Video summary

Every Prompt to Make Unlimited Free AI Videos

Main summary

Key takeaways

Technology

Overview

This tutorial explains how to generate unlimited free AI videos by combining AI agents (Claude desktop/code) with an open-source tool called HyperFrames.

Setup / Installation Workflow

  1. Open Claude desktop and create a new session.
  2. Create a new local folder/directory for the project, then set Claude code to work in that folder.
  3. Configure options:
    • Use auto mode / accept edits so the agent applies changes unless it needs permission.
    • Choose a model:
      • Opus 4.8 for Mac
      • Sonnet 5 recommended for “Pro” users for token efficiency
  4. Run the install command: install HyperFrames.

Prompts Provided (with example outcomes)

1) Beginner Prompt (8-second explainer)

  • Goal: Create a fast-rendering 8-second landscape animated explainer.
  • Topic example: health benefits of sunlight
  • Requirements: one strong animated visual metaphor for instant understanding.
  • Output demonstrated using: concepts like vitamin D, mood, energy, sleep.

2) Intermediate Prompt (brand/context-based; 30 seconds)

  • Goal: Make a 30-second faceless explainer.
  • Example topic: high-protein diet benefits
  • Includes benefits such as:
    • satiety (eat less)
    • blunting blood sugar spikes
    • muscle rebuilding
  • Also includes: a suggested daily intake section.
  • Iteration guidance:
    • After preview, if something is off, ask the AI agent to modify and iterate until satisfied.

3) Intermediate Prompt (external context → launch/product video; 25 seconds)

  • Goal: Use external info (e.g., a company website) as context.
  • Uses a pre-built workflow/skill called “website to video”.
  • Prompt requirements:
    • Capture brand colors, typography, UI/brand feel
    • Create an Apple Keynote-style cinematic product reveal
    • Include dramatic pacing, animated zooms, three benefit cards, voiceover, subtle sound effects, and a final CTA
  • Demonstrates the workflow can also assemble your own assets (product images/video clips).

Optional Prompt Logic: Talking-head Video → Edited Explainer

If you provide a raw talking-head video, the workflow:

  1. Transcribes what you say with Claude.
  2. Cleans the transcript by removing:
    • false starts
    • stuttering/repeats
    • filler words
    • pauses longer than ~0.3 seconds
  3. Breaks the script into key concepts.
  4. For each concept, generates a strong visual metaphor.
  5. Applies a consistent visual system, including:
    • theme (example: light + blue primary color)
    • fonts/type settings
    • motion parameters
    • recurring graphic rules (e.g., cards for metaphors/quotes/comparisons/steps/titles)
    • transparent overlays near/over the face
  6. Syncs graphics to timestamps in the spoken script.
  7. Exports MP4 matching the aspect ratio of the source footage:
    • vertical shorts for short recordings
    • landscape/appropriate format for long-form videos

Main Speakers / Sources

  • Main speaker: the tutorial author/instructor (referred to as “I’m going to…”; no specific name provided in the subtitles)
  • Tools / mentioned sources:
    • Claude (Claude desktop / Claude Code)
    • HyperFrames
    • “faceless explainer” workflow
    • “website to video” workflow
    • Apify / Blutato (for scraping/transcript context)

Original video