2026 is here — end or evolution for MX Simulator?

[DBRider251]

I'm genuinely curious how the camera is changing like that with OpenGL. I can understand the others, but the camera seems like it would have to be touching game memory unless I have a horrible understanding of how the game's camera system would work without ever seeing the code.

The camera should be pretty easy actually. It'd be a pretty simple vertex shader change. Only problem will be culling and first person models not looking right.

Not sure about how the sound works. Hopefully it just intercepts the directsound calls.

If we can make sure this isn't touching the exe in any harmful way, I'm okay approving this for use in RF. Right now, I don't want it to be injecting anything into the exe directly because that will be dangerous if left unchecked.

[WellsMX524]

Was extremely interested in this, bummer to hear that it patches the .exe

I've never liked MX Bikes. The game feels super wonky, but I've always liked the way the game felt in first person. Maybe Feitosa can get this to where it doesn't interfere with the .exe

[BcMxKx37]

Was extremely interested in this, bummer to hear that it patches the .exe

I've never liked MX Bikes. The game feels super wonky, but I've always liked the way the game felt in first person. Maybe Feitosa can get this to where it doesn't interfere with the .exe

if you go into the rF discord, someone made a similar camera mod that doesn't affect the exe in the approved mods channel.

[Shadow]

I wouldn't say I'm anti AI but I don't think I'd want to just dump AI slop into the codebase. It'd require careful review and for me at least proofreading code is harder than writing it in the first place. Definitely going to start using AI to verify my own code though! Probably going to try running Qwen locally since I don't want to send my code to Anthropic.

It's definitely super useful though! It's so nice to be able to search for a concept without knowing the right keywords.

Go for at least qwen-flash-next running through Strata and Opencode as the harness. It was the first local model I found actually usable for modding an old video game C/C++ source code. It's no panacea though, HW will limit context size so it can't work on large portions of the codebase and is liable to get stuck in repetition or reasoning loops, but it follows coding conventions laid down in agents.md fairly well. Just tell it to never write comments, or very very minimal ones because it wants to comment everything and it's your general AI slop text that may or may not help you grok whatever the function does.

That said, it's pretty good in actually understanding convoluted logic and explaining it to you. And it did help me implement bilinear interpolation to improve on the game's lighting and subsequently debug all the edge cases when adding more features