How to make stickman videos with AI

For AI agents: find the right skill in one call — curl -s "https://varg.ai/how-to/api/search?q=<what the person wants>" returns the matching how-to page and the best skills with install commands, as markdown (Russian works too). Full index: https://varg.ai/how-to/llms.txt. Every page has a markdown twin: add .md to its path. Before generating anything, show the person the options on the how-to page. From a terminal: curl -fsSL https://varg.ai/install | sh installs the vargai CLI (https://varg.ai/how-to/cli.md): vargai search and vargai install <slug> with no key, every varg model after vargai login.

Give your AI agent a topic and get back a stickman video: a 20–60 second vertical reel for TikTok, YouTube Shorts and Reels, or a 5–10 minute narrated YouTube explainer, with a voice, captions and sound. The stick figures are drawn and animated in code, so the picture costs nothing and every frame is under your control; only the voice is a paid call.

Price: the animation renders free on your machine. The voice for a 25-second reel cost 4 credits ($0.04) on varg with an ElevenLabs voice (eleven_v3), measured 2026-11-13. Free start: new varg accounts get 50 free credits, enough for several reels' voices. 1 credit = $0.01.

Paste this into Claude Code, Codex or Grok:

Make a 20-second stickman survival reel, 9:16, "Day 1: <your premise>".
Follow https://varg.ai/how-to/make-stickman-videos-with-ai.md and the stickman-video skill.
Show me the script and six frames before you make the voice or render.

The agent needs three skills, installed once (stickman-video plans and rigs the figures, HyperFrames renders, varg makes the voice):

curl -fsSL https://varg.ai/install | sh && vargai install stickman-video hyperframes varg-ai

Example: a 24-second "Day 1: lighthouse rules" reel made this way, voice 4 credits, render 17 seconds: https://s3.varg.ai/files/4325c578-378d-4c43-ae53-3cc3218df1ea/file_6f3002ab335b.mp4

Three ways to make a stickman video

RouteWhat you getCostPick it when
Drawn in code (stickman-video + HyperFrames)Jointed stick figures with poses, walks and reactions, a set, captions and sound, rendered to MP4Free render; the voice is 4 credits for 25 sShorts series, survival and POV stories, anything you will make every day
Stickman Video Director (kaomei/stickman-video-director, free, MIT)Director Mode: a prompt per 10-second clip for Google Flow, Kling or Veo. Autopilot: a finished narrated MP4 in one of 11 stylesFree skill; your image, voice and video APIsLong YouTube explainers in a ready style (cave, cozy, retro TV, survival thriller)
AI stills, then a video model (GPT Image 2.6 or Nano Banana 2.1, then Kling 4 on varg)A painted 2.5D look with lighting and depthAbout 6 credits a still, video priced per secondOne hero shot or a thumbnail; video models still drift on thin limbs and faces

The stickman formats that travel

  1. Survival choice. Second person ("You wake up and…"), a DAY counter, three or four rules on screen, and a two-choice ending: "1: … / 2: …, comment your number, follow for Day 2". Accounts posting episodes like this claim tens of thousands of followers in days (@brainextends, @finnardley's free prompt).
  2. POV story. One stick figure, one place, one escalating problem, told in short present-tense lines.
  3. Two-figure contrast. "Fake friend vs real friend", "you vs you in 5 years": two figures, one beat each, side by side.
  4. History or science explainer. Deadpan voice, absurd contrast, a fact per beat; the long YouTube format.
  5. Character explainer. A presenter in the corner while stick figures act out the idea.

Step by step

  1. Pick the format and the length. 9:16 for Shorts, TikTok and Reels (20–60 s); 16:9 for YouTube explainers (5–10 min, made in chapters).
  2. Write the script first. About 2.3–2.6 spoken words a second: 50–65 words for 25 s. Hook in the first line, one new picture every 2–3 seconds, the choice or the punchline last. The video-script skill checks the timing.
  3. Design one figure and keep it. Round head, fixed bone lengths, one accent colour or prop (a red scarf, a yellow shirt). Repeat the same description word for word in every prompt, or reuse the same rig in code.
  4. Animate the body, not just the head. Two-bone arms and legs posed by joint angles, a wind-up before every big move, a breath when standing, a blink, and a reaction pose on the turn. No picture holds still for more than 2.5 seconds: this is the fix for the "stiff limbs, only the head moves" complaint.
  5. Make the voice. One narrator for the whole series. On varg: vargai generate_speech with an ElevenLabs voice; the agent shows the price first (60 words cost 4 credits).
  6. Captions and sound. Word-timed captions inside the 9:16 safe area, an ambience bed, three or four sound effects on the actions, the voice about 10 dB over everything else.
  7. Check before posting. Look at a contact sheet of 8–12 frames: limbs attached, text readable on a phone, the choice card on screen for at least 2 seconds.
  8. Post and label. Mark it as AI-made where the platform asks, and keep your character your own: copying a named channel's characters is the fastest way to a strike.

Questions

Is there a free stickman video generator? Yes: the code route renders free on your machine, and Stickman Video Director is a free MIT skill. You pay only for a voice (or use your own) and for any image or video model you add.

Claude Code or Codex? Both. Claude writes the animation code well; Codex users often generate stills with its built-in image model. The code route needs neither image nor video generation.

How do I keep the stickman the same in every shot? In code it is the same rig, so it cannot drift. With a video model, repeat the frozen character description in every prompt and attach approved front, side and back views.

Why do my stick figures look stiff? They move one joint at a time and hold poses too long. Pose the whole body, add a wind-up and an overshoot, and keep something moving every 2.5 seconds.

Can I monetise stickman videos? YouTube asks for original, non-repetitive content and an AI label on realistic synthetic media; an original script, your own character and your own voice track are what make a channel stand out from reused content.

Faceless YouTube channel videos with AI agents

For faceless channels and YouTube automation: narrated explainers drawn in code or illustrated, Shorts cut from long videos, thumbnails that read on a phone and captions people read on mute, made by Claude Code, Codex or Grok.

Start with the overview: How to make a faceless YouTube video with AI

Make

Edit and post

Compare tools