How I Create Viral Historical POV Videos with AI (Step by Step)

You’ve seen these: someone wakes up as a Roman soldier, a medieval knight or a Viking about to go into battle. One Instagram account with fewer than 20 of them hit 1.2M and 5.5M views and grew to almost 40,000 followers in under a month. Each video takes less than 20 minutes.

Step 1: Plan the story with the free GPT

Open my Historical POV Shorts GPT (link above), click Start, and answer three questions:

1. Era: e.g. ancient Rome
2. Region: Rome
3. Your role: noble

In seconds you get a full production plan: six scenes telling one story from waking up to the end of the day, each with a first-person photorealistic image prompt and a matching video prompt. You don’t write a single prompt.

Step 2: Images

On Higgsfield (50% off) → Image → Nano Banana Lite (cheapest and fastest, so you can experiment):

• Aspect 9:16 (Reels/Shorts), quality high, 4 images per prompt so you can choose
• Paste scene 1’s prompt and generate
• Repeat for each scene with the same settings, picking the version with the fewest artifacts and most natural light

Step 3: Bring them to life

Video → Gemini Omni Flash. Paste the scene’s video prompt, upload the scene image as the reference, set 7 seconds, 9:16, and generate. The sound and camera motion feel cinematic.

Fix tip: with Gemini Omni Flash you don’t have to start over. Click Edit video, draw over the problem (I circled a chair) and type “remove the chair.” Gone.

Repeat for all six scenes, swapping the reference image each time.

Step 4: Edit

In CapCut:

1. Drop the six clips in order.
2. Add smooth background music and lower it. You can mute the clips so only the music plays.
3. Add a POV title (e.g. “POV: you’re a Roman noble”) using a text style with a background box so it pops.
4. Export.

📝 Your notes on this video

Saved on this device only

Copy notesDownload .txt