Technology (decisions to validate, no code)

Document 05 · 4 min read

Client engine

Option	Pros	Cons
Unreal Engine 5	Lumen and Nanite get closer to the reference (night lighting, volumetrics, open world with World Partition); mature first person and animation	Heavy for an MMO (many players on screen are expensive); 5% royalties above US$ 1 million in gross revenue; the team needs to know C++/Blueprints
Unity 6 (HDRP)	Lighter, large ecosystem, good for small teams	Giant open world and night lighting require more manual work
Godot 4	Free and open	Open-world tooling and MMO networking far less mature

Platform: PC only at first (owner's decision), with a VR port planned. Engine proposal: UE5, because the reference is above all lighting (the whole cycle, with night as the high point) and scale. Confirm in Phase 1 with a test scene running at 60 fps on a mid-range GPU (RX 6500 XT / RTX 3060), since not every card handles Lumen. UE5 also has mature VR support (OpenXR), which weighs in its favor for the future port; Lumen in VR is expensive, so there are two lighting profiles (decided): full PC (Lumen) and reduced VR (baked lighting plus simplified dynamic light), with a 90 fps budget defined from Phase 1A. The art is made to read well in both profiles (cool × warm contrast through materials and emissives, not just global light).

Server

An MMO requires an authoritative server: the client never decides damage, loot or final position. The server validates every hit (speed, range and trajectory); VR hand input is treated as untrusted.
Concurrency goals (decided): designed for 10,000 players per world at launch, ~500 per region server, ~100 visible in a city; clan wars limited to 50 vs 50, alliance wars 100 vs 100, in instanced battlefields.
Regional servers (decided): Americas first, with São Paulo and US East; goal of latency < 80 ms within the region. Other regions later.
Options: SpatialOS-like/in-house (zones split into cells, each cell in a process), UE dedicated server + own sharding layer, or frameworks like Nakama (social, accounts, chat, auction) together with in-house game servers.
Persistence in Postgres (accounts, characters, items) + Redis (session, queues).
Start with one region = one server and only split into cells when population demands it.

Network

First-person action combat needs client-side prediction and server reconciliation (shooter style), harder than WoW's tab targeting.
Goal: 30 ticks/s in combat, 10 in exploration; ~100 players visible in a city with aggressive level of detail; latency < 80 ms in the region.

Open world

Streaming by blocks (World Partition in UE5), levels of detail for giant trees (impostors), instanced vegetation.
Dynamic light for the whole cycle: sun and moon as moving directional lights, sky and cloud system, real-time global illumination (Lumen). Baked light cannot be used, because the time of day changes. Lanterns and windows are emissives that turn on at dusk, with a few real dynamic lights near the player.
Weather as world state: rain, fog, storm and snow, with wet surfaces and particles.
Clock and events on the server: a scheduler keeps the time, moon phase, season and weather of each region and triggers the dynamic events; the client only displays. This way all players see the same time and the same event.

Content tools

Procedural terrain with manual editing on top (Houdini/UE5 PCG) to reach 16 km² without a huge team.
Generative AI only for concept drafts and internal reference; final assets must be original or clearly licensed, to avoid rights problems.

Costs and team (order of magnitude)

An MMORPG is the most expensive genre to produce and maintain. To plan honestly:

Phase 1 (vertical prototype): 1–3 people, 3–6 months, almost no server cost.
Playable alpha with two regions (the 16 km² launch block): team of 8–15 people for 18–30 months.
Complete world (1,120 km², 18 dungeons, 4 raids, 17 paradises, 4 layers, boats): outside this estimate; needs its own budget and team, and only enters in phases after launch (see 06).
Servers, storage and anti-cheat become a fixed monthly cost as soon as there are players.

Distribution, wallet and age (decided)

PC distribution: own launcher + Epic Games Store. Steam is not planned (its policy bars games with blockchain and NFTs; recheck before launch).
Wallet: embedded and non-custodial (e.g., Immutable Passport); the studio never holds players' assets. Conversion to cash only through a licensed on/off-ramp partner.
Minors: wallet, token, market, NFTs and voice chat only for verified 18+ accounts; players under 18 play the full game without the economy features.