What's the opinion on locally generated code completions like in modern JetBrains IDEs? They work offline even on a laptop, but use a LLM, which is generative AI. What if I run a mid-size language model on my gaming GPU to help me code?
Yes, I noticed the resize issue too but did not want to go over the allotted dev time.
To be honest, almost all the code is from the example project or generated. I checked each line of code but was happy with the implementation and only tweaked some of the values manually to make the physics feel right. I'm considering typing more of the code manually next time, the hosted LLM feels like an unfair advantage.
Regarding the high score: I managed only 351, but my girlfriend got to 974 m!
That's a really cute concept! Does it ever end? The pieces kept getting smaller and smaller and then I restarted by accident. Anyway, I had lots of fun.
Nice entry! Glad you added the checkpoint, I was worried that this would be hard to finish, even the second jump was kind of a challenge for me, but i made it in the end.
The look of the glue pickups is nice, but maybe having the glue meter partially regenerate would let players avoid having to jump off the edge if they run out.
Cool game! It really captures that retro platformer vibe and I really like the level of challenge you made. I got to the final stage twice, but the timer got me before I could actually reach the finish.
