How to make a video game

The engine matters less than finishing. Every engine above has shipped successful games, and the skills you learn in one transfer to the others. Pick by the game you want to make first. For a 2D game, Godot, GameMaker and GDevelop are all comfortable. For 3D, Godot and Unity are friendly to beginners, and Unreal is the choice if you're after realistic graphics and don't mind a steeper climb and a powerful computer. If you can't code at all and want results this weekend, GDevelop or Scratch will get you there.

Licences are worth a glance. Godot is free under the MIT licence with no royalties. Unity and Unreal are free to start, with fees or royalties once a game earns past their thresholds; both have changed their terms in the past, so read the current page before you commit. GameMaker is free for non-commercial use with paid tiers for selling on some platforms.

A tiny game, like a single-screen arcade game, can take a weekend once you know an engine. A small polished game takes most people a few months of evenings. A first commercial game on Steam is often one to three years for a solo developer. The single best way to shorten that is to shrink the idea: fewer levels, one mechanic done well, one setting.

Game jams, where you make a game in 48 hours or a week to a theme, are the fastest way to learn. They force a small scope and give you a deadline, and itch.io lists dozens running at any time.

New developers expect code to be the hard part. Usually it's everything around it: menus, a pause screen, saving, sound for every action, a title screen, settings, what happens when you lose. Budget time for it. Sound in particular is cheap to add and changes how a game feels more than almost anything else; even simple effects for jumping, hitting and collecting make a prototype feel finished.

If you're not an artist, lean on a limited style: flat colours, low-poly shapes, or a small pixel palette. Free asset packs with a clear licence (CC0 means you can use them for anything) are a perfectly respectable way to ship.

Most first games are solo projects, and that's fine. When you get stuck, the engine's own forum or Discord is usually the fastest place to get an answer, and searching the exact error message solves more problems than any book. If you want collaborators, game jams are where artists, musicians and programmers meet; teams formed for a weekend often keep working together afterwards.

Keep a short devlog as you go: a screenshot and two lines a week. It keeps you honest about progress, and when you publish, it becomes the story you tell players about how the game was made.

AI game makers now write the whole game for you. On Modder you describe the game in a sentence, and an agent plans it, writes the code in a real engine (three.js with Rapier physics), picks or generates art and sound, then plays the game in a browser, looks at screenshots and fixes what it finds. A first build takes five to twenty-five minutes, and you change it by asking: "make the jump higher", "add a shop".

It isn't a replacement for learning an engine if you want to make a big, original game, or to sell on consoles. It's great for testing whether an idea is fun before you spend months on it, for making something to play with friends tonight, and for people who will never have time to learn Unity. Making games on Modder needs a plan or a credit top-up; playing them is free.