Sep 22, a little after two in the afternoon, I typed: “Still cruzing? I see no progress happening”. Soon after that, I asked the question this whole chapter hangs on:
“Did we make all the assets for this game yet? I’m not seeing armors, proper buildings, walls, beasts, items… Not sure why we’re testing if all the game assets aren’t even in there.”
And right behind it: “wasn’t the full game map supposed to be rebuilt? … idk what your strategy is or how you’re justifying it.”
That question changed the whole plan. This is the last chapter of Season 0, six days long, and it’s the week we stopped polishing what was there and started building what wasn’t.
What got built
An honest count
First we counted everything in the game against everything in the design. No rounding up. Here’s where PRIDE actually stood on Sep 22:
- Armor: 12%. 41 of 58 armor pieces had no 3D model at all. The 17 that did were sharing 5 models between them.
- Weapons and items: 55%. Every one of the 419 items had an icon. But 204 of them had no body in the world, and 68 different weapons all showed up as the same 8 shapes.
- Buildings: 35%. 30 building types were sharing 14 shells. A prison was just a courthouse. A smithy was just a forge.
- Walls and forts: 25%. 16 wall, gate, tower and bridge pieces were sitting on disk, and the game used 2. Bridges never showed up anywhere.
- Beasts: 33%. 12 of 27 species were finished. 5 had no model at all.
- Characters: 62%. All eight races were locked, but the skins I’d approved weren’t even switched on by default.
- The map: 35%. A whole new world was designed, with 2,300 meter mountains, real rivers and lakes, and 440 natural wonders. It had never been built.
No new kind of asset had gone into the game since July 27. We’d spent weeks tuning frame times and lighting on a world that was missing most of its stuff.
Content first
Within minutes there was a new plan, and it was simple. One goal: a world dressed in finished content that four friends can play together for two hours with no problems, where a brand new player can make it through their first hour without help. Performance and polish got parked until that was real. When we parked it, the game was running at a median of 10.8 milliseconds a frame at 1440p on Ultra in the spots we tested, so it could wait.
That night I added one more rule. We’d just moved to Unreal Engine 5.8, so I asked: “We fully updated to ue5.8 right? We have queued up all the new changes and tools that gives us?” Once I had my answer, I wrote: “get us there. Assure we’re using cutting edge best practices to achieve our mission that should be at the forefront of our plan.”
The next day I added another. From then on, finishing meant something: a piece of work was finished when it was proven in the game, not when it was written. I put it this way at the time: “to note, i want to assure our plan has us finishing lanes of development in an intellent way.”
The first things to land
The content started landing that same afternoon:
- The skins I approved went on. The approved character skins became the default.
- Towns got their real walls. Authored gates, towers and wall pieces went in, 19 wall pieces and 129 building models in the first import.
- Sound. Footsteps by surface, for dirt, wood, mud, sand and snow. Calls for the creature families that had been totally silent. UI sounds and work sounds, all from free, properly licensed sources. We started a license ledger so nothing ships on a guess.
- Work you can see. The first work animations: hammering, grinding, tanning, and chopping wood.
- An item factory. A setup that takes an item from an image to a 3D model to the game, piloted on 12 items. By Sep 24 it was 142 of 201 through.
- The first hour. Eating, drinking, resting, and working in another town all got wired for a new player.
When I saw the new town frames that evening, I didn’t sugarcoat it: “Need major adjustments, these do not look like manorlords quality or anything close to it.” So that became the bar. Towns at Manor Lords quality, in our own look. Terracotta and lime plaster, not thatch. The way we wrote it down: “Style is ours; quality is theirs.”
The town grid had to go. Towns had one main street and evenly spaced side streets. After the change, roads grew out from the market and followed the land, the way a real town would. New house shapes went in too: a great hall, two-story houses, L-shaped houses, courtyard houses, cottages and stores.
A new world, rebuilt from the ground up
The other half of my question was the map. The whole world got rebuilt through a process we call the bake, and it fought us the whole week.
By Sep 27 the rebuilt world had these rules, and all of it was still changing:
- Rivers ran downhill. There were 59 rivers, and 54 of them reached the sea. None climbed more than half a meter.
- Lakes sat where lakes go. Turns out “every lake in the world had been drawn at SEA LEVEL until now.” After the fix, 198 lakes were drawn whole at their real height.
- Coasts got a real shelf. The sea floor fell away gradually instead of dropping off a cliff, so things could stand in shallow water. There was still no sea travel, and this world had no word for a ship.
- You could see the world. At full zoom out, the camera tilted so you could see up to 1,267 meters ahead. There was no minimap in PRIDE, so landmarks on the horizon had to pull you toward them.
- History was on the map. Ruins, watchtowers, forts and camps had been hardcoded, mostly at spots that weren’t even on the map anymore. After the change, 868 sites and 17 ancient falls were placed from the world’s own history on day one.
- Food out in the wild. I said “things should be distributed around the WORLD in addition to the starting town locations,” because if you walked out of town there was nothing to gather. After the change, no spot of land was more than a kilometer from something you could forage.
- A bed on night one. Every settlement had somewhere to sleep, from an inn in a town to bedrolls by the campfire.
- Only rivers counted as water. On Sep 23 I ruled that only rivers got to shape a town around them. Dry canyons and washes didn’t.
On Sep 27 at 3 in the morning I made my favorite call of the week. A batch of cliffside valleys on the coast kept getting dropped because their spots landed in the ocean. I wasn’t having it: “make the shuffle not be in the ocean? … we need a beautiful immersive love to explore passion adventure map.” So the rule became that when a place worth finding can’t go where it was planned, it moves to land that deserves it instead of getting deleted.
What went wrong
A new player couldn’t earn a single coin
We had a test player play all three of the game’s openings for an hour each. It earned zero. Nothing. When we dug in:
- Nobody could find their way. The whole world’s navigation covered one 2 kilometer box, and the places people actually live were 17 to 41 kilometers away from it. In one hour, 281 different people tried to walk somewhere and failed.
- There was nowhere to work. One town had filled all five of its jobs on its own, and 149 people were stuck doing odd jobs.
- There was nowhere to sleep. No inn existed anywhere in the game.
- There was nothing to pick up. One opening woke you up nearly 2 kilometers from any town, with nothing to gather in sight.
All of it got fixed in code that week. On the navigation I ruled with one line: “study best practice for this and implement the solution”. On work for strangers I asked for it to be “logically built out and implemented”. The result was real navigation where people live, a labor market for strangers, beds, and food in the wild. The proof run was still waiting on the new world when the week ended.
The 6 cm house
A model file says what units it uses, and the importer ignored it. So we got a house 6 centimeters tall, and a gate that measured 0.1 meters.
The bake, step by step
Rebuilding a whole world means it breaks in creative ways:
- The ocean covered the entire map, including a crater 300 meters below sea level. The ocean’s outline had simply never been set. After the fix it followed a traced shoreline with 927 points.
- Wells to nowhere. A scan of the land found 34 holes deeper than they were wide, one of them 900.9 meters deep.
- A blue hole ate a river. A deep sea pit sitting on the Sulsrun’s path drowned the last 5.6 kilometers of it.
- Two towns ended up underwater, one under 4.5 centimeters of lake and one under 22.
- A river ended inland because a step that simplifies shapes dropped the point where two rivers meet.
- All 198 lakes failed to build. The fix was bending every lake edge out by 2 centimeters. Then all 198 worked.
- The engine ran out of room. Unreal 5.8 had a table with 1,048,576 slots, and loading every distant piece of the world at once overflowed it.
Four players, two dropped
The first four-player test lost two of its three friends while they were still joining. Building their copy of the world took longer than the 60 second join window. The window was raised to 180 seconds. That run didn’t count as proof, and we said so.
What I learned
Count before you polish. We’d been tuning a world that was missing most of its content. The census took one afternoon and changed everything. The question that decides what’s next is simple: does a player in a four-player game actually run into this?
Done means proven. A piece of work is finished when it’s proven in the game, not when the code is written.
Use what the engine gives you. Knowing about engine features doesn’t count until it’s checked against the engine itself. Installing 5.8 was step one. Using its best stuff is the real work.
Don’t delete the good stuff. If a place is worth finding, move it somewhere that deserves it. Losing it is the last resort.
Even the reviewer can be wrong. At one point a review flagged a problem as critical, and the review itself turned out to be wrong. We wrote that down too.
Next up
That’s Season 0, from the first commit on Feb 4 to Sep 27. The Forge moves on to weekly devlogs from here.
September 22 to 27 in numbers: 382 commits · 6 of 6 days with a commit · busiest day Sep 23 (112) · Unreal Engine 5.8 · 5,206 total by day 235.
What’s the one thing a game has to have in its first hour to hook you? Drop it below.
Colophon
Season 0, chapter 7 of 7. The images are real captures from the build, and the chart is drawn from the record.
Following the build? Sign the book so I know you’re out there
From the toolbox The Toolbox · A random post · All tools →
Free
Join the Guild
The free Weekend Field Manual, dev log alerts, and guild dispatches for announcements. Zero spam.
Join the Guild
Keep reading
Next in the Forge
From the Forge archive: the night it started (February 2026)
The Forge
Devlog 008: no magic anything
Throw five on it