Another update to my ghibli-themed train sim game! added midnight mode and a dedicated garage town with sick updates. built with GPT-6 Astra, Crayon and Threejs. Play here:

@usecrayon Very enjoyable.

@usecrayon Thank you!

@usecrayon I’m going to play

@usecrayon Thanks! please do :D

such fun!

Thank you big man

@usecrayon This model really can help us achieve our dream

@usecrayon Sure can :D

@usecrayon Can you tell me before creating any game what principals should I follow to write better prompts to my actual idea

@usecrayon Can you drop a DM :)

@usecrayon Actually I can't dm to premium users

@usecrayon DM'd

@usecrayon This visual ❤️

@usecrayon Thanks!

@usecrayon Wonderful, so healing.

@usecrayon Thank you ✨

@usecrayon This is sick 🔥, amazing work man

@usecrayon Thank you!

@usecrayon wow!!

@usecrayon Appreciate it :)

@usecrayon 美しい😍

@usecrayon Thank you!

@usecrayon Nice ! Might have a cool concept !

@usecrayon Appreciate it! Hmu 🤝🤝

@usecrayon Fantastic work! 🔥

@usecrayon Thanks!

@usecrayon Nice visuals! Trains are hot categories on steam!

@usecrayon Possibly haha, Thanks!

@usecrayon I'm building a Barton Springs diving game with AI, so I'm curious how you're keeping the look consistent between the original world and garage town. Reference screenshots, a written style guide, or mostly manual tweaks?

@usecrayon Sounds great, I do a mix of all. Constantly updating the context helps. But after first prototypes most of the work is done after testing things personally and iterating

@usecrayon you are a magician!

@usecrayon Thank you!

@usecrayon Woa nice track! Will give it a try!

@usecrayon Calling a 3js sim a game is a stretch

@usecrayon my existing players would disagree haha but all good

Update :D

@usecrayon 音樂響起 回億來了

@usecrayon midnight mode + garage town on the same train world is the right kind of iteration was the lighting pass mostly prompt or did you hand tune the three.js night cycle

@usecrayon This is so fucking cool 😂 I want the source so I can try to add a few more rail cars and my realistic train physics from Only with much lighter weight of course.

@usecrayon the next metric I want is human interventions per accepted task the failure taxonomy will be more useful than another best-case demo