no credit card, and HeyGen offers no self-serve trial of a paid plan, so Free is the way to learn the editor. It covers AI Studio, a library of 500+ stock avatars, one Digital Twin slot, three Photo Avatars, one voice clone and three Video Agent videos a month, and it uses no credits at all. Its limits are the short, watermarked videos and the licence, which rules out ads, client work and anything you sell. Treat Free as a rehearsal space, not a production tool.

HeyGen's pricing page lists Creator at $29 a month billed monthly, or $24 a month billed yearly ($288 a year), with 600 credits a month, 1080p export and videos up to 30 minutes.^[1] Credits, not the plan price, decide how much video you actually get, and our HeyGen pricing guide maps every tier along with the rollover and expiry rules. Creator also removes the watermark and allows commercial use. Pro adds 4K export and Proofread for translations. Business adds seats, SSO and 60-minute videos.

Start a project in AI Studio or Video Agent

Every video begins at Create > Create a Video, where you choose landscape or portrait and a starting point: a blank project, photo to video, a template, a PPT or PDF upload, or Video Agent. The first fork matters. AI Studio puts you in control of every line; Video Agent drafts the whole thing from a prompt and hands you something to correct.

AI Studio is the hands-on route, and it is script-based rather than timeline-based. You write the narration, and HeyGen times scenes and elements around it. If you come from a timeline editor, expect an adjustment. You set when an image or text block appears by dragging two markers inside the script, not by sliding clips along a track. Projects save automatically and keep a version history, so experimenting is cheap.

Video Agent is the hands-off route. You describe the video and can attach decks, PDFs, images or clips; it drafts the script, scenes, visuals, voice-over and captions. You approve a Video Plan before anything renders, and planning and chatting cost nothing. Rendering costs about 80 credits a minute in Standard mode, or about 240 in Seedance mode, a Pro feature. HeyGen shows the total estimate on a confirmation card first. Lengths run from 15 seconds to 3 minutes, or Auto; some lengths need a paid plan. When the draft is close, use Edit a Copy to finish it in AI Studio.

Choose the avatar, the engine and the voice

AI Studio opens with a default stock avatar named Annie; swap in any public or custom avatar. Then pick the engine, because it decides both realism and cost. Avatar III loops existing footage with lip sync and is the cheapest, at 4 credits a minute with a video look or 7 with a photo look. Avatar IV, the default engine, animates a single photo and accepts motion prompts, at 16 credits a minute with a photo look or 31 with a video look. Avatar V gives the highest fidelity at 48 credits a minute and needs a video look. In AI Studio, Avatar IV charges only for the seconds of avatar generation.

Voices come from the Voice Library, your own clone, an AI-designed voice or an integration such as ElevenLabs. You can also upload or record real audio instead. Voice Director adds tone presets such as Calm or Excited and needs the Panda voice engine. Voice Mirroring shows your line like a teleprompter, and the avatar copies your pacing and emotion from the recording. Previewing voices costs no credits. If a clone sounds flat, Improve Voice lets you compare engines and save the best one. Our HeyGen AI avatar guide covers every avatar type, and our voice cloning guide covers clones.

Making your own Digital Twin

To appear as yourself, open Avatars and choose Clone a Real Person. For Avatar V you record about 15 seconds of continuous footage facing the camera; a webcam or phone works. A consent clip follows, in which the person on camera reads HeyGen's text and a four-letter code. It must be the same person as the footage, and screen recordings or AI clips fail. The twin is usually ready in about 10 to 20 minutes. Fill the frame with your face: when it takes up too little of the base look, lip sync degrades. HeyGen strongly recommends recording a voice clone at the same time. You can redo an avatar once a month.

Write the script the way the editor reads it

Type, paste or upload the script, or let the AI Script Writer draft one from a short description of your idea. Keep each section under 2,000 characters. An import splits any line or paragraph over 1,000 characters into separate scenes, so break long paragraphs yourself where you want the cuts. Uploads accept .docx but not .doc files. Each scene holds one character. Two presenters mean two scenes.

Pauses go in 0.5-second steps. A pronunciation fix applies only to that word in that one scene, so recurring brand and product names belong in the Brand Glossary, which applies across the project. Avoid mixing languages in a single script. HeyGen warns that the voice output turns unpredictable, and it advises separate scenes or separate projects for each language instead.

Around the script, AI Studio adds stylized captions synced to each word, a music library that also takes your own tracks, transitions such as Fade, Slide, Zoom and Pop, and, on Business and above, a built-in screen recorder that turns a capture into an editable scene. Images and video clips upload at up to 200 MB each, audio at up to 100 MB. Captions are burned into the picture. Avatar videos get no closed-caption track that viewers can switch off.

Generate, wait, then share and translate

Here is the part that surprises new users. The editor never shows the moving avatar. You can preview the audio, but lip sync renders only after you submit.^[2] So check the script by ear before generating. Most edits are free: images, layouts, captions and timing cost nothing, and neither do voice previews. Re-rendering an avatar scene after a script, avatar or voice change spends credits again.

When you press Generate, you set the title, folder, frame rate (25, 30 or 60), resolution and format, and decide on the watermark. Any paid plan can switch the HeyGen watermark off at this step, and it stays off for later videos until you turn it back on.^[3] Creator exports at 1080p; Pro, Business and Enterprise reach 4K. A submitted video cannot be edited in place. Duplicate it to change anything.

Rendering takes time. Free processes one video at a time, Creator and Pro three, Business six, and extra videos wait as Pending. HeyGen's help center says the regular queue averages just under 24 hours on weekdays. Moderation can also hold a video. Failed or cancelled videos are never charged. By our arithmetic, a two-minute Avatar IV video with a video look uses 62 credits, the same video on Avatar V uses 96, and Creator's 600 monthly credits cover about 150 minutes of Avatar III with a video look. As a ceiling they equal about 12.5 minutes of Avatar V, but Creator includes Avatar V only with limited usage.

The finished video opens on a share page that tracks views, average watch time and completion rate. Downloads let you pick the resolution and take the full video, the captions only or the audio only; sharing covers an embed, a link, email, LinkedIn and a GIF. Translating from the same page creates a new draft in the chosen language. Translation costs 4, 6 or 10 credits a minute per language for Audio only, Speed or Precision, with up to 10 languages per job and Proofread on Pro and above. Only spoken audio is translated, never on-screen text. Our video translation guide walks through each engine.

The bottom line

Learning HeyGen takes an afternoon on the Free plan: one project in AI Studio, one avatar, a short script and a render. The habits that save money come next. Draft and check scripts by ear, because the moving avatar appears only after submission. Pick the engine on purpose, since Avatar V costs twelve times Avatar III with a video look. Leave slack for the queue. Move to a paid plan when the video is for business use, runs past a minute or needs to lose the watermark, then size the plan by credits per minute, not by its monthly price.