hvi

Sounds like czech law enforcement takes metin2 pervers seriously. Do we have any other example in other countries? Strictly for copyright infringement

That sounds like hitting limits to me You'd probably get better results with something like [Hidden Content]

I don't let them refactor as I simply do not trust an AI with properly structuring the code. Not once I got a proper structure, not even with claude. I use deepseek flash for bugfixes only, it's a relatively small model, I wouldn't trust it with bigger modifications. Didn't try the pro version which is substantially bigger. Kimi and GLM are the real deal, both offer coding plans like Claude but with much higher usage limits at the same price points. Kimi is usually cheaper as GLM likes to spew out insane amount of tokens for reasoning. [Hidden Content] [Hidden Content] EDIT: For big refactoring the context window is probably your biggest bottleneck. GLM 5.2 offers 1M tokens context, so that's probably your best bet for your use case. Minimax M3 should be able to handle it aswell, but I didn't have to chance to try it yet.

[Hidden Content] This benchmark is aimed at measuring coding through agentic workflows (ie: how AIs are able to leverage the tools that different harnesses provide them) Personally I'm using Kimi 2.7 and Deepseek V4 Flash, depending on the scope. I tried Kimi 2.6 and GLM 5.1 aswell and was getting good results. GLM 5.2 just come out and seems promising, didn't have a chance to try it yet. They're usually on par with Claude, in between Sonnet and Opus.

The texture gets uploaded once on the GPU and reused. Since it's not 2004 anymore and we're not as memory constrained it's perfectly fine. CPU particles updates are much more likely to be your bottleneck here

I've been working on a Godot client reimplementation for about a year now, so I can give you some pointers. I couldn't find much of help besides these 2 tools: [Hidden Content] For which you're gonna need a custom shader [Hidden Content] But it's gonna need some fixes and changes, especially for root motion. However, from the video, it looks like you already managed to get the models converted? There was also but personally I found the code to be unusable, so I ended up reimplementing everything myself. Overall it's a lot of work as Metin2 uses a lot of custom formats.

Knowledge and understanding. The exact thing your LLM is lacking, even in the response you posted to my message. A LLM does not have the capability of understanding the words or the code it uses: it's purely statistics. Sure. But it's not understanding anything. And that's why your project is stuck at the enter game. I did start a server rewrite in Go in the past, just as a proof of concept, and I managed to get in game in about a week of work. Manually, with no AI. I'll release it if anyone is interested. Hence my recommendation to do it manually, you're just shooting yourself in the foot by using AI. https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/ Security comes from understanding architecture and surface of attacks. Something LLMs cannot do AI security agents are just scanning mostly for buffer overflows: and sure, it's something that gets there by distraction but that a developer would find easily if they would just take the time to look for them. It's just that nobody sits there and starts looking for it, not black magic nor exceptional capabilities. The articles you can find online are just marketing and you can easily find examples of vibe coded software that gets immediately hacked. Not being 2021 has nothing to do with it as, once again, LLMs are incapable of understanding what they're doing. Because multithreading requires being able to understand the program flow. Yes, and frankly it's kinda offensive. It looks like you're not just limiting yourself to using AI as a translation tool, given your message structure but that you're letting it craft your answer and points. So this will be my last response here, since if I wanted to talk to chatgpt I would just go talk to chatgpt. Best of luck I wholeheartedly agree with you. The players see the client, not the server. They don't care about the backend as long as it does the job.

Would've been nice if you took at least the time to write the posts yourself, instead of having an AI do it. I think people would appreciate talking to people instead of LLMs. With that said you're rebuilding critical infrastructure, expecting a LLM to be able to do it is straight up delusional. Rebuilding such project requires architecture knowledge and understanding that LLMs will never be able to achieve. And even if it were to manage to get something working you're only setting yourself up for security issues. Still it is a nice project that can teach a lot and I'd really encourage you to write it yourself as you'd get a lot of experience out of it. C# is fine for it as the server ticks every 40ms if I'm not wrong, so the garbage collector is not an issue. The real benefit of a rewrite would be making it multi threaded, to leverage all the cpu cores. But once again a LLM will not be able to do it.