Thrones of Gandhar

The pitch I kept repeating to myself was this: you build a society that can sustain a war machine, while you command that war machine.

Every resource . Every log, every stone, every gold bar gets physically carried by a little person from one building to the next. If your sawmill is on the other side of the map from your woods, you will feel the strain.

You control an army and a hero. A hero who levels up and keeps their gear between games, armies with formations and stances, towers you can garrison, spells, veteran ranks. Once your soldiers exist, it should feel like a proper RTS, not like a spreadsheet with swords.

The two systems stay separate but they lean on each other. The society makes the army, the army protects the society. Lose a key supply chain and your army should actually hurt.

I want to be upfront about this because I think it’s the interesting part. I didn’t hand-write the code. I worked with Claude Code as my engineer and I was the director, designer and the most annoying playtester alive.

My job was to know what the game should feel like and to keep saying “no, not that” until it did. I made the art & designs and played it every evening and wrote up what sucked.

Each playtest became a “bug fixing round”. Some rounds were tiny fixes. Some threw out the entire economy.

Version one was a straight-up clone of Warlords Battlecry III running in a browser tab. Plain HTML canvas and JavaScript with no build step

Then I bolted the Settlers economy onto it: around 30 goods, 25 professions, 33 buildings, carriers who walk goods around, woodcutters who actually chop trees, foresters who plant saplings that grow into forest.

It worked! It also looked like a pixel-art prototype, and that wasn’t the game in my head. I moved the whole thing to Godot 4 and started again on the art.

I built a Pinterest board and gave myself a brief: not low-bit pixel art. Stylised, texture-painted 3D, rendered down to isometric sprites. Fantasy, but a little gritty. Not too cartoony, not grimdark either.

To actually decide, I rendered the same barracks in three looks and put them side by side, both close up and at real in-game size on the terrain.

This is the part I’m kind of obsessed with. Every building is modelled in Blender stone by stone and tile by tile. Every roof tile, plank, beam and log is actual geometry, then a hand-painted shader adds worn edges, grime creeping up from the ground, moss, cracks, and the odd missing tile for grit.

Every map is a randomly generated island. Hills and ridges, rivers that actually flow downhill to the sea and join up as tributaries, ponds, beaches. You can wade across a river, slowly.

Later rounds added tall mountains nobody can cross except flyers (the game cuts a pass if they’d wall a kingdom off), and biomes. Desert slows you down and won’t grow crops. Tundra is even slower, and horses hate it.

Characters were their own little saga. My first rule for them was “pre-rendered and realistically proportioned, like Warlords”. Then I played it and realised I actually wanted the Settlers feel: stocky little people you can read from far away.

So we went stocky. Bigger heads, bigger hands and feet, slightly shorter legs, oversized tools. And every trade got its own signature hat so you can tell who’s who at a glance: green felt cap with a red feather for the woodcutter, a brass helmet with a lamp for the miner, a straw hat for the farmer, a white toque for the baker.

This was my verdict after round five, and honestly it’s the most important thing that happened on this project.

By then the economy was super deep. Households slowly grew settlers if you fed them. Recruits came out of your population. There were tools for every profession, a toolsmith, an armsmith, around 30 goods. On paper it was everything I loved about Settlers. In practice I was spending all my time staring at shortages and none of it having fun.

So I rewrote the spec. Round six threw out the population mechanic completely and squeezed the economy down to five resources, each with exactly one processing step:

My two launch requirements were: a good readable UI. So the UI went back to the Warlords layout I grew up with. Slim bar on top with your five resources, food, mana, army size and carriers. At the bottom: minimap, hero portrait with health, mana and XP bars, a display area for whatever you’ve selected, and a command grid.

For the skin, I built dark bronze buttons that light up molten gold when you hover, bronze-rimmed slots with a grungy texture, parchment strips, dark panels.

A few of my favourite things on the military side:

• Tower capture duels. Regular troops can’t knock down a defended tower, they have to take it. At 70% health it opens up for one-on-one duels: one defender steps out, one attacker steps up, everyone else waits. Winner holds the ground.
• The phalanx. Four or more Valerian men-at-arms or pikemen can lock shields. They get a lot tougher and arrows barely touch them, but they’re slow, and dragons absolutely wreck them
• Blessings. The temple turns wine into mana, and mana levels up a whole branch of your army at once (foot, archers, riders or mystics), including soldiers already on the field
• Food morale. A well-stocked kingdom fights better. Well Fed, Feasting, Abundance, each one a bit more damage and armour
• Weather. Rain makes fields grow faster and arrows worse. Snow slowly creeps over everything, slows everyone down, and eventually freezes the rivers so your fishermen just stop

The game’s called Thrones of Gandhar for a reason, so a faction rooted in Indian history was always the plan. So I brought in the Aryavartans.

They share every system with the Valerians but have their own names, recipes and art. Their homes are haveli-themed. Their food comes from walnut farms, mango farms and cattle farms, and their mana comes from a Soma Brewery instead of wine. They train war elephants that wreck buildings and Garuda as flyers, and once a game they can call down Indra, a demigod who throws lightning.

Pretty art has a cost, and I found out the fun way: scrolling started stuttering. Turns out the unit sprite sheets were huge and uncompressed, around 150 MB of video memory per unit type. Oops.

I shrank the sheets, compressed them properly, loaded them in the background when a game starts, and moved the minimap drawing onto the GPU (it had been freezing the game for about 145 ms every second). An early game went from about 40 fps to a solid 60 on my M2 MacBook.

Same energy on the computer opponent: finding a spot for a farm used to take 210 ms. It now takes 8 ms, with identical results.

Play it, every day. Every big decision on this project came out of a playtest. “It isn’t fun” is a useless sentence in a meeting and the most useful sentence in the world after an hour of actually playing.

Compare things side by side, at real size. The style tests saved me from at least two wrong calls. Always look at the thing the way the player will see it.

Rules make a world. Symmetrical windows and hollow chimneys sound like nitpicks. But a hundred buildings following the same rules is what makes it feel like a place.

Directing is a real skill. As a designer my job is mostly taste.

More peoples are sitting in the data waiting for their turn, along with wall designs and a proper balance pass on the computer opponents. It runs on Mac and Windows already. If you grew up on these games like I did and want to try it, send me a message.