AI Avatar Video Generator: Create Talking Avatar Videos in Cliptude
Cliptude's Avatar Video format combines AI avatar scenes with the platform's existing documentary pipeline. Instead of turning every scene into a presenter shot, Cliptude selectively adds a talking avatar video where it improves clarity, then mixes those scenes with maps, charts, motion graphics, B-roll, text reveals, and scene-based voiceover. The result is a more flexible AI avatar video generator workflow for YouTube videos, explainers, internal training, and AI spokesperson content.
Example video
Sample Avatar Video directed by Claude Opus 5.5: talking-avatar scenes mixed with data boards and B-roll.
What avatar video generation means in Cliptude
Cliptude does not treat avatar video generation as a single static presenter track. It uses an avatar as one visual layer inside a broader edit. That means a scene can be full-screen avatar, split-screen avatar with B-roll, avatar with text reveal, or a normal non-avatar scene if maps, charts, or footage tell the story better.
This makes the format especially useful when you want a presenter-led feel without sacrificing visual variety. It is a good fit for avatar videos for YouTube, software explainers, product education, onboarding videos, compliance training, course modules, sales enablement, and AI spokesperson clips that still need supporting visuals on screen.
How the avatar video workflow works
Choose format
In Create Video, pick Avatar Video instead of a visual-only format.
Pick an avatar
Use a public look, or upload a photo to create a reusable presenter.
Lip-sync the voice
Scene voiceover is reused as-is for lip sync — no second narration track.
Mix the visuals
Only some scenes get a presenter. The rest can still use charts, maps, B-roll, and text.
In the Create Video setup popup, Avatar Video adds an Avatar tab right after Basics (Choose your presenter). Two switches sit at the top: On screen — Throughout (default) or Intro only, about 8 seconds at the start and then documentary style; and B-roll — Stock footage (default) or AI footage, AI video and images with your avatar in some scenes. Below, pick a presenter from Built-in, Favourites, My avatars or Public, or click Create an avatar. Every avatar renders with HeyGen Avatar IV; it is the only avatar engine offered.
Earlier versions called these switches Avatar Intro Only and AI Image + AI Video; the API flags keep those names. REST, MCP, and CLI inherit them from a saved template and can override them per job.
Make it from Claude or ChatGPT (MCP)
Connect the Cliptude MCP server (https://mcp.cliptude.com/mcp) to Claude, ChatGPT or another agent and
ask for the video in plain words. In Claude and ChatGPT, Cliptude's tools open pickers and cards inside the chat.
This is the order the agent follows:
- Your templates first.
list_templatesshows your saved templates as a gallery. Pick one. - No template fits? The agent builds one with the pickers, one at a time:
voice, then the presenter (
list_avatars), then music, background and style. - Type next. In Claude a click saves your pick and the picker says
"Saved. Type next in the chat to continue." Type
nextand the agent reads it. In ChatGPT the pick is sent as your next message. - Script. The agent researches the topic, writes the script and lets you read it.
- Review card. Every setting, any warnings, a checkbox for anything you turned off, and the credits, with an Every setting tab and a Script (N words) tab. Nothing is built or charged until you click Confirm.
- Job card, then the publish kit. A job card shows progress and, when the job waits for it, the script to approve. When the video is ready, the publish kit plays it and gives you thumbnails, titles, a description, chapters and tags to copy.
Make me a 10-minute Avatar Video about [topic]. Show me my saved templates first and let me pick one. If none fits, set one up with me using Cliptude's pickers, one at a time (voice, then the presenter, then music, background and style). Write the script, let me read it, and show me the review card before you build anything. Show me the publish kit when the video is ready.
Claude Code, Codex, Cursor and other agents that show no widgets give you a cliptude.com/pick link with the same
picker instead. It works once, for 15 minutes, and only
for you when you are signed in. The agent reads the result with get_pick. From a terminal:
cliptude pick <pick_id> --url <link> opens it and waits.
Setup for each agent: Claude, ChatGPT, Claude Code, or any other MCP agent.
Public avatars vs photo avatars
| Avatar type | Best for | How it works | Tradeoff |
|---|---|---|---|
| Public avatar | Fast presenter scenes from the built-in catalog | Select a look and use it immediately | A shared look, ready without an upload |
| Photo avatar | A reusable presenter from your own image | Upload a photo, wait until it is ready, then reuse it across videos | Relies on the quality of the source photo |
If you want the fastest path to a talking avatar video generator workflow, start with a public avatar. If you want your own on-screen presenter identity, use a photo avatar and keep reusing it.
Create avatar videos from a photo
Cliptude supports the create avatar video from photo workflow directly in the avatar selection step. Upload a clean image, wait for the photo avatar to become ready, then reuse that avatar in future projects without recreating it each time.
- 1 Go to the avatar selection step after choosing the format and voice.
- 2 Upload a front-facing image with clear lighting and a visible face.
- 3 Wait until the avatar is marked ready, then select it for the video.
- 4 Continue with background selection and finish the normal Cliptude workflow.
Voiceover and lip sync
Cliptude does not create a separate narration just for avatar scenes. It uses the same scene voiceover clips already generated for the video. That gives the avatar renderer scene-accurate timing and keeps the spoken pacing aligned with the rest of the edit.
- • Existing scene-level voiceover is reused for avatar lip sync
- • Only avatar scenes are sent for avatar rendering
- • Long final videos are supported because each avatar request is scene-based, not full-video based
- • Voiceover can still be part of a mixed video with B-roll and overlays
If you want to bring your own narration instead of generated speech, see Voiceover Upload and Voiceover Selection.
Avatar videos for YouTube, training, explainers, and marketing
Avatar videos for YouTube
Use presenter scenes to introduce sections, bridge narrative transitions, or appear on-screen during commentary while the rest of the video stays documentary-driven with maps, archive footage, charts, and text overlays.
AI spokesperson video for products
Use a public or photo avatar as the face of a launch video, landing-page explainer, product walkthrough, or paid social creative where a consistent presenter improves trust.
Training and internal enablement
Training teams can use avatar scenes for intros, process overviews, or compliance summaries while retaining diagrams, checklists, and screenshots in the companion visuals.
Explainers and educational content
Use split-screen layouts to keep a presenter visible while revealing bullet points, product interfaces, step-by-step demos, or visual evidence on the other side of the frame.
How Cliptude decides which scenes get avatars
What gets selected
- • Introductory sections that benefit from a presenter-led setup
- • Transition scenes where a human-like narrator adds clarity
- • Sections that work well as full-screen presenter or split-screen explainer shots
- • Scenes whose voiceover clip is short enough to stay inside the avatar renderer's request limits
What stays non-avatar
- • Scenes where charts or maps communicate the idea more clearly
- • Long stretches that would feel repetitive as presenter-only video
- • Shots where B-roll or animated overlays carry the story better
- • Any avatar scene that fails validation or rendering, which is demoted back to a normal scene instead of breaking the whole video
The goal is visual diversity, not wall-to-wall avatar footage. That is why avatar scenes are mixed into the broader documentary edit instead of replacing every visual.
Avatar Intro Only
On the Avatar tab, set On screen to Intro only (the default is Throughout; the API calls it avatar_intro_only). When enabled, HeyGen appears only at the beginning of the video — typically the first spoken sentence, and never shorter than about five seconds so the presenter is clearly visible. The opening window can still render full-screen or as a split/collage layout. Everything after that intro stays documentary-style without additional avatar scenes.
- • Works with built-in public avatars, public catalog avatars, and your own photo avatars
- • Mid-video and closing presenter windows are skipped
- • Ideal when you want a presenter-led hook without paying for intermittent avatar coverage across the full runtime
- • REST, MCP, and CLI support this toggle: save it on the template, or override a job with
avatar_intro_only/--avatar-intro-only
AI Image + AI Video
On the Avatar tab, set B-roll to AI footage (the default is Stock footage; the API calls it AI Image + AI Video). When enabled, Cliptude skips stock and YouTube B-roll and generates AI video clips and AI stills instead. The selected avatar can appear inside some of those generated scenes — in the environment, doing a task — not only as a talking overlay. Other scenes illustrate whatever the script is talking about without that face. HeyGen still handles real lip-sync talking windows.
Sample on YouTube — Avatar Video with AI Image + AI Video (generated AI clips and stills plus talking-avatar windows).
REST, MCP, and CLI support this toggle. Saved templates store it (and Avatar Intro Only). Jobs inherit the template unless you override that job with avatar_ai_footage / --avatar-ai-footage (omit to keep the template). See API, CLI, and MCP.
- • About 5 seconds of each finished minute is generated video; the rest is AI stills with camera moves
- • The talking avatar opens and closes the video, and on longer videos appears for a few seconds about every seven minutes in between
- • Charts, animations, and text overlays still follow the visual-enhancement toggles you already set
- • This option is 52 credits per finished minute. It replaces the documentary minute rate and any avatar surcharge, including when Avatar Intro Only is also on
Sample channels
AI Image + AI Video is a strong fit for presenter-led channels that need kitchens, workbenches, gardens, or other sets you would otherwise have to film. The same avatar can appear inside those generated scenes, with a consistent face and voice from episode to episode.
Cooking & Recipes
A weekly cooking show with one host at the counter — chopping, plating, and talking through the dish in a generated kitchen.
DIY & Home
Long repair and build videos where the presenter works the project on camera, without a workshop or a camera crew.
Gardening
Planting calendars, pruning demos, and plant-care series that stay on a seasonal upload rhythm.
Education
Lesson-style explainers from a written script, delivered by the same on-screen tutor every episode.
Health & Lifestyle
Routines, wellness advice, and habit breakdowns that look filmed in real rooms — without booking a studio.
How-To Videos
Any process you can write as a script becomes a long walkthrough, published as often as you have the next topic.
Scene layouts: full-screen vs split-screen avatar scenes
Cliptude supports multiple avatar layouts so the video does not feel static. Some scenes render the avatar full-screen, while others use split-screen compositions that pair the presenter with motion graphics, text reveals, maps, or B-roll.
Full-screen avatar
Best for intros, direct address, concise explanations, and spokesperson-led scenes where the presenter should dominate the frame.
Split left / split right
Best for product explainers, tutorial-style content, and educational scenes where the avatar should stay visible while the evidence or visuals appear beside them.
Avatar plus text reveal
Best for definitions, frameworks, takeaways, and stat-driven scenes where the narration should stay conversational but key points still need to land on-screen.
Best practices for better avatar videos
- → For photo avatars, use a sharp, front-facing image with even lighting and minimal background clutter.
- → Pick one look and keep it across the video so the presenter stays recognizable.
- → Let documentary visuals handle dense evidence, timelines, maps, and comparisons instead of forcing every scene into presenter mode.
- → Keep intros, transitions, and CTA moments presenter-led for a stronger sense of continuity.
- → If you need a real uploaded presenter clip instead of an AI avatar, use Talking Head Upload.
Pricing, credits, and key options
AI Image + AI Video is 52 credits per finished minute. That price replaces the documentary minute rate and any talking-avatar add-on, including when Avatar Intro Only is also on. Avatar Intro Only is 55 credits per finished minute on Cliptude\'s HeyGen key, the same as a talking avatar throughout. Creating a photo avatar with the system key is a one-time 55-credit charge. Reusing that avatar does not repeat the fee. A HeyGen key you bring yourself skips the photo-avatar creation fee.
What does a minute cost?
Drag to a monthly minute target. The panel picks the smallest plan that covers it.
Minutes are the full monthly credit balance spent on that option alone. See Credits System and Pricing & Plans.
Frequently asked questions
Can I create an avatar video from a photo?
Yes. Cliptude supports photo avatars, so you can upload a photo, wait for the avatar to become ready, then reuse it across future avatar videos.
Can I use my own HeyGen key?
Yes. Cliptude supports a bring-your-own-key workflow and stores user keys in encrypted form. AI Image + AI Video stays at the flat finished-minute rate either way. A key you bring yourself skips the photo-avatar creation fee.
Do all scenes become avatar scenes?
No. The system uses only selected scenes for avatar rendering so the final video keeps visual diversity and can still use maps, charts, animations, and B-roll where those visuals are stronger.
How many minutes of avatar video do I get on each plan?
With AI Image + AI Video at 52 credits per finished minute: Starter about 21 minutes ($0.90/min), Lite about 55 minutes ($0.89/min), Creator about 135 minutes ($0.88/min), Pro about 365 minutes ($0.87/min), Scale about 1,010 minutes ($0.87/min). A talking avatar or Avatar Intro Only (55 credits/min) is about 20, 52, 127, 345, 955 minutes on those same plans. Use the calculator above to match a minute target to the smallest plan that covers it.
Can I skip stock footage and use only AI visuals?
Yes, on Avatar Video. Set B-roll to AI footage on the Avatar tab, or on a saved template. REST, MCP, and CLI inherit the template and can override per job with avatar_ai_footage / --avatar-ai-footage. Cliptude then skips stock and YouTube B-roll and generates AI video and AI images instead. The selected avatar can appear in some of those scenes; other scenes illustrate the script subject without that face. That option is 52 credits per finished minute.
Can I use an avatar only for the intro?
Yes. On the Avatar tab, set On screen to Intro only, or set it on a saved template. REST, MCP, and CLI inherit the template and can override per job with avatar_intro_only / --avatar-intro-only. The avatar appears only at the start of the video (full-screen or split/collage), for roughly the first sentence and at least about five seconds, then the rest of the edit continues without additional HeyGen scenes.
Can long videos still use avatar scenes?
Yes. Cliptude renders avatar scenes individually with scene-based audio clips, then composes them back into the full video. The final video can be much longer than any individual avatar scene.
What is the difference between public and photo avatars?
Public avatars come from the built-in catalog and are ready to use immediately. Photo avatars are created from your uploaded image and are better when you want a reusable presenter that matches your own brand or persona.
Related guides
Video Formats
See where Avatar Video fits alongside Cliptude's other visual formats.
Voiceover Selection
Choose the narration style that will drive your avatar scene lip sync.
Voiceover Upload
Bring your own audio when you want narration control before the avatar scenes are rendered.
Talking Head Upload
Use a real uploaded presenter clip instead of an AI avatar when you want your own recorded footage.
Ready to create an avatar video?
Start from the Create Video flow, choose the Avatar Video format, select a public or photo avatar, and let Cliptude blend presenter scenes with the rest of your documentary visuals.
Create an Avatar Video