I Turned a Tokyo Drift Scene Into a Game With AI
Long-form14:19GamingPublished Oct 8, 2026, 6:51 PM · 2w ago
Views
3,714
+1 views/hour
Views / hour now
1
measured 9h ago
Peak views / hour
1
1d ago
First 24 hours
—
not tracked from upload
Engagement
2.9%
likes + comments / views
Likes
84
2.3% of views
Comments
22
0.59% of views
Est. ad revenue
$26–$59
lifetime, $7–16 RPM
Checked every 12 h while the video is this age · last check 9h ago · next in 129 min. YouTube itself updates public counts with a short delay.
Avg views / day
262
since upload
vs channel median
0.4×
this video's views ÷ the channel's usual
Rank in channel
#9
of 21 tracked
Views over time
4 readings
Total views Views gained per hour
Description
I grew up on these films, so I wanted to drive that yard myself. Not a game "inspired by" the Tokyo Drift drift-lesson scene, but the same yard, metre for metre. I gave the job to Claude Opus 5.5 and a small team of AI agents: one builder, a critic that tries to fail everything, and a tester.
The first thing they found out: it was never Tokyo. The yard is in an old shipyard on Terminal Island, Port of Los Angeles – confirmed with a port survey photo from 2000 that nobody had linked to the film.
In this video:
• The location detective: one website, a 2000 survey photo, a 2006 aerial – and the sun giving away the time of day
• From public maps and a LiDAR scan to a 1:1 yard in Godot (74 of 74 buildings within 25 cm)
• Our game laid over the film, frame by frame
• A car built from blueprints… that I stopped, and the Tripo model that arrived one metre long
• Drift physics, a camera jitter hunt, and an engine sound rebuilt from single firing pulses (PSOLA)
• Ten rounds of me playing, complaining, and the AI fixing
• Why it took ~42 hours, and what it cost: ~$566 at API list prices
Chapters:
0:00 Intro
1:21 The setup: one orchestrator, three roles
2:04 Where was it really filmed?
3:35 From map data to a 3D yard
5:17 The car
7:32 Into the game engine
8:05 Drift physics
9:17 Ten rounds with me
12:00 The result
12:39 Why it took so long, and the cost
13:29 What I learned
Let me know in the comments: do you like this format, and would you play this game?
Tools: Claude Opus 5.5 (orchestrator, builder, critic, QA) · Blender (Python scripts) · Godot 4.7 · Tripo (car body) · GPT-6 Astra (image references, HUD design)
Fan project – not affiliated with or endorsed by Universal Pictures or Mitsubishi Motors. The game is not released or distributed. Film frames © Universal Pictures, shown briefly for comparison.
Sources & credits: Port of Los Angeles, Architectural Survey (Jones & Stokes, 2000) · Imagery: LARIAC, © Los Angeles County · NAIP / USDA · USGS 3DEP LiDAR · © OpenStreetMap contributors · LA County Building Outlines 2020 · Natural Earth · Sound effects: Pixabay
Music: "Dismantled Toys" by The Grey Room / Density & Time · "Maple" by Dyalla (YouTube Audio Library)
If you want to follow what we build next, subscribe.
Tags
tokyo drifttokio driftfast and furiousfast and furious tokyo drifttokyo drift scenedrift lessonlancer evolutiondrift gameai game developmentgame devindie game devgodotgodot 4blenderclaudeclaude opus 5.5claude codeanthropicai agentsgpt-6 astratripo aiai 3d modelai codingdevlogmovie scene recreationfilming locationdrift physics
Details
- Published
- Oct 8, 2026, 6:51 PM
- Length
- 14:19 (859 s)
- Format
- Long-form
- Category
- Gaming
- Audio language
- EN
- Quality
- HD
- Captions
- Yes
- Made for kids
- No
- Licensed content
- No
- Video ID
- pdD52-sM6V8
More from AI Dev Challenge
AI Agents Built My Space Shooter in 2 Days with Claude, GPT, Customuse, Blender and Godot274 views · 7h ago · +30/h
Sonnet 5.5 is incredible in gamedev740 views · 1w ago · +2/h
I Gave Claude Sonnet 5.5 Four Concept Images. It Built the Game.10.1K views · 1w ago · +3/h
Claude Opus 5.5 Rebuilt Asteroids in Godot1.9K views · 1w ago · +2/h
AI built the Drift Game798 views · 1w ago
AI Built This Tokyo Drift Game 🤯 #gamedev #devlog #claude #opus55777 views · 2w ago
We're Building a Survival Game With AI… Can You Survive It?639 views · 2w ago
Claude Opus 5.5 Ran a Team of 9 AI Agents. Here's What They Built.354 views · 2w ago
Statistics from the YouTube Data API, tracked by FameX. Views per hour, revenue and comparisons are FameX estimates.