Metin2 on another game engine? UE5/Unity

[mrspeak 3]

Hello everyone,

 

I’m not here to waste your time talking about trying to recreate Metin2 from scratch on a different game engine, like many others have done before giving up. However, I’m curious to know why those projects were abandoned, and for those who considered it but never started, what held you back?

 

Some might say, “Making a game from scratch is not easy,” and I wouldn’t disagree nothing worthwhile is easy. But I’m interested in understanding why some people gave up after starting, and why others only thought about it but never took the first step.

 

I have two questions:

 

1. For those who started but then gave up, what was the reason? At what point did you decide it wasn’t achievable?

2. For those who never started and only thought about it, why didn’t you? What stopped you?

 

I’m also considering recreating Metin2 in Unreal Engine 5, but if I do, it will be open-source. That way, if I decide to abandon the project, someone else might be able to continue.

[Kaiser 1246]

There are very few people who do this job and each one does it individually. So they are not a team. This means that the entire workload falls on one person. It is not easy to transfer all the objects, effects, maps and everything else.

And these people are probably doing this as a fantasy. Even if they continue and complete it, I don't think they will be able to open and run a server with this project. Because if GameForge warns and shuts down the server (which is very likely), it means all their efforts will be in vain.

[lime 261]

1 hour ago, blaxis said:

There are very few people who do this job and each one does it individually. So they are not a team. This means that the entire workload falls on one person. It is not easy to transfer all the objects, effects, maps and everything else.

And these people are probably doing this as a fantasy. Even if they continue and complete it, I don't think they will be able to open and run a server with this project. Because if GameForge warns and shuts down the server (which is very likely), it means all their efforts will be in vain.

"GameForge warns and shuts down the server (which is very likely)"
shuts down
very likely

Didn't happen, won't happen. 

[Kaiser 1246]

2 minutes ago, lime said:

"GameForge warns and shuts down the server (which is very likely)"
shuts down
very likely

Didn't happen, won't happen. 

Yes, it is possible. Because this situation once happened in Turkey.

However, the game had original graphics. It was sued and closed down because it only had too many players.

[mrspeak 3]

3 hours ago, blaxis said:

There are very few people who do this job and each one does it individually. So they are not a team. This means that the entire workload falls on one person. It is not easy to transfer all the objects, effects, maps and everything else.

And these people are probably doing this as a fantasy. Even if they continue and complete it, I don't think they will be able to open and run a server with this project. Because if GameForge warns and shuts down the server (which is very likely), it means all their efforts will be in vain.

I never mentioned anything about transferring assets. While you can more or less just drag and drop models, animations, and maybe effects into Unity, unfortunately, I can’t say the same for UE5. It works, but it involves a few more steps.

 

Nowadays, there are many AI tools that allow game developers to create models with animations and effects using just a prompt and a few clicks. I won’t even get started on how easy it is to make maps and how many free, copyright-free assets are available to choose from.

 

In my opinion, the most “challenging” aspect is the time investment—both for those who already know what they’re doing and for those who still need to learn. In Unity, you need to know C#, whereas in UE5, you don’t necessarily need programming skills, just a basic understanding of concepts like variables, structures and stuff like that.

 

[Criminus 673]

Metin2 is a perfect game in my eyes. I would never play a high quality graphics metin2 server. It just sux so bad. The graphics (terrains, skybox, trees. shades) are too realistic and the characters aren't. Same for all models; mobs, mounts, etc.

But to answer your question, it's too hard for one man to do such project. I've been playing on localhost to replicate an old browser game and at some point it gets boring and not worth the hassle. Copyright would be the second part. All your work will be wasted. Thinking about using A.I, you can scrap the idea. 

[ValkeryE 124]

As the friends above wrote. Usually it's one person who, as a fan of Metin2, wants to try their hand at porting Metin to Unreal Engine 5. They think too generally. What could be the problem? Metin is a simple game after all. You'll get the hang of it quickly. But really, when they add the first 3D building model, warrior + animations and the first "scripting" starts, they suddenly open their eyes to how much code there is to write and they simply give up.

[rcngrgl 3]

I worked at a MMOFPS gaming company for couple of years. I know how hard is to develop and maintain the game especially if it is online. But despite that I still decided to develop a replica of Metin2 a week or so ago. I will keep you updated once I give up

[lime 261]

On 10/25/2024 at 7:18 PM, blaxis said:

Yes, it is possible. Because this situation once happened in Turkey.

However, the game had original graphics. It was sued and closed down because it only had too many players.

So, what about Elveron, WoM, Rubinum, Rodnia, Oryum coming soon... what about them? 

[rcngrgl 3]

I guess I start to understand why. Whenever I am doing something I am looking if someone did it or how they did it. They are always doing it manually.

Most people who tried to recreate Metin2 didn't go to the path for writing parsers. Simplest example for this is to terrain. Game files have already data for height map, shadows, tiles, etc. What I see is people are importing height map in game engines but using the other parts. So they are painting the terrain manually.

Another example is, game objects like buildings have separate collision file. Need to write parser for that to convert them, or manually do everything. People usually picking the manual option.

So, as I guess most people don't have capacity or experience to write parsers for all the required files. Because of that it makes development process much harder. Or they don't have experience at all on MMO game development and they just get overwhelmed.

[LoKnarash 82]

1 hour ago, rcngrgl said:

Immagino di iniziare a capire perché. Ogni volta che faccio qualcosa, guardo se qualcuno l'ha fatto o come l'ha fatto. Lo fanno sempre manualmente.

La maggior parte delle persone che hanno provato a ricreare Metin2 non hanno seguito il percorso per scrivere i parser. L'esempio più semplice è il terreno. I file di gioco hanno già dati per la mappa delle altezze, le ombre, le tessere, ecc. Quello che vedo è che le persone stanno importando la mappa delle altezze nei motori di gioco ma usando le altre parti. Quindi stanno dipingendo il terreno manualmente.

Un altro esempio è che gli oggetti di gioco come gli edifici hanno un file di collisione separato. Bisogna scrivere un parser per convertirli, o fare tutto manualmente. Le persone di solito scelgono l'opzione manuale.

Quindi, come immagino la maggior parte delle persone non ha la capacità o l'esperienza per scrivere parser per tutti i file richiesti. Per questo motivo, il processo di sviluppo diventa molto più difficile. Oppure non hanno alcuna esperienza nello sviluppo di giochi MMO e si sentono semplicemente sopraffatti.

Yes, surely what you said is true, that is, there is certainly a lack of experience in creating an mmo on Unity or Unreal. Surely doing it on Unity is faster thanks to the asset store but the fact remains that creating metin on Unity makes no sense in my opinion. If you want to make an mmo on unity or unreal you don't do metin.

I think that if metin were remade on another engine it would lose the "charm" and experience that metin offers.

 

[rcngrgl 3]

11 hours ago, LoKnarash said:

Yes, surely what you said is true, that is, there is certainly a lack of experience in creating an mmo on Unity or Unreal. Surely doing it on Unity is faster thanks to the asset store but the fact remains that creating metin on Unity makes no sense in my opinion. If you want to make an mmo on unity or unreal you don't do metin.

I think that if metin were remade on another engine it would lose the "charm" and experience that metin offers.

 

I don't think so. When I am playing my version, all the graphics, animations, etc. reminds me the old days. It is not losing the "charm", on the contrary it is bringing the "charm" back. Because it is simple like old days and I know that there will be no cheaters.

[rcngrgl 3]

I still continue on my development. If I will have significant progress I will open discord channel to share my updates as well as forum posts.

I still watch a lot of videos of attemps to remaking the Metin2. I didn't change my mind. It is always the same flaw. People are not trying to convert the existing assets, instead trying to make them by themselves. These includes map tiles, model collisions, effects etc.

• If creating a map takes some days, it will take days for each map.
• If creating a collision for an object takes 5 minutes, it will take 5 minutes for each object.

Instead

• I've spent a lot of days for a single map. Now others will take minutes.
• I've spend a lot of days for a single object with a collision. Now others will take seconds. They will be also put in a map automatically because there are data for that too!

You got the idea.

Another issue that I see is, not tackling the big parts. I am working on effects now. I've spent my days just to understand how they are working. I need have a conversion for their effect data to the engine I use. Otherwise game will not be playable without the effects. If I focus on every single monster, every map, every item, every something before this. It will be a huge loss. So I am trying to make one or two things work from main features. Once I get the MVP, I can always add the remaining parts.

It is a long time after you opened this post, but I hope I was helpful.

[duwen123 72]

14 hours ago, rcngrgl said:

I still continue on my development. If I will have significant progress I will open discord channel to share my updates as well as forum posts.

I'd definitely be interested, and I'm sure others around here, to follow your progress. Including the technical, developmental details. Me personally to learn and better understand more advanced concepts.

So, if you decide to open a Discord server for your devlogs, count me in! 

[Mdev02 4]

On 12/15/2024 at 5:20 PM, rcngrgl said:

I still continue on my development. If I will have significant progress I will open discord channel to share my updates as well as forum posts.

I still watch a lot of videos of attemps to remaking the Metin2. I didn't change my mind. It is always the same flaw. People are not trying to convert the existing assets, instead trying to make them by themselves. These includes map tiles, model collisions, effects etc.

• If creating a map takes some days, it will take days for each map.
• If creating a collision for an object takes 5 minutes, it will take 5 minutes for each object.

Instead

• I've spent a lot of days for a single map. Now others will take minutes.
• I've spend a lot of days for a single object with a collision. Now others will take seconds. They will be also put in a map automatically because there are data for that too!

You got the idea.

Another issue that I see is, not tackling the big parts. I am working on effects now. I've spent my days just to understand how they are working. I need have a conversion for their effect data to the engine I use. Otherwise game will not be playable without the effects. If I focus on every single monster, every map, every item, every something before this. It will be a huge loss. So I am trying to make one or two things work from main features. Once I get the MVP, I can always add the remaining parts.

It is a long time after you opened this post, but I hope I was helpful.

How has it been going? I have been on the same route you have but for about 1,5 years now. Few things I can tell everyone that some what confirms what this poster says.

 

Main issues:

Parsers, parsers and parsers. Unless your main desire is to retest, rebalance everything it wont be possible.

Heightmaps:

About heightmaps etc. Smaller issue than is made out to be. This is just common amongst mmorpgs and nobody runs via sweeps. Its pre-baking work that makes server & client load cheaper and prevents tons of movement sweeps. Cells/grids.

Effects:

I did manage to pull off effects to the point that it resembles metin 1:1. There's no difference. But its ALOT of work. You need parsers again just to read all the msf, mse and msa files. Else you have to rewrite it all or spit through every single msf msa and mse file. Let me tell you, its quite alot. 

Collision:

Collision around buildings etc. can infact be done automatically to save alot of time. Maps is where it gets a bit more complicated, you could build a seed generator and use metin2 layers in UE landscapes with a custom layer system. You need to create your own modules to truly mirror metins layering / paint system.

 

Graphics:

Either go with UE5 graphics or build your custom rendering pipeline on top of ue so you can get rid of most blooming/realistic lighting crap. UE APIs can help but it still requires tons of manual work, testing etc.

Shadows:

You need custom shadows, not unreal dynamic shadows, not unreal static shadows. Those are heavy GPU eaters in unreal. That wouldnt represent metin2, it also looks way realer than simple shadows from 2004.

Also get rid of all HDR lighting, build your own lighting renderer that resembles metin2. 

Recreate skyboxes, clouds etc. just to mirror the same skyboxes. Metin2 has very old skyboxes so just dropping it in and betting on HDR/HDRI is never gonna work.

 

Characters/meshes:

 

Mobs &. Npcs? Easy. Controllable characters? Needs manual tweaking in Blender, keyframe transferring to UE compatible root bone. Recreating root motion because root motion seems to dissapear in 10/10 export->import paths.

Another issue: bone scaling, socket rotations, inverted axises, weapons not having proper origin points so offset is always there, which does not exist in metin2. Infact, EVERY offset/rotation is incorrect.

 

Now all this is mostly what the players see.

 

Next guardrails: custom networking, UE server overhead vs Metin2 old engine. Alot more ram per server. You need to know how dbs work, how scriptings work, how you program c++ with ues custom c++ layer. Minimizing rpc due to rpc overhead. Reliable vs unreliable, entire custom/rewritten backend.

 

The main issue I genuinely think most have is the amount of time. Not perse that its extremely hard (definitely requires a good amount of knowledge). But it is just alot of work which isnt required by just starting a private server, like anyone else does.

 

Then: IP, Copyright, potential risks, live ops, server hosting + a super backend that can handle multiple unreal ran server instances which run on alot more raw RAM and cpu cost than Metin2 does.

 

Nontheless, I did not stop. I did not give up and I will finish this project. But it is not without difficulties and quite frankly, its only possible for me because of my current situation. I would declare someone with little time crazy for even trying.

 

(fyi: 12-16 hours of work a day, for 1,5 years. Nonstop. I only take 1 day off a week so I don't burn out and im still not even close. I could say im at about 50% to actually release this)

[HAMLET 15]

16 hours ago, Mdev02 said:

How has it been going? I have been on the same route you have but for about 1,5 years now. Few things I can tell everyone that some what confirms what this poster says.

 

Main issues:

Parsers, parsers and parsers. Unless your main desire is to retest, rebalance everything it wont be possible.

Heightmaps:

About heightmaps etc. Smaller issue than is made out to be. This is just common amongst mmorpgs and nobody runs via sweeps. Its pre-baking work that makes server & client load cheaper and prevents tons of movement sweeps. Cells/grids.

Effects:

I did manage to pull off effects to the point that it resembles metin 1:1. There's no difference. But its ALOT of work. You need parsers again just to read all the msf, mse and msa files. Else you have to rewrite it all or spit through every single msf msa and mse file. Let me tell you, its quite alot. 

Collision:

Collision around buildings etc. can infact be done automatically to save alot of time. Maps is where it gets a bit more complicated, you could build a seed generator and use metin2 layers in UE landscapes with a custom layer system. You need to create your own modules to truly mirror metins layering / paint system.

 

Graphics:

Either go with UE5 graphics or build your custom rendering pipeline on top of ue so you can get rid of most blooming/realistic lighting crap. UE APIs can help but it still requires tons of manual work, testing etc.

Shadows:

You need custom shadows, not unreal dynamic shadows, not unreal static shadows. Those are heavy GPU eaters in unreal. That wouldnt represent metin2, it also looks way realer than simple shadows from 2004.

Also get rid of all HDR lighting, build your own lighting renderer that resembles metin2. 

Recreate skyboxes, clouds etc. just to mirror the same skyboxes. Metin2 has very old skyboxes so just dropping it in and betting on HDR/HDRI is never gonna work.

 

Characters/meshes:

 

Mobs &. Npcs? Easy. Controllable characters? Needs manual tweaking in Blender, keyframe transferring to UE compatible root bone. Recreating root motion because root motion seems to dissapear in 10/10 export->import paths.

Another issue: bone scaling, socket rotations, inverted axises, weapons not having proper origin points so offset is always there, which does not exist in metin2. Infact, EVERY offset/rotation is incorrect.

 

Now all this is mostly what the players see.

 

Next guardrails: custom networking, UE server overhead vs Metin2 old engine. Alot more ram per server. You need to know how dbs work, how scriptings work, how you program c++ with ues custom c++ layer. Minimizing rpc due to rpc overhead. Reliable vs unreliable, entire custom/rewritten backend.

 

The main issue I genuinely think most have is the amount of time. Not perse that its extremely hard (definitely requires a good amount of knowledge). But it is just alot of work which isnt required by just starting a private server, like anyone else does.

 

Then: IP, Copyright, potential risks, live ops, server hosting + a super backend that can handle multiple unreal ran server instances which run on alot more raw RAM and cpu cost than Metin2 does.

 

Nontheless, I did not stop. I did not give up and I will finish this project. But it is not without difficulties and quite frankly, its only possible for me because of my current situation. I would declare someone with little time crazy for even trying.

 

(fyi: 12-16 hours of work a day, for 1,5 years. Nonstop. I only take 1 day off a week so I don't burn out and im still not even close. I could say im at about 50% to actually release this)

So what is the point to move to unreal if you spend your days to  keep the old 2004 graphic and remove everything that  Unreal gives you? Cause i see none but worthless work. 

[rcngrgl 3]

I am happy to see someone understands what needs to be done and works on it, really liked your detailed progress report.

On 6/14/2026 at 8:41 PM, Mdev02 said:

Main issues:

Parsers, parsers and parsers. Unless your main desire is to retest, rebalance everything it wont be possible.

Yeah, parsers are a must, but once they are done, you can use every data, model and that's the beauty of it.

On 6/14/2026 at 8:41 PM, Mdev02 said:

Heightmaps:

About heightmaps etc. Smaller issue than is made out to be. This is just common amongst mmorpgs and nobody runs via sweeps. Its pre-baking work that makes server & client load cheaper and prevents tons of movement sweeps. Cells/grids.

Heightmaps is simple as you said. I guess my only problem was creating custom shader for blending between areas, originally it was done just by blur or something like that.

On 6/14/2026 at 8:41 PM, Mdev02 said:

Effects:

I did manage to pull off effects to the point that it resembles metin 1:1. There's no difference. But its ALOT of work. You need parsers again just to read all the msf, mse and msa files. Else you have to rewrite it all or spit through every single msf msa and mse file. Let me tell you, its quite alot. 

This took way a lot of time for me. I created parser and custom shader and tweaked lots of times. For some of the effects I've got 1:1 resemblence but still not perfect for all.

On 6/14/2026 at 8:41 PM, Mdev02 said:

Collision:

Collision around buildings etc. can infact be done automatically to save alot of time. Maps is where it gets a bit more complicated, you could build a seed generator and use metin2 layers in UE landscapes with a custom layer system. You need to create your own modules to truly mirror metins layering / paint system.

I don't don't how it is in UE but it was pretty straightforward. I was using heightmap values to create mesh collision and it worked perfectly for maps. And landscapes contain collision data, so it was automatic too.

On 6/14/2026 at 8:41 PM, Mdev02 said:

Graphics:

Either go with UE5 graphics or build your custom rendering pipeline on top of ue so you can get rid of most blooming/realistic lighting crap. UE APIs can help but it still requires tons of manual work, testing etc.

It's either you need to upgrade your graphics or why to even use UE at this point? I was using Godot and happy with what I see.

On 6/14/2026 at 8:41 PM, Mdev02 said:

Shadows:

You need custom shadows, not unreal dynamic shadows, not unreal static shadows. Those are heavy GPU eaters in unreal. That wouldnt represent metin2, it also looks way realer than simple shadows from 2004.

Also get rid of all HDR lighting, build your own lighting renderer that resembles metin2. 

Recreate skyboxes, clouds etc. just to mirror the same skyboxes. Metin2 has very old skyboxes so just dropping it in and betting on HDR/HDRI is never gonna work.

This part I didn't work on yet, it is right to pick anything. Because you can make static shadows and gain performance or make dynamic shadows and have day and night cycle.

On 6/14/2026 at 8:41 PM, Mdev02 said:

Characters/meshes:

 

Mobs &. Npcs? Easy. Controllable characters? Needs manual tweaking in Blender, keyframe transferring to UE compatible root bone. Recreating root motion because root motion seems to dissapear in 10/10 export->import paths.

Another issue: bone scaling, socket rotations, inverted axises, weapons not having proper origin points so offset is always there, which does not exist in metin2. Infact, EVERY offset/rotation is incorrect.

 

Now all this is mostly what the players see.

Yeah, I remember there were a lot of issues with rigging. I don't remember much but I added basic support for adding any rigging easily and had basic character movements to move around map.

On 6/14/2026 at 8:41 PM, Mdev02 said:

Next guardrails: custom networking, UE server overhead vs Metin2 old engine. Alot more ram per server. You need to know how dbs work, how scriptings work, how you program c++ with ues custom c++ layer. Minimizing rpc due to rpc overhead. Reliable vs unreliable, entire custom/rewritten backend.

 

The main issue I genuinely think most have is the amount of time. Not perse that its extremely hard (definitely requires a good amount of knowledge). But it is just alot of work which isnt required by just starting a private server, like anyone else does.

 

Then: IP, Copyright, potential risks, live ops, server hosting + a super backend that can handle multiple unreal ran server instances which run on alot more raw RAM and cpu cost than Metin2 does.

 

Nontheless, I did not stop. I did not give up and I will finish this project. But it is not without difficulties and quite frankly, its only possible for me because of my current situation. I would declare someone with little time crazy for even trying.

 

(fyi: 12-16 hours of work a day, for 1,5 years. Nonstop. I only take 1 day off a week so I don't burn out and im still not even close. I could say im at about 50% to actually release this)

The main reason why I wanted to do this project was to prevent hackers. My childhood ruined because of them. So I was sending keyboard and mouse movements and matching the movements retroactively based on the ping between server and client. I tried with packet drops, high ping simulation etc. and it was working smoothly.

I already worked previously on some MMO company with up to 50k-60k CCU, so I pretty much know what needs to be done on DB, backend, and server. For me this part is the easiest part since it's my main.

Headless servers should not use that much resources much but again I don't know much about unreal. For my case it will be for sure higher than the original server since server will calculate character movements, mob hits etc. to prevent hackers but still if you have a baremetal server it should easily handle 5-10k players.

I appreciate your afford and I wish you will finish this and have what you hoped. I wish I could have same opportunity. I have a full time work and was working on this project at the weekends etc. The main reason why I stopped is not what needs to be done, but I needed a lot of money for advertisement. Every month some popular server starts with 25k$ rewards and ads. Since mine is not something I can try again and again, I will have one shot. So I decided I need to have at least 50k$ to release it. That's why I didn't continue.

I had a lot of missing things for sure. I don't want to discourage you but I worked on this project 2.5 months partially on weekend and made a lot of progress and in that short time. I don't know your experience it changes a lot but seems like you are moving slow. If I roughly calculate I worked on this project around 200 hours and you around 6500 hours which needed to make a lot of gap in progress in between us. If I kept my phase with that 6500 hours I would finish this.

I am sorry if I discouraged you. Again I don't know your experience and what you doing is quite remarkable. Nobody will have that much free time and dedication to do this. So I hope you will be moving forward day by day and finally provide us something good.

I didn't work on the project for 1.5 years, so I am quite rusty. But if you want to talk anything about the project except when it is related to unreal, feel free to reach me.

[StormyFF 47]

On 6/15/2026 at 12:32 PM, HAMLET said:

So what is the point to move to unreal if you spend your days to  keep the old 2004 graphic and remove everything that  Unreal gives you? Cause i see none but worthless work. 

It's not a matter of "removing" from Unreal, because the Unreal Engine, being an engine suitable for both cinematic rendering and real-time in-game rendering, has great scalability and modularity. Consequently, it gives you the ability to use optimization tricks directly from within the engine, specifically designed to scale well without significantly losing AAA graphics (one of the most important systems in this regard is the nanite system, which, when used well, is incredible for performance).
Attached is a clip of how metin2 looks on Unreal Engine.

[HAMLET 15]

23 hours ago, StormyFF said:

It's not a matter of "removing" from Unreal, because the Unreal Engine, being an engine suitable for both cinematic rendering and real-time in-game rendering, has great scalability and modularity. Consequently, it gives you the ability to use optimization tricks directly from within the engine, specifically designed to scale well without significantly losing AAA graphics (one of the most important systems in this regard is the nanite system, which, when used well, is incredible for performance).
Attached is a clip of how metin2 looks on Unreal Engine.

Thank you, i know what Unreal is and what is capable of, but the man just said that he is stripping off what Unreal graphic gives so he get the old metin2 graphic in Unreal. So next time when you copy paste your answer from GPT, read the context. Also to say NANITE is an optimisation tool it means you do not know what tf you are doing in Unreal. 

[StormyFF 47]

15 hours ago, HAMLET said:

Thank you, i know what Unreal is and what is capable of, but the man just said that he is stripping off what Unreal graphic gives so he get the old metin2 graphic in Unreal. So next time when you copy paste your answer from GPT, read the context. Also to say NANITE is an optimisation tool it means you do not know what tf you are doing in Unreal. 

I copied from google translate gnegne, and nanite is an optimization system ofc, one of the best in unreal engine, but i don t expect you to understand what i wrote because you have probably no experience about unreal engine. and the guy just said u what he is doing... and what he is doing is completelly normal on unreal engine... the ultra realistic graphic is mostly used for cinematic trailer, but when you send a game in production ofc you don t need to burn your gpu after some seconds .
Pass to a graphic engine doesn t mean "ei i want the best graphic".... use unreal engine means avoid hacks, scalability, already made optimization systems, visual editing (ye we are in 2026), the possibility to run your game in editor, integrated profilation tools...  it means switch from a 2004 source to an engine... if you port metin2 to unreal just for graphics probably you have no idea what is unreal engine... or probably you just saw some videos on youtube without never run a single session by you  i love people who talks without any kind of idea about what they say, good subject for flame a bit... but atm i have no time for flame, i m in holiday, so have a great day and go study a bit unreal engine if you want talk about unreal engine to a guy who ported metin2 to (bet?) unreal engine and has been working on it for 3 years. ignorant

Edited June 26 by Metin2 Dev International
Core X - External 2 Internal

[Mdev02 4]

On 6/16/2026 at 2:02 PM, rcngrgl said:

I am happy to see someone understands what needs to be done and works on it, really liked your detailed progress report.

Yeah, parsers are a must, but once they are done, you can use every data, model and that's the beauty of it.

Heightmaps is simple as you said. I guess my only problem was creating custom shader for blending between areas, originally it was done just by blur or something like that.

This took way a lot of time for me. I created parser and custom shader and tweaked lots of times. For some of the effects I've got 1:1 resemblence but still not perfect for all.

I don't don't how it is in UE but it was pretty straightforward. I was using heightmap values to create mesh collision and it worked perfectly for maps. And landscapes contain collision data, so it was automatic too.

It's either you need to upgrade your graphics or why to even use UE at this point? I was using Godot and happy with what I see.

This part I didn't work on yet, it is right to pick anything. Because you can make static shadows and gain performance or make dynamic shadows and have day and night cycle.

Yeah, I remember there were a lot of issues with rigging. I don't remember much but I added basic support for adding any rigging easily and had basic character movements to move around map.

The main reason why I wanted to do this project was to prevent hackers. My childhood ruined because of them. So I was sending keyboard and mouse movements and matching the movements retroactively based on the ping between server and client. I tried with packet drops, high ping simulation etc. and it was working smoothly.

I already worked previously on some MMO company with up to 50k-60k CCU, so I pretty much know what needs to be done on DB, backend, and server. For me this part is the easiest part since it's my main.

Headless servers should not use that much resources much but again I don't know much about unreal. For my case it will be for sure higher than the original server since server will calculate character movements, mob hits etc. to prevent hackers but still if you have a baremetal server it should easily handle 5-10k players.

I appreciate your afford and I wish you will finish this and have what you hoped. I wish I could have same opportunity. I have a full time work and was working on this project at the weekends etc. The main reason why I stopped is not what needs to be done, but I needed a lot of money for advertisement. Every month some popular server starts with 25k$ rewards and ads. Since mine is not something I can try again and again, I will have one shot. So I decided I need to have at least 50k$ to release it. That's why I didn't continue.

I had a lot of missing things for sure. I don't want to discourage you but I worked on this project 2.5 months partially on weekend and made a lot of progress and in that short time. I don't know your experience it changes a lot but seems like you are moving slow. If I roughly calculate I worked on this project around 200 hours and you around 6500 hours which needed to make a lot of gap in progress in between us. If I kept my phase with that 6500 hours I would finish this.

I am sorry if I discouraged you. Again I don't know your experience and what you doing is quite remarkable. Nobody will have that much free time and dedication to do this. So I hope you will be moving forward day by day and finally provide us something good.

I didn't work on the project for 1.5 years, so I am quite rusty. But if you want to talk anything about the project except when it is related to unreal, feel free to reach me.

Thanks for the reply. I only mentioned a very little part of the project, it's alot larger and I cannot write it all out. It was mostly a summary of some fundamentals. I must also mention that i'm thinking alot about scalability here, I don't set the goal just to re-create Metin2 as is, in the simplest way. My goal is alot bigger than that, hopefully understandable, I don't want to spill all the tea ;). You definitely did not discourage me, I love the detailed reply and i'm glad we still have some people with knowledge roaming around in the community. I do have to reply to the part about the graphics though, because that's not only what Unreal Engine offers, also i'm not stripping out everything to the point the Engine does not support it. What I can say is that the approach is to support both, we have a mixed community and alot that Metin2 offers, is the old feel. For it to actually work and succeed, the best approach is to support both. Keep the engine support when it comes to any upgrades, newer versions, potential revamps.

The time part is something I think you are underestimating though. I also had to scrap a initial prototype after the first few months due to some unexpected setbacks. Which led me to restart and the current project i'm working on has been since somewhere in October last year. However even with 6500 hours, I genuinely think you are underestimating how much extra work comes when creating a MMO, Unreal is not plug and play. It still requires alot of manual work, manual testing, profiling, stress testing etc. Now think about gameplay, quests, item balancing, player balancing, pvp/pve balancing, the thousands and thousands of items. Whilst revisioning everything to a better, scalable architecture. I think the amount of time can be highly underestimated once you start touching the real problems and not surface-level problems that can be solved in minimal time.

The major thing Unreal Engine offers though is absolutely not the graphics. The engine offers alot of development and designer support. We cannot compare the capabilities of a engine like Unreal vs a Metin2 engine, the long-term scalability of a engine like Unreal is unmatched when you compare that to the one Metin2 offers.

The heightmap/collision part you mentioned does suggest that your approach relies heavily on standard engine collision and terrain. That is perfectly fine, but my own goals are slightly different because I am optimizing for MMORPG-scale scenarios. Default Unreal movement is extremely capable, but it is basically designed around lower players in the same zone and characters that can afford very expensive movement and collision calculations. Large MMORPGs mainly move towards much simpler server-side representations such as grids, cells, baked height information and simplified validation because the cost per player becomes extremely important once hundreds of players are in the same area. That does not mean Unreals approach is wrong. It simply optimizes for a different problem than the one I am trying to solve. Unreals own documentation suggests not to use this and create a custom model instead when trying to scale above their own given limit which is something like 100-150+ players (I don't have the exact numbers, i'm writing this off the top of my head).

About headless server, ofcourse a headless server is cheap but it's not as cheap as the one from Metin2 because Unreal brings alot of the engine and pre-done work into the server. Think of Actors, reflection, uobject management, servers also initially carry alot of general-purpose work that Unreal wants to support which I might not want to use currently or at all. There's no 1 press button to switch it to my liking so it'll be optimizing & testing over-time.

I do also have to say, i'm not going off parsers with everything, I am, in-fact, re-creating entire data tables, data assets etc. Filling them in partially manually and sometimes using import tools to fill them partially. I am doing alot of manual work, the game will be Metin2 but I am basically rewriting every single system of Metin2 into a much better system for both developer, designer and user.

Another thing that is highly underestimated: assets from 2000's era in a hyper-realistic world is not what you think it is. Those are designer choices and decisions that have alot more impact in a game than just a 5 minute youtube video that looks cool.

On 6/25/2026 at 9:36 PM, HAMLET said:

Thank you, i know what Unreal is and what is capable of, but the man just said that he is stripping off what Unreal graphic gives so he get the old metin2 graphic in Unreal. So next time when you copy paste your answer from GPT, read the context. Also to say NANITE is an optimisation tool it means you do not know what tf you are doing in Unreal. 

I actually did not say stripping off, that's something completely different. I only mentioned what the options are for graphics. I built my own rendering on top of Unreal and disabled from Unreal what does not represent Metin2. That does not mean it's gone, it's currently disabled during development. The Engine still offers the option to upgrade, to change graphics and perhaps on release it'll be alot different. I do have to correct you, Nanite is absolutely an optimization tool. Just to give you 2 things to look into when it comes to Nanite and optimization that you might not have thought of: Draw calls & LOD workflows. Nanite is not just bumps in the ground. Now whether it actually improves performance depends on the content and target hardware. There's a degree of truth that Nanite is probably not optimal for hardware from 2000. But for most cases, Nanite is absolutely a plus for the hardware 99% of consumers use nowadays.

I think the main thing I can say. Unreal is alot more than just graphics, Unreal is also alot more than just a better engine compared to Metin2. It gives you massive control as a game creator. I am not trying to recreate Metin2's implementation however, I do not want their systems and the way they set everything up. I am trying to recreate the Metin2 experience while building on a modern foundation.

Edited June 29 by Mdev02