The fastest way to stall a stealth prototype is to build before you decide what “good” means. This lesson keeps you in design mode long enough to define a mission fantasy, success conditions, fail states, and a hard scope box. When you open Unreal in Lesson 2, you will already know what one “done” mission looks like.
Lesson Objective
By the end of this lesson you will have:
- A one-paragraph player fantasy and target session length for the slice
- Three design pillars that steer every feature decision
- Written success conditions, fail states, and a cut list for v1 of the vertical slice
Step 1: Name the fantasy in one breath
Answer this in plain language, no engine terms:
“In this mission, the player is someone who [identity/role] and must [verbs] inside [place] while [pressure] is trying to stop them.”
Strong example: “An infiltrator who must reach a rooftop relay and plant a device while patrolling guards and cameras tighten security as time passes.”
Weak example: “A stealth game with cool AI.” (No place, no objective, no pressure.)
Keep the fantasy specific enough that you could sketch the opening shot of a trailer. If you cannot imagine that shot, the fantasy is still fuzzy.
Step 2: Pick target session length and cadence
Stealth lives or dies by pacing. Choose one primary session window for this vertical slice:
| Target playtime | Best for |
|---|---|
| 5 to 8 minutes | Portfolio slice, one core loop proof |
| 10 to 15 minutes | Greenlight-style vertical slice |
| 20+ minutes | Usually too large for a first UE vertical slice unless you already have production tools |
Write your chosen window at the top of your design doc. Every system you add must justify its cost inside that window.
Pro tip: session length is a budget, not a wish. If your beat sheet exceeds the budget, you are writing a full game, not a slice.
Step 3: Define three experience pillars
Pillars are the filters for ideas. Pick three short labels plus one sentence each.
Example set for an industrial infiltration slice:
- Readable tension – Players always understand why they were seen or heard.
- Fair pressure – Alerts come from clear rules, not random noise.
- Route variety – At least two viable paths through each major beat.
Test every feature spike with: “Which pillar does this serve?” If the answer is vague, defer it.
Step 4: Write success conditions (definition of done)
Success conditions should be observable. Use this checklist format:
- Win state: What exact world/event marks mission complete? (e.g. “Player interacts with exit marker after objective A and B.”)
- Skill check: What must the player demonstrate? (e.g. “Bypass or neutralize at least one patrol without full combat.”)
- Showpiece moment: What moment proves the slice is special? (e.g. “A search phase where guards coordinate along a single flank route.”)
If you cannot verify a condition in a playtest, rewrite it until you can.
Step 5: Map fail states and fail-forward rules
List how the player can lose or slip, and what happens next.
- Soft fail: Spotted, enters search; player can recover by breaking line of sight and hiding.
- Hard fail: Health reaches zero, or mission-critical alarm locks doors.
- Fail-forward (optional): Combat unlocks as a worse path (loud route) instead of instant game over.
Decide now whether combat is in scope for v1. Pure stealth vs. stealth-with-escape-combat changes AI, UI, and animation workload massively.
Step 6: Cut list and scope lock
Create a “not in v1” list before opening Unreal. Examples:
- No day/night cycle
- No loot economy
- No multi-floor open world
- No fully voiced cinematics
Put 8 to 12 bullets on that list. This document is your shield when you want to add “just one more system.”
Mini challenge
Draft a beat sheet with 5 to 7 timestamps (e.g. “0:00 rooftop entry”, “3:00 first patrol puzzle”, “6:00 mini climax”). Each beat must fit your session budget from Step 2. If two beats need the same mechanic back-to-back, merge them.
Troubleshooting
| Problem | Fix |
|---|---|
| Fantasy keeps expanding | Force a single primary objective; file everything else under v2 |
| Pillars sound generic | Tie each pillar to one mechanic you plan to ship in the slice |
| You cannot define win/lose | Prototype a paper map of the mission before greybox |
| Scope creep in “optional” features | Anything “optional” needs a cut line date, or it ships as a note only |
Summary
- Lock fantasy, session length, pillars, win/fail, and cuts before deep production.
- Treat the vertical slice as a measurable proof, not a mini open world.
- Next you will mirror this plan in Unreal 5.5 project setup and folder standards so assets and Blueprints scale without chaos.
In Lesson 2: Unreal 5.5 Project Setup and Folder Standards, you will create the Unreal project skeleton, naming conventions, and source-control habits that keep this mission traceable as art and AI land.
Further reading
- Unreal Engine documentation hub – official reference for UE 5.5 features and tooling.
- Unreal Engine guides – broader engine context on this site.
- Unreal Engine 5.5 packaging troubleshooting – bookmark for the packaging lesson.
FAQ
What if I want an open-world stealth game later?
Start with a linear slice anyway. Open-world stealth needs tooling you will discover only after one shipped-quality encounter loop.
How strict should session length be?
For learning, ±20 percent is fine. If you double the target without adding time, you are re-scoping.
Do I need a GDD before Lesson 2?
You need this lesson’s outputs, not a publisher packet. One page is enough.
Combat in stealth: required?
No. If combat is out of scope, write “detection equals puzzle reset” or similar so designers know the boundary.
Bookmark this lesson when your pillars are stable, and share your one-line fantasy with a teammate or friend. Fresh eyes catch vague verbs early.