Comment

Comment by Drache93

Main reason at the start was blatant curiosity of what the pros and cons would be and if AI could manage it!

Since the games systems were already well defined - aka generating a race, tight budget per tick, automated tests - and it can all be replayed and compared. So every tick things happen, we can compare the hash with output of Lobster, prove the game has been ported cleanly. With lobster we can play hundreds of games in a few seconds to test, check balance etc. So the performance boost was worth it already.

When I realised that was possible I decided to give it a go and see what other benefits! Having it run on android, faster tests, faster generation. And now moving to voxels for units so I can generate their art more reliably. Realising no engine gives a lot of freedom, and planning to release it as an RTS engine to build on now

No optimisation plugins or anything. Just tight control of claude. Every time something slow; tests, load time, unit generation (creating the art) - I solved it architecturally doing a deep dive and rework to solve the issued - new tooling etc

parent Drache93 7 days ago

No comments yet

Be the first to share your thoughts.

Live from the official Hacker News API

A modern reader for Hacker News · SvelteKit + Skeleton