Dragon Shelter · Chapter 06

Polishing the World

Seven hours of Dragon Shelter exists. This is the chapter where we stopped adding and started fixing: the hundreds of small corrections that turn a working build into a game that feels good to touch.

We sat down and played it for eight hours

At some point you have to stop building and just play your own game like a stranger would. So we did — roughly seven to eight hours straight through the current build, notebook open. The verdict was honest and a little uncomfortable: there's a real game in here, but the experience keeps tripping over itself. After a couple of hours the core loop starts to buckle under its own scale. The same actions repeat without escalating, feedback lags behind your progress, and a newcomer simply can't tell what half the systems want from them.

We wrote it all down as a pre-alpha audit and grouped the mess into six honest buckets: how the systems connect (quests, farm, kitchen, dragons, town, seasons, elements all exist but don't yet click into one loop), repetition at scale, onboarding, UI consistency, controls, and pure game feel. Seasons should change how you care for dragons and run the farm; dragon elements — fire, water, earth, electricity — should do something to the farm and the town instead of just being flavor. Naming these problems out loud, in one place, is what let us stop guessing and start fixing.

Another reworked house — cleaner silhouette and warmer detail in the project's style.
Another reworked house — cleaner silhouette and warmer detail in the project's style.
A working game and a game that feels good are two different builds. We spent this stretch closing the gap.

Readability first: make the world easy to read

When we audited the look of the live build, one priority rose above everything else: readability. Before anything pretty, the player has to be able to see — pick out objects, read the gameplay, and not get lost in visual noise. We deliberately reached for the fixes that were fast to ship and paid off immediately, which meant starting with terrain and effects.

So we went through the world surface by surface. The grass terrain now matches between the farm and the town instead of feeling like two different places. We toned down the ground tiles so the character actually stands out on them, and made the grass and plants less noisy and more varied in color — better to look at and easier to parse mid-action. We hunted down the blown-out, overbright spots in the lighting and pulled them back. Small things on their own; together they make the whole scene calmer and clearer.

The pickup effects got the same treatment. The collectible highlight is now a crisp outline that reads clearly without shimmering and shouting. Berries get a quieter glow; fireflies are yellow, flowers are purple, and you can finally tell them apart at a glance. Garden-bed effects and plant visuals were corrected so a planted row is legible at speed.

Rebuilding the houses, sharpening the UI

In parallel we reworked the town's buildings. Three houses went through a full before-and-after pass, and the difference is the kind you feel before you can name it — cleaner silhouettes, warmer detail, better fit with the rest of the world. Just as importantly, we tightened our own pipeline along the way: the time to get a finished house in the project's style dropped from about a week to roughly two days, working in small, fast iterations with concepts and references up front.

The audit was blunt that the single most critical visual problem is the UI — different screens still look like they came from different games. So that's the next front: a shared visual language of colors, icons, and state indicators, an audit of the key screens (inventory, quest log, orders, the dragon menu), and a single control map so keyboard-and-mouse and gamepad finally agree on what the primary action does in each context. All the visibility work this period lives in our art/visibility_fixes branch, ready to fold into the build.

Sketches & process

before-afterWindmill before-and-after — clean build beside the upgraded version with crops, lanterns and well.
Windmill before-and-after — clean build beside the upgraded version with crops, lanterns and well.
before-afterBefore and after comparison of the painted pine trees, tightening color and density for better readability.
Before and after comparison of the painted pine trees, tightening color and density for better readability.
in-engineDragon Shelter gameplay in the build: a dragon bouncing on a crafted trampoline prop in the lush forest world.
Dragon Shelter gameplay in the build: a dragon bouncing on a crafted trampoline prop in the lush forest world.
in-engineScale and staging check pairing the player character beside the textured dragon companion in the Maya scene.
Scale and staging check pairing the player character beside the textured dragon companion in the Maya scene.

What we built