Last time I wrote about wanting movement to feel good. This is the follow-up: a record of the decision to throw out a feature I had built myself, all for the sake of smooth movement.
I couldn’t build “traversal” well
This game has movement where you climb and vault over ledges and walls — the common thing where you clamber over an obstacle to keep going.
At first, I built this myself. When you find a wall, grab it, pull your body up — assembling the motion one piece at a time. It works as an action. But it doesn’t come out smooth. Somewhere it catches; the character clips into the wall; the tempo after the vault breaks down.
I spent days on small adjustments, but it was a constant loop of fixing one spot and breaking another. What I thought was a single feature turned into something I couldn’t climb out of.
Smoothness wasn’t a “feature” — it was a result
As I dug into it, one thing became clear.
That natural, smooth motion isn’t something you can build as a standalone feature. It’s a texture that emerges as a result of continuously picking, in real time, the motion that best fits the situation from a huge database of animations.
In other words, the way I was doing it — assembling motion by hand, one piece at a time — could never reach that smoothness in principle. No amount of patching the seams would produce the real texture.
I dropped my own work and built on the real thing
So I decided to throw out the traversal I had built.
Instead, I’d take the movement system that Unreal Engine provides officially and use it directly as the foundation. On top of that, I’d lightly layer my own things — my character’s appearance, and the elements specific to this game (like the flick-to-dash from last time).
My approach is to build the parts worth building myself, and use ready-made things where they do the job. What I want to make is something that feels good to play. In solo development, where you carry everything, I came to think that the most important skill is judging what to build deeply yourself and what to leave to something already made.
Borrow the foundation, add only your own color
This isn’t limited to traversal — it’s become the overall approach to how I build this game.
Smooth movement, traversal, multiplayer sync — for these foundational parts, I use what the specialists have built, without hesitation. On top of that, I layer “this game’s color” — combat with real decisions, the feel of movement, your own character — carefully, without breaking the foundation.
Keep what’s good about what you borrowed, and add only your own character on top. For now, I see it as the realistic way to finish something at a decent level of quality on your own.
Next time, I’ll write about the design behind not letting lag delay your inputs in co-op combat.
