This guide takes you from zero to your first AI-generated video. No prior video editing experience required. Total setup time: about 10 minutes.

Prerequisites

Before installing VidNo, make sure your system meets these requirements:

  • Operating System: Linux (Ubuntu 20.04+, Fedora 36+) or Windows 10/11 with WSL2
  • GPU: NVIDIA RTX 3060 or better (voice synthesis requires CUDA)
  • RAM: 16 GB minimum, 32 GB recommended
  • Storage: 10 GB for VidNo + models, plus space for recordings and output
  • Node.js: v18 or higher
  • FFmpeg: v5.0 or higher (VidNo will check for this on first run)

Check the full system requirements page for GPU benchmarks and storage calculations.

Step 1: Install VidNo

VidNo installs via npm as a global CLI tool:

Stop editing. Start shipping.

VidNo turns your coding sessions into YouTube videos — scripted, edited, thumbnailed, and uploaded. Shorts included. One command.

Try VidNo Free
npm install -g vidno

Verify the installation:

vidno --version
# vidno v2.4.1

On first run, VidNo will download the required AI models (~3 GB). This is a one-time download.

Step 2: Configure Your Claude API Key

VidNo uses Claude for script generation. You need an Anthropic API key:

vidno config set claude-api-key sk-ant-xxxxxxxxxxxxx

Your API key is stored locally in ~/.vidno/config.json. It is never sent anywhere except Anthropic's API endpoint during script generation.

Step 3: Record a Coding Session

Use any screen recording tool you prefer. OBS Studio, SimpleScreenRecorder, or even the built-in screen recorder on your OS all work. VidNo accepts MP4, MKV, and WebM files.

For your first video, keep it short -- 10 to 20 minutes of recording. Something concrete works best: building a small feature, fixing a bug, setting up a project. The screen recording tips guide covers optimal settings.

Save your recording somewhere accessible. For this example, we will use ~/recordings/auth-feature.mp4.

Step 4: Generate Your First Video

Navigate to the root of the git repository you were working in (this helps VidNo detect commits), then run:

cd ~/projects/my-app
vidno process ~/recordings/auth-feature.mp4

VidNo will display a progress indicator as it moves through each pipeline stage:

[1/6] Analyzing frames...       ████████████████ 100%
[2/6] Extracting git diffs...    ████████████████ 100%
[3/6] Segmenting scenes...       ████████████████ 100%
[4/6] Generating script...       ████████████████ 100%
[5/6] Synthesizing voice...      ████████████████ 100%
[6/6] Rendering video...         ████████████████ 100%

Output:
  Full tutorial:  ~/recordings/auth-feature_full.mp4
  Quick recap:    ~/recordings/auth-feature_recap.mp4
  Highlight reel: ~/recordings/auth-feature_highlight.mp4

Step 5: Review the Output

Play back the full tutorial first. Check for:

  • Script accuracy: Does the narration correctly describe what is happening in the code?
  • Voice quality: If you have not set up voice cloning yet, VidNo uses a default developer voice. It is good, but your own voice is better -- see the voice cloning guide.
  • Pacing: Are cuts too aggressive or too loose? You can adjust sensitivity with vidno config set cut-sensitivity (0.1 to 1.0).
  • Scene transitions: Do chapter breaks align with logical stopping points?

Step 6: Iterate (Optional)

If the script needs tweaking, you can edit it and re-render without re-analyzing the recording:

# Edit the generated script
vidno script edit ~/recordings/auth-feature.vidno

# Re-render with the edited script
vidno render ~/recordings/auth-feature.vidno

This re-render step only takes 1-2 minutes since it skips the analysis and script generation stages.

What to Record Next

Now that you have the workflow down, here are formats that work well with VidNo:

  • Bug fix walkthroughs: Debug a real issue start to finish
  • Feature builds: Implement a feature from blank file to working code
  • Code reviews: Walk through a PR and explain your decisions
  • Tool setups: Install and configure a development tool from scratch
  • Refactoring sessions: Clean up messy code with explanations

The key insight: you do not need to perform for the camera. Just code naturally. VidNo handles the performance layer.

Next Steps

Set up voice cloning to use your own voice. Check out the batch processing guide if you want to queue multiple recordings. And if you want to automate YouTube uploads, the upload automation guide covers that workflow.