How to make viral faceless AI animations for $0

03 Nov 2026 04:07 25,384 views
You can now create short, faceless AI animations with voice, sound, and stickman visuals for $0 using a free open model and an AI agent that runs on your computer. This guide walks you through the full workflow, from story writing to directing and exporting vertical and horizontal videos.

Faceless AI videos are everywhere right now. Simple stickman stories are pulling in tens of millions of views, and many tools now charge $50–$149 a month to help you make them. The good news: you can build almost the same style of animation for $0 using a free open model, an AI agent on your computer, and a reusable "skill" that handles the hard parts.

What you’ll be able to create

The workflow in this guide lets you build a full short-form video in about 30 minutes:

• 20–30 seconds long
• Stickman-style animation with a consistent world and characters
• AI voiceover running locally on your machine (no voice subscription)
• Sound effects and basic music
• Horizontal and vertical versions, ready for Shorts, Reels, and TikTok

Everything except the model "brain" runs on your own computer: the drawing, voice, sound, and rendering. The brain itself is a free open model that you access via an AI agent called OpenCode.

What you need (all free)

You only need three pieces to make this work:

1. A story from an AI writer (or yourself)
Use a free chatbot like ChatGPT free tier or Claude free tier to write the script. This doesn’t touch your video-building usage, so you save your free model calls for the actual animation.

2. OpenCode as your local AI agent
OpenCode is a free, open-source AI agent that runs on your computer. It can read your files, call tools, and follow multi-step plans to build your video from start to finish.

3. A reusable "skill" folder
This is a pre-built folder that teaches the model how to make the video: the visual style, how characters move, how to structure the audio, and how to check its own frames before you ever see them. It’s what makes a free model behave like a specialist video tool.

Step 1: Write a short “what if” story

To save your free model usage for the heavy lifting, start by writing the story outside of OpenCode.

Open your favorite free AI chatbot and prompt it something like:

“Write a 30-second ‘what if’ story for a stick figure video. Around 60 words, short sentences, one idea per sentence. Make it engaging and easy to follow just by listening.”

You can use ideas like:

• What if you got paid for every word you said?
• What if a zombie apocalypse started today?
• What if everyone knew the exact day they would die?

Or tell the AI your niche (horror, finance, relationships, self-improvement) and ask for 10 "what if" ideas. Pick the one you would actually click on at midnight while scrolling.

Once you’re happy with the story, save the text. You’ll paste it into OpenCode later.

If you want more ideas on turning these concepts into content that can actually earn money, you might like this guide on real ways to make money with AI tools in 2026.

Step 2: Install and launch OpenCode

OpenCode is the AI agent that orchestrates everything. It connects to a free open model (the "brain") and then uses tools on your computer to draw, generate audio, and render.

Install OpenCode

• On macOS or Linux: open your terminal and paste the install command provided in the skill’s documentation or community post. Hit Enter and follow the steps.
• On Windows: OpenCode recommends using WSL (Windows Subsystem for Linux). Install WSL, then open a terminal inside WSL and paste the same command.

Start OpenCode and pick a free model

Once installed:

1. Open a new terminal window.
2. Type opencode and hit Enter.
3. If you see a "connect" or sign-in prompt, you can skip it for this setup.
4. Type /models search free to see available free models.
5. Choose a fast model (for example, something like GLM 5.3 Flash or a similar "Flash"/"Lite" model if suggested).
6. Press Ctrl + T until you see a mode like "High" or the recommended quality setting.

At this point, you have a free AI agent running on your machine. It can think and plan, but it still doesn’t know how to create faceless videos. That’s where the skill comes in.

Step 3: Install the faceless video skill

The real magic isn’t the model—it’s the skill. A free model is not as smart as top-tier models like Claude or GPT-4, but it doesn’t need to be if you give it a strong, opinionated skill that already knows how to build this exact type of video.

What the skill actually does

The faceless video skill is a folder that encodes everything learned from making many of these animations:

• The stickman visual style and world
• How characters move and react
• How to structure a script for the ear, not the eye
• How to generate and sync voiceover
• How to add sound effects and basic music
• How to check frames before showing them to you

Because the instructions are so detailed, the free model just has to follow them. That’s why this works without an expensive subscription.

How to install the skill

1. Download the faceless video ZIP file from the shared drive or community post.
2. Save it into your project folder. Do not unzip it—OpenCode can read it as-is.
3. Go back to your OpenCode terminal.
4. Paste the provided install prompt (usually something like “install this skill from faceless_video.zip”).
5. Let it run. It will set up voice models, tools, and dependencies (around 1 GB).
6. Wait for the self-test to pass. If it asks about a large music model (around 15 GB), you can skip that for now if you want to save space.

Once the test passes, your environment is ready to generate faceless videos.

Step 4: Generate your first 30-second stickman video

Now it’s time to turn your script into an actual animation.

Feed in your story and shots

1. In OpenCode, type /new to start a new task.
2. Paste your full story from earlier.
3. (Optional but powerful) Add rough shot ideas under the story, like:
• 0–3s: Close-up of character with word counter on screen
• 3–10s: Wide shot of character talking, counter rising fast
• 10–20s: Partner in background on phone, counter slows
4. Then write a clear instruction, such as:
“Make this a 30-second stickman video, horizontal and vertical, using the faceless skill.”

OpenCode will show you a plan. Review it quickly, then tell it to go ahead.

After a few minutes, you’ll get a first version: animation, voice, sound, and exports.

How to give notes like a director

Your first version probably won’t be perfect—and that’s normal. Treat the AI like a junior editor and give it clear, timestamped feedback.

Tips for effective feedback

1. Start with 30 seconds and one video at a time
Longer scripts and multiple jobs in parallel are more likely to hit free usage limits or break mid-run. Begin with a single 30-second video to keep things stable.

2. Always provide the full story and shot list
Don’t send just one line at a time. The model does better when it sees the entire story and your rough shot ideas together.

3. Use timestamps in your notes
Instead of “make the counter bigger,” say: “At second 5, make the money counter much larger and easier to read. It should be the main focus of the screen.” This makes it easier for the agent to map feedback to specific frames.

4. Watch the video yourself
Sometimes the AI will say “everything looks good,” but details will still be off. Always watch the export with your own eyes before you consider it done.

With a few rounds of notes, you can usually get from a rough first pass to a clean, publishable short—all still at $0 while your free usage lasts.

How the voice and sound work (no subscriptions)

One of the biggest advantages of this setup is that the voice runs locally on your computer. That means:

• No 11 Labs subscription
• No API keys to manage
• No credit-based billing for voiceovers

The skill configures a local text-to-speech model and uses it to narrate your script. It then layers in sound effects and optional music using tools that run on your machine.

The catch: free usage limits on open models

There is one important limitation: open models are free only up to a point.

• The model provider enforces a free usage limit, but the exact number of calls or tokens isn’t always clearly documented.
• In practice, you might be able to run for an hour or two without issues—or you might hit a limit after running multiple videos in parallel.

For example, running two videos at once for under an hour can be enough to trigger a “free usage limit reached” message. Other times, you might get a couple of hours of uninterrupted use.

The takeaway: this is a great way to start and experiment, but it’s risky to build an entire channel on free usage alone.

What to do when you outgrow the free tier

Once you know you want to publish regularly, it’s worth paying for a stable setup:

• OpenCode’s paid option (like OpenCode Go) is around $10/month.
• With pay-as-you-go pricing on the backend model, you might pay roughly $0.60 per video for 100 videos, depending on your usage.

By then, you’ll already have your skill, your workflow, and maybe even your first audience. Upgrading just gives you more reliability and scale.

If you’re serious about using AI to ship content and projects faster, you may also find it useful to explore more advanced AI agent skills, like the ones covered in this breakdown of Claude Code skills.

Hardware and platform requirements

You don’t need a monster PC to run this, but there are a few basics:

• Computer, not phone: This does not work on mobile. The agent needs access to your file system and folders.
• Disk space: Allocate at least a few gigabytes. The tools and voice models are the main weight; the model brain itself is remote.
• Operating system: Works on macOS, Linux, and Windows (with WSL). It can also run through Chrome if installed correctly.

Common questions

Do I need my own visual style or Hex-style config?
No. The skill already comes with multiple pre-made stylesheets (around five at the moment), including a tested stickman style. You can start with those and customize later if you want.

Does this require any paid API keys?
No for the basic setup. The free open model, the local voice, and the tools used here do not require an API key. If you later want custom image styles via third-party APIs, those would be pay-as-you-go.

Can I run this entirely in the browser?
Not fully. Even if you trigger it from a browser, the tools need to run on your local machine with access to files and folders.

Can I use other AI video tools instead?
Yes—there are many tools that do something similar but charge monthly. This guide focuses on a $0 path using open models and local tools so you can test your ideas before committing to a subscription.

Why this is a powerful way to start

This setup is ideal if you’re just getting into AI content creation and don’t want to spend money before you see results:

• You learn how AI agents, skills, and tools actually work together.
• You get a repeatable process to turn ideas into short, faceless videos.
• You can experiment with different hooks, niches, and storytelling angles with no upfront cost.

Once you find a format that resonates, you can decide whether to keep iterating with free models or move to a paid setup for more speed and reliability. Either way, you’ll already have the most important piece: a skill-driven workflow that turns text into engaging, watchable animations.

Share:

Comments

No comments yet. Be the first to share your thoughts!

More in Video Generation