Jump to content

Recommended Posts

Posted

I never liked fake players or even fake counters to any server..

So i dont think developing this one its a good idea about money it is but for gameplay is not..

 

Posted
18 minutes ago, SantaPaul said:

The only way this could work if you got over 300-400 real online players so you put like 100-200 fake players. This can be "hidden"

The best case scenario is both what you said and also grand opening.

Posted
15 minutes ago, Nightw0lf said:

if you sell this kind of shit i will share acis/h5 version for fake players and destroy l2 for good

i forget we are at 2k17 and l2 is w/o fake players. wtf, every server already using 50% of fake players 

 

 

Posted
1 hour ago, Nightw0lf said:

if you sell this kind of shit i will share acis/h5 version for fake players and destroy l2 for good

Feel free to do so. You will save me so much time.

1 hour ago, AchYlek said:

i forget we are at 2k17 and l2 is w/o fake players. wtf, every server already using 50% of fake players 

 

Yeah i did a bit digging around and it seems that everyone is using shared phantom.

Posted
2 hours ago, Nightw0lf said:

if you sell this kind of shit i will share acis/h5 version for fake players and destroy l2 for good

You're all amateur kids behind a 50 euro monitor trashtalking as if you control the world in your hands. 

"If you do this, i'll share this and then boom whole l2j community is dead." .. Yes maybe you destroy the whole internet for good? 

As if all humanity wait for your fail coding engines to quit L2J. I wouldn't respond but when i see so high self-est people writting down posts, it provoke me.

Posted

After a bit of testing and benchmarking, it seems that my current implementation won't fly. After 600 players the server load is 99%. I pretty much knew that but i wanted to see what the load would be with individual thinking tasks.
I will go with the lighter approach of batching per AI type.

Other than that everything else seems to be going well.

Posted
2 minutes ago, Nightw0lf said:

i've been there you cant use individual tasks but one task for each move skill using moving attacking talking and you can load 4k players with 1% server load

I've done it before, i wanted to see what the difference of "before" and "after" would be.
I'd rather have one task per AI for now and see how that goes.
If it seems heavy then i will go even higher.

  • 3 weeks later...
Posted

So i finally have 3 days off to work on this. I started by remaking the way the bots think about how they will do stuff.

Previously i had one task running per player just to see how far this could go and it because obvious that after 200 players the thing went kraaaa.

 

However i reworked the way this logic in handled and now with a reasonably low cpu usage.

 

To test it i spawned 666 bots of two types, AttackerAI (general purpose) and SpellHowlerAI (class specific) and the load is kept low, so im happy withe the performance side of things.

 

jVzWcdS.jpg

 

Next thing in the todo list is to create individual class based AI for combat and social interaction.

Bots should behave differently in different scenarios and even though fake, their decisions should at least look authentic.

Posted

Some wizard combat stuff. The players are unbuffed and just use spiritshot thats why some of them stop casting for a sec some times when all skills are on cooldown.

 

 

Posted

Been working on the decision making and skill picking. Some skills are more likely to be used by others.

 

Also bots have automatic internal rebuffing and refueling of consumables (shots/arrows/bones) when they are about to run out.

Performance is solid.

 

 

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



  • Posts

    • Hi, I'm selling Interface sources. I was working with one of bigger Essence projects (for ±1 year) delivering them Interface source KR. Per this period gathered knowledge about UE, UE Scripting, saw a lot of tools and their progression. So finally I decided that I can open up my own sales topic. I'm selling source for any protocol - If I don't have source, I will need you to provide whole system folder or for Essence/main - provide me protocol number. Sources are acquired by my OWN decompiler - written in Java, so we're not using popular UEExplorer upgraded by anyone. Decompiler extracts already compile ready sources. It outputs clean code, however constants are mapped 60-70%. Why only 60-70%? Well tl;dr when you compile any sources, they take at that time constant value and place it instead of = CONST_NAME to = THAT_CONST_NAME_VALUE, so it's not possible to have const name instead of value. Example of output: https://pastebin.com/cjyXcJvT for bigger .uc class (557p). p.s pastebin messed up encoding, but KR symbols are preserved - you won't see "?" symbols, rather you will see proper original korean text. I will give you guide as well, how not to mess it up with your editor VS Code/Notepad++ or what you use it to make them preserve it. You will get source + my compiler wrapper (written in java) - works on Windows/Mac/Linux - it has small GUI but supports CLI. Small notice, who thinks 'this is decompiled source' - basically all sources that are on sale are made by decompiling .u and applying changes. I think powerhouse few years ago was ArtToKill/kick (kick added support for INT64 to the publicly available compiler sources) etc. So no - there are no straight from 'creator' sources anywhere, all sellers offer the same - just the quality is different. UEExplorer upgraded was a big step, however, it introduces still a lot of nesting, if statements wrapping incorrectly which could lead to bugs. There are many things how to identify poor quality sources, as if the seller tries to sell 'out-of-box' when they see it compiles - sad news, some bugs are in the logic due to seller focusing only on 'it compiles' rather than issues that their decompiler is not capable so solve. On request I can restore ±100% const mapping.   Later I can attach compiled .u to try. Price is 400EUR. Payment methods USDT (Crypto). I'm still in debate with myself, wether to start selling XDat Editor with structures and ClientDat with 100% all files openings (no unk namings) and improved experience (such as comments support in .dat files, mass sync, improved performance and so on). P.s. If you ask me to give you some project's custom interface - the price may vary - there are couple things per project that need to be hand checked and adjusted (for example Innadril completely changes tokens which then I need to adjust my program to their token Ids). So it might take more time = more money. P.s.s. I have more things in the garage, like my own dsetup source that can allow you 'unlocking' system with your desired encrypting - you don't need to wait till 'patched' systems start appearing and so on, but I think at this time everyone who needs it make them selves it. AAC crypt removals and so on, BuildZone, everything that 'unlocks' client, however, I don't plan to make them on sale as mentioned, everyone can do them.  
    • It has been a while since I updated this thread. I am currently actively working on a full-scale port of the shader library to ensure the fast and accurate transfer of all effects. I have decompiled the original game and created a tool that fully logs the entire *Lineage 2 Interlude* particle system. This allows me to extract the original formulas hardcoded into the game, resulting in a high-fidelity port. For instance, *Lineage 2* uses an `appRnd` system that generates unique randomization patterns for different skills, giving them distinct visual characteristics; there are countless formulas and technical nuances like this. Ultimately, I will derive all the key formulas and replicate the geometry with high precision. Twelve basic geometry formulas have already been ported. The effect was created in 10 minutes using a .uc file, HLSL formulas, and SpriteEmitter particle blending formulas.   https://prnt.sc/_3UDLQVX_Y-_
  • Topics

×
×
  • Create New...

Important Information

This community uses essential cookies to function properly. Non-essential cookies and third-party services are used only with your consent. Read our Privacy Policy and We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue..