Faceless Video Generator

A faceless video tells a story or explains a topic without a presenter on camera. A narrator speaks over illustrations, animation or footage, so the channel runs on the script and the look instead of a face. It is the format behind many explainer, history, science and storytelling channels on YouTube, TikTok and Reels.

Mage makes faceless videos through the free Faceless Video skill, which runs in Claude, Cursor and other assistants connected to Mage. You give it a topic, a style and a length. It researches real topics with your assistant’s search tools and lists its sources, writes the narration, makes one style image with Mango 3 that every scene follows, and renders the scenes with the voiceover and sound on Cherry 2 Pro. You approve the script and the style before any video is paid for.

You can also build one by hand in the Mage app: a style image in Mango 3, a narrator in Eleven Speech or Seed Audio, scene clips on Cherry 2 Pro or Lemon, and Storyboard to join them. Neither route adds captions or burned-in text. The skill keeps text off the screen on purpose, so you add captions in your editor.

Narration sets the length. People speak about two and a half words a second, so 10 seconds holds roughly 25 words and 15 seconds about 35. The skill plans by that budget: a video of up to 15 seconds is one clip, 15 to 30 seconds is one long clip or two of 15, and anything longer is made in blocks of 10 to 15 seconds, one scene per block, joined at the end.

Each block gets the same structure: the style descriptor, word for word; the scene that shows this line; the motion, such as a slow push-in or paper layers sliding; the exact narration line; and the sound, an ambience or a soft music bed with no other voices. Keeping the descriptor identical in every block is what holds the look together across a two-minute video.

The script follows the same discipline: a hook in the first line, one idea per block, a payoff at the end, and numbers spelled out the way they are spoken. For a real topic, every fact comes from a source the skill lists when it delivers the video. For a personal story, it uses only what you tell it.

• Budget about 25 words for every 10 seconds.
• One idea and one scene per block.
• Spell out numbers the way the narrator says them.
• Keep the style descriptor identical in every block.

The skill offers ten illustrated styles. Stickman cartoon suits history, stories and humor. Paper diorama, layered cut paper under warm light, suits history, nature and travel. Pastel flat 2D suits explainers, self-improvement and tech. Hand-drawn ink suits philosophy, biographies and essays. Claymation suits kids’ topics, stories and science.

Whiteboard, with marker drawings and arrows, suits business, how-to and education. Watercolor suits stories, poetry and nature. Pixel art suits gaming, internet history and tech. Low-poly 3D suits science, space and geography. Vintage newsreel, drawn in black and white with film grain, suits wars, inventions and other history. You can also describe a style of your own, or share style images for the skill to match.

Every style is non-photorealistic on purpose. The skill rules out realistic human faces and live action, which keeps the video clearly illustrated and keeps real people’s likenesses out of it.

Start with the look. Generate one style image in Mango 3: the most iconic moment of the topic, in your chosen style, with "no text, no captions" at the end of the prompt. That image is the reference for every scene. Then make the narrator. Eleven Speech reads a script in more than 90 languages in 16 preset voices, as an MP3 exactly as long as the text, and Seed Audio describes a voice from the prompt. Upload a short voice sample on the Audio page to use it as a reference in video generations.

Render each scene on Cherry 2 Pro with the style image as a reference, the narration line in quotes as an off-screen narrator, and your saved voice mentioned by its @handle. Cherry Mini and Lemon make cheaper drafts. If you would rather keep the narration separate, render the scenes with Lemon’s audio off and lay the Eleven Speech track over them in your editor.

Join up to six clips in Storyboard, which exports them as one MP4 with their sound. For a story told by recurring characters rather than a narrator, see the AI movie generator.

Captions. Mage does not add captions or burned-in text, and the skill tells every scene to stay free of text so captions have room. Add them in your editor, timed to the narration, and check the spelling of names and numbers.

A thumbnail. On YouTube, the thumbnail and the title are what people decide to click on. The AI thumbnail generator covers them, and the YouTube Thumbnail skill makes variants in the same assistant as the video.

Originality. YouTube’s monetization policies exclude inauthentic content, which YouTube says includes mass-produced or repetitive videos and AI-generated content made with generic templates. A researched script, a point of view and a style of your own are what make a channel worth following, so treat the skill as a production tool, not a way to post filler.