Video summary

How to Make Viral Dog Rescue POV Videos with AI

Main summary

Key takeaways

Technology

Overview

The video explains how to create “viral” first-person POV dog rescue short videos that look extremely real but are fully AI-generated from prompts. It claims the workflow is fast enough to start a YouTube channel immediately and references earnings and prior results (e.g., a niche short reaching 16M views).

Core concept: AI-generated POV rescue videos (not real footage)

  • The target format is first-person POV featuring intense dog rescue/danger scenarios (e.g., rescues, dangerous events, disasters, intrusions).
  • The creator emphasizes that you can produce similar videos using one tool/workflow and prompts, instead of filming.

Step-by-step workflow + tools

1) Analyze a competitor video with Google AI Studio

  1. Go to Google AI Studio (free AI assistant).
  2. Open the chat and paste a link to a competitor’s video.

  3. Ask the AI to analyze the video:

    • what it’s about
    • a breakdown of style/structure

  4. Request a similar video prompt, such as:

    • “Make a similar video in the same style with the same graphics”
    • Must include a first-person person + dog
    • Include a strong storyline

2) Generate a multi-part storyboard prompt for Sora 2

The prompt is broken into three parts:

  1. Beginning (setup of the rescue/danger)
  2. Middle / most dynamic moment
  3. Ending (resolution)

The guide suggests:

  • Keep the same dog breed (color may change).
  • Use a strong scenario (examples: hurricane approaching, thief entering, etc.).

Result: AI generates three high-quality prompts, one per scene.

3) Create video scenes using Leonardo AI (Text-to-Video)

The creator recommends using Leonardo AI instead of generating directly in Sora 2, claiming it’s faster and gives better style quality based on testing.

In Leonardo AI:

  • Go to Text-to-Video
  • Select generation model: Sora 2
  • Choose duration: 8 or 12 seconds (avoid 4 seconds as “too short”)
  • Resolution: 9:16 (vertical format for Shorts)
  • Enable private mode
  • Paste the first prompt and generate

Continuity across scenes (important)

For scene 2 and scene 3, use the “plus” / reference feature:

  • Add the previous generated video as a reference
  • This helps preserve:
    • location
    • characters

Notes from the creator:

  • Camera angles may shift slightly—regenerate if needed.
  • They plan to trim glitches/angle changes later in editing.

Handling prompt restrictions

If the neural network blocks the prompt due to restrictions:

  • Go back to Google AI Studio
  • Ask it to rewrite the third prompt properly, keeping everything else the same.

Editing and finishing in CapCut

  1. Download the generated scenes and import them into a video editor.
  2. Use CapCut:
    • Upload all videos
    • Drag them onto the timeline in order
    • Trim unnecessary parts (glitches, camera-angle jumps)

Scene count guidance:

  • For longer videos, generate 5–6 scenes instead of only three.

Subtitles

  • Add subtitles using CapCut’s automatic subtitle generation
  • Choose a text style similar to the competitor’s video to match the look

Export

  • Export using “best settings”
  • Publish

Example outcome shown in the tutorial

The video shows a generated rescue-style clip where:

  • the dog assists
  • the person is pulled free
  • dialogue like “you’re safe now” is implied

The creator asks viewers to comment on whether it worked.

Key product features / emphasis points

  • Google AI Studio
    • Video analysis via competitor link → produces breakdown + scene prompts
  • Sora 2 (via prompts)
    • 3-scene storyline structure for slightly longer Shorts
  • Leonardo AI
    • Text-to-video generation using Sora 2
    • Reference video continuity to maintain characters/location
    • Regenerate if camera angles drift
  • CapCut
    • Timeline trimming to fix continuity issues
    • Automatic subtitle generation
    • Subtitle style matching to the competitor

Main speakers / sources

  • Main speaker: the video creator/host (tutorial author providing step-by-step voiceover)
  • Tools referenced: Google AI Studio, Leonardo AI, Sora 2, CapCut

Original video