Vsub

Vsub is an AI narration tool. A faceless video generator that turns a script or voiceover into a narrated, illustrated video. Its template system clones the drawing style of any reference clip you upload, and the whole job can also be triggered from Claude, ChatGPT or Cursor over MCP.

Vsub is a browser based production tool for faceless video channels. The premise is simple: you supply a script or a voiceover, and the platform breaks it into scenes, generates narration, draws each frame and cuts the result into a finished video. The distinguishing feature is the template system. Instead of offering a fixed library of visual styles, Vsub lets you upload a reference video you already like and clones its drawing style into a reusable template that stays in your workspace. The workflow is built around a paste and generate loop. A script can run up to 50,000 characters, though the finished output must fit inside the maximum length allowed by your plan. Narration is generated automatically, or you can upload your own recording, in which case the script is transcribed from the audio and no synthetic voice is layered over it. Scenes are drawn with the chosen template, and the assembled video opens in an editor where individual scenes can be redrawn. Failures are recoverable: a retry resumes from the step where generation stopped, and the voiceover, scenes and images already produced are preserved and not billed again. Ready made templates cover the styles that dominate faceless YouTube: stickman explainers, history and ancient themed drawings, finance clips, cartoon, paper diorama, quiz and commentary formats. These can be used as they are, so a first video needs no template work at all. Custom cloning is there for creators who want a look nobody else is publishing. Reference clips work best when they contain spoken narration and still or moving footage, run at least 30 seconds, and sit around three minutes for the strongest match. Beyond the web editor, Vsub exposes an API endpoint for creating videos from a script and template, and an MCP connection that lets assistants such as Claude, ChatGPT or Cursor request a video directly. That puts it in the company of other script to video generators, but the template cloning and the assistant driven path are what separate it. It suits solo faceless channel operators and small teams publishing frequently, less so anyone who needs hand authored motion design or frame level creative control.