The goal of this world was not to recreate another section of Plants vs. Zombies exactly. The project uses PvZ as a stress test for a reusable gameplay framework, so the original world needed to demonstrate that the framework could support new content designed independently of the original game.
The world was intentionally designed on paper first, with a limited amount of time available. The goal was not to prove that every mechanic would be fun or balanced before implementation. Instead, the goal was to reach a sufficiently coherent first implementation, explicitly identifying questions that could only be answered through playtesting.
The world theme chosen was an underground basement.
An early question was whether the world should occur before or after Roof. Two possibilities were considered:
If it occurs after Roof, Basement could function as a new "Night" equivalent of Roof, introducing a new mechanic after the player has already learned the previous worlds.
If it occurs before Roof, Basement could instead become a late-world "put everything you've learned together" world, with Roof remaining the final world.
This was not fully resolved because it did not need to be resolved before the core world design could proceed.
Three initial world-mechanic concepts were considered.
The first concept was a concrete/dirt floor covering part of the lawn. Plants could not normally be planted on these tiles and would require pots to use them.
This was compared directly to Roof.
Roof already has:
pots required across the lawn
roof-specific projectile interactions
The proposed mechanic therefore felt like only a partial version of an existing Roof mechanic: essentially, "some of the lawn requires pots."
The concept was rejected.
The problem was not that it was impossible to implement, but that it did not provide enough of a distinct gameplay identity. It felt too much like taking one component of Roof and weakening it rather than creating a new world mechanic.
The second concept was a periodically changing visibility mechanic.
The idea was that the lights would randomly go out for short periods, potentially flashing on and off over several waves.
It would give Plantern a clear purpose.
It has some conceptual similarity to Fog.
It creates a recognizable basement atmosphere.
It could introduce uncertainty and tension.
The main concern was that intermittent darkness might feel more annoying than interesting.
Fog works partly because its visibility restriction is spatial and relatively predictable: the player knows which part of the board is obscured.
Randomly losing visibility over the entire board could instead interrupt normal actions:
"Oh, come on, now I can't see."
The concern was therefore not simply difficulty, but player experience and agency. A mechanic that makes the game harder is not necessarily a mechanic that creates interesting decisions.
Lights Out remained viable but was considered riskier than the other option.
The third concept was a terrain object placed on the lawn.
The initial idea was that dirt piles would:
block projectiles
potentially slow or otherwise debuff zombies
exist directly on the field rather than being something the player initially creates
The important clarification was that the dirt pile is terrain, not a plant and not a defensive object the player freely places.
The mechanic became more interesting when considering that the terrain could potentially be moved to another tile.
This transformed it from a passive obstacle into something that could potentially allow the player to manipulate the battlefield.
That introduced several design questions:
Should the object be visually a pile or a wall?
Exactly how much should it block?
Should it slow zombies?
How should the player move it?
Should the shovel move it?
Should a plant move it?
Should movement have a cooldown?
Could moving terrain become too powerful?
Could the player construct overly centralized defenses?
Would this feel too complex compared to PvZ1?
Rather than resolving these questions immediately, the decision was made to design plants and zombies around the mechanic first.
This became an important turning point in the design process.
Initially, the process was approaching the world mechanic as something that should be finalized before designing the plants and zombies.
That approach was reconsidered.
The reasoning was that PvZ's mechanics are highly interconnected. A terrain mechanic cannot necessarily be evaluated independently from the plants and zombies that interact with it.
For example, while considering how the player might move dirt piles, the existing Cob Cannon interaction was remembered.
Cob Cannon demonstrates an interaction pattern in which:
the player selects/clicks the plant,
the player selects a destination on the lawn,
the plant performs an action at that location.
This suggested that a terrain-manipulating plant could potentially use an already-familiar interaction pattern rather than requiring a completely new control system.
This led to a change in design workflow:
Instead of finalizing the world mechanic first, design the plants and zombies while allowing them to change and clarify the world mechanic.
The world mechanic would then be finalized enough for implementation after seeing how the content interacted with it.
This became one of the most important methodological decisions of the process.
The first major plant designed around the world mechanic was Tumbleweed.
The current concept is:
It can be planted on top of a dirt pile/wall or another plant.
The player can activate it and select another tile.
The Tumbleweed moves the selected plant or terrain object to that destination.
After moving the object, the Tumbleweed rolls to the right.
It hits the first zombie it encounters and deals damage.
Its damage is intentionally secondary to its terrain/plant manipulation role.
Its cooldown should be medium-high so the player cannot constantly rearrange terrain whenever necessary.
The exact damage was deliberately left unresolved. An initial thought was approximately two peas' worth of damage, but this was explicitly treated as a testing parameter.
The cooldown was also left as an implementation/testing parameter rather than being precisely determined on paper.
The important function of Tumbleweed is not its damage.
Its primary function is to make terrain manipulation an actual player decision.
Without Tumbleweed, movable dirt piles risk becoming a complicated tool with unclear controls.
With Tumbleweed, the player potentially has to think about:
where terrain should be
when terrain should be moved
which lane needs it
whether a plant should be moved instead
whether moving something now is worth spending the cooldown
how the resulting Tumbleweed attack affects the board
This creates the possibility of terrain becoming a strategic resource rather than simply an obstacle.
The exact strength of this mechanic was intentionally left for playtesting.
One possible strategy was identified:
The player could potentially move many dirt piles into one lane and place Garlic in front of them, effectively concentrating the terrain into a defensive structure.
This was not automatically considered a problem.
The reasoning was that Garlic itself has limitations, and eventually zombies would destroy it. The player would also be sacrificing defensive flexibility in the other lanes.
The conclusion was:
This is a strategy worth testing rather than something that should automatically be prevented.
This reflected a broader design principle used during the process:
A strategy that looks unusual or powerful on paper is not automatically a design flaw.
If it turns out to dominate the game, it can be adjusted after playtesting.
A second plant concept was created:
Power Plant
Plants in a 3×3 area receive a multiplier to their behaviour speed.
The initial values considered were around 1.5× or 2×, but the exact value was deliberately left for testing.
The concept fits the existing PvZ naming style and was also retained partly because the name itself felt appropriately PvZ-like.
Its important design role is to interact with the project's existing concept of plant behaviour speed rather than introducing an entirely unrelated mechanic.
The first zombie designed specifically around the movable terrain was the Shovel Zombie.
The proposed behavior was:
It deletes dirt piles.
This can disrupt a player's terrain-based strategy.
It can synergize with other zombies because removing terrain allows them to move normally again.
It could potentially be countered by Magnet-shroom removing its shovel.
It could also simply be dealt with using normal defenses.
The intended player experience was potentially panic:
"Oh shit, the thing I've built my strategy around is being removed."
However, several problems emerged.
The most obvious dedicated counter was Magnet-shroom, which already has a relevant interaction with equipment.
A plant specifically created to make more dirt would be problematic because it would undermine the premise that the dirt is a world-level terrain mechanic rather than something the player can freely create wherever they want.
There was concern that the Shovel Zombie could make players stop using Tumbleweed.
If the player sees their dirt piles being systematically destroyed, they may conclude:
"Why bother building a terrain strategy?"
rather than:
"I need to adapt my terrain strategy."
The key unresolved question was whether the zombie would create interesting panic or simply punish the player for engaging with the world's main mechanic.
This was considered a significant risk.
The Shovel Zombie was scrapped.
The important decision was to remove an interesting-sounding mechanic once its interaction with the broader system looked unhealthy, rather than keeping it merely because the original level structure needed a zombie.
A spreadsheet analysis was performed on the original PvZ level progression.
The analysis found a useful structural pattern when excluding minigame and belt levels:
Every normal level introduces a new plant reward.
New zombies alternate between introduction levels and levels without a new zombie.
Therefore, across four normal levels, the expected structure is:
Level
New Plant
New Zombie
6-1
Yes
Yes
6-2
Yes
No
6-3
Yes
Yes
6-4
Yes
No
This changed the design problem significantly.
Instead of freely inventing four plants and an arbitrary number of zombies, the world now had a clear PvZ-like progression constraint.
The original idea of using a minigame on 6-5 was retained as a possible post-release addition, but the initial world design did not depend on having one.
This was effectively treated as a design constraint rather than something to fight against.
6-1 needed to introduce:
the basement's dirt-pile mechanic
a new plant
a new zombie
The world mechanic was intended to be introduced gradually.
The player should initially be allowed to understand what dirt piles do without immediately having the ability to manipulate them.
This led to the idea that Tumbleweed should not be introduced until 6-2.
The first zombie therefore needed to be relatively understandable and not overwhelm the first introduction to the world mechanic.
An obvious basement-themed idea was an Excavator Zombie: a zombie operating an excavator.
The concept fit the visual theme strongly.
However, it was rejected for the first level.
The reasoning was not only visual.
In PvZ, large zombies communicate danger through their presentation and usually correspond to significant durability/threat.
A large excavator therefore implicitly tells the player:
"This is a major threat."
Introducing that kind of enemy at the beginning of the world felt inappropriate.
The concept could potentially fit later in the world, alongside the kinds of larger threats represented by Zomboni and Gargantuar, but it was not considered appropriate for 6-1.
This was an example of rejecting an idea that fit the theme because it did not fit the difficulty/progression role.
A Crawler Zombie was then considered.
The initial inspiration came from observing a pattern in PvZ's zombie introductions:
Newspaper Zombie changes behavior when attacked.
Ducky Tube Zombies create an accessibility/range problem.
Jack-in-the-Box Zombie can bypass normal defenses and create an explosion.
Bungee Zombie behaves differently from a conventional walking zombie.
This suggested a possible design principle:
New zombies often introduce a threat that cannot simply be solved by normal projectile damage.
The proposed Crawler would potentially:
move in a crawling posture
be difficult or impossible to hit with normal projectiles
potentially have a range-based attack
require unusual counters
The design quickly accumulated too many interacting mechanics.
Questions appeared around:
whether Puff-shroom could hit it
whether it should be fast enough to overwhelm Puff-shroom
whether it should have ranged attacks
whether it should carry a weapon
how plants would defend against it
whether it would be too strong for a first-level zombie
Eventually, the design was abandoned.
The important lesson was that an interesting theme can easily turn into a mechanically overloaded enemy when multiple "special" properties are added to make it distinctive.
The ranged component of the Crawler idea was then separated from the crawling idea.
This produced a much cleaner concept:
Slingshot Zombie
attacks plants from range using a slingshot
creates a threat to Night's common reliance on inexpensive Puff-shrooms
introduces a new kind of threat without requiring multiple unrelated special mechanics
This was considered more appropriate for 6-1.
The reasoning was particularly connected to Night's resource structure:
Sun is more constrained.
Puff-shroom is attractive because it costs nothing.
A ranged zombie creates a vulnerability in relying exclusively on cheap, fragile plants.
The zombie therefore doesn't simply say:
"Don't use Puff-shroom."
Instead, it creates a situation in which the player has to consider protecting their plants.
That was considered a cleaner first-level interaction.
The proposed response to Slingshot Zombie became Cottonball.
Its basic concept is a plant that absorbs incoming projectiles.
This creates a relationship:
Slingshot Zombie → ranged projectile threat
Cottonball → projectile protection
The design was compared conceptually to Umbrella Leaf and its relationship to Catapult Zombie.
The exact implementation and strength were not finalized.
One additional idea was considered:
If a Slingshot Zombie repeatedly fails to damage plants because of Cottonball, it could eventually stop using its ranged attack and resume normal movement/eating.
The exact number of failed attacks was deliberately left as a testing parameter.
This prevents the counter from necessarily becoming a permanent hard-disable.
The resulting 6-1 concept became:
New world mechanic: Dirt piles
New zombie: Slingshot Zombie
New plant: Cottonball
The intended learning sequence is:
The player learns how dirt piles affect the lawn.
The player encounters a zombie that threatens plants from range.
The player receives a tool that can address that threat.
The important design decision is that Cottonball does not need to manipulate dirt piles directly.
It can instead address one of the consequences created by the world.
6-2 was selected as the point where the player receives the ability to manipulate dirt piles.
The reasoning was deliberate:
If Tumbleweed and the zombie that disrupts terrain were introduced simultaneously, players might never form the intended terrain-based strategy.
Instead:
6-1: understand the terrain
6-2: receive the tool that manipulates it
This gives the player an opportunity to discover the mechanic before being asked to defend it against a specialized threat.
The new zombie slot is intentionally empty.
Because 6-3 required another new zombie, Shovel Zombie was initially placed there.
This actually fit the intended progression well:
6-2 gives the player Tumbleweed.
6-3 introduces a zombie that disrupts the terrain strategy.
However, further analysis showed that the Shovel Zombie's interaction with the system was questionable.
The decision was therefore made to remove it rather than force it into the roster simply because the level structure required a zombie.
The replacement was a Cheer Zombie.
The concept:
The Cheer Zombie follows another zombie.
It increases that zombie's behaviour speed through a multiplier effect.
This immediately interacts with the existing gameplay framework and creates a new kind of group threat.
It also creates an interesting relationship with a proposed plant:
Spikeshroom
Spikeshroom has a piercing attack:
it hits one zombie
the projectile continues
it can hit a second zombie
The result is that a zombie which increases the speed/behavior of another zombie can indirectly make piercing attacks more valuable against groups.
Unlike Shovel Zombie, Cheer Zombie does not directly invalidate the world's central mechanic.
It introduces an additional combat pressure that can interact with the new plant roster.
The proposed 6-3 plant became Spikeshroom.
Its initial mechanic:
A projectile can pierce through one zombie and hit the next.
The exact number of targets and damage were not treated as finalized balance values.
Its relationship with Cheer Zombie provides an initial reason for the plant to be useful without requiring it to be a hard counter.
This is an intentional distinction:
The plant does not need to be:
"Cheer Zombie Counter Plant."
Instead, the zombie creates conditions in which the plant's existing strength becomes more valuable.
The resulting 6-3 concept is:
New plant: Spikeshroom
New zombie: Cheer Zombie
The intended progression is approximately:
The player has already learned terrain manipulation through Tumbleweed.
A new zombie creates a different combat problem.
The player receives a plant whose properties become particularly useful against groups of affected/faster zombies.
This avoids directly attacking the player's newly learned terrain strategy.
Power Plant was placed on 6-4.
Its initial mechanic:
Plants in a 3×3 area receive a multiplier to their behaviour speed.
The exact multiplier remains a playtesting parameter.
The plant is not required to be a direct counter to a zombie introduced on the same level.
This was recognized as a useful point during the design process: the PvZ progression pattern provides structural constraints, but not every plant/zombie pair needs to form a perfectly isolated counter relationship.
Power Plant can instead serve as a new capability that modifies how the player's existing plants operate.
A world-specific final-flag mechanic was also identified.
Existing PvZ worlds have special ways of introducing additional zombies during the final wave/flag:
Night uses graves.
Pool/Fog use the pool as part of their world-specific spawning behavior.
Roof uses Bungee Zombies.
For Basement, the natural equivalent became:
Zombies emerge from dirt piles / from above the underground environment and fall onto the lawn.
The visual explanation is tied directly to the world theme:
The zombies are underground and can emerge from above, dropping onto the lawn.
This makes the dirt-pile mechanic affect not only combat and positioning, but also zombie spawning.
A plant does not necessarily need to be a dedicated counter to this ambush.
The existence of Umbrella Leaf as a possible interaction was considered, but the design does not require every world mechanic to have a dedicated counter plant.
The current proposed progression is:
Level
Plant
Zombie
Primary purpose
6-1
Cottonball
Slingshot Zombie
Introduce dirt piles + ranged threat
6-2
Tumbleweed
—
Give player terrain manipulation
6-3
Spikeshroom
Cheer Zombie
Introduce group/speed interaction
6-4
Power Plant
—
Introduce plant behaviour-speed synergy
Final flag
—
Dirt-pile ambush
Make the world mechanic part of spawning
The exact numbers, cooldowns, damage values, and several interaction details remain experimental.
The design intentionally does not attempt to answer every question on paper.
Examples include:
How strong should projectile blocking be?
Should they slow zombies, and by how much?
Should they be called piles or walls?
Exactly how should terrain movement work?
What cooldown feels appropriate?
How much damage should its rolling attack deal?
How long should its cooldown be?
What happens if the destination is occupied?
How flexible should its movement be?
Does its manipulation remain useful when Shovel Zombie does not exist?
How strong should the ranged attack be?
How durable should Cottonball be?
How many blocked attacks should cause the zombie to switch behavior?
Does the interaction create interesting protection decisions?
What speed multiplier should Cheer Zombie provide?
How much should the piercing attack contribute?
How many zombies should Spikeshroom be able to hit?
Does the combination create interesting positioning?
Should its multiplier be 1.5×, 2×, or something else?
What counts as "behaviour speed" for its purposes?
Is the 3×3 area sufficiently interesting?
How many zombies should appear?
How frequently?
Where do they land?
How much warning should the player receive?
Does Umbrella Leaf interact with it naturally?
Does concentrating dirt piles into one lane with Garlic become dominant?
Does Tumbleweed remain valuable when players discover defensive terrain strategies?
Does terrain manipulation create interesting decisions or merely increase micromanagement?
Does the world mechanic remain relevant throughout the level progression?
These were deliberately not treated as blockers because they are questions that implementation and playtesting can answer more reliably than paper design.
The basement theme was useful because it generated concepts naturally, but the process did not blindly accept ideas just because they were thematic.
The Excavator Zombie is a good example.
It fit the theme extremely well but was rejected because its implied threat level was inappropriate for the first level.
This demonstrates an important distinction:
Theme supports mechanics; theme does not dictate mechanics.
The spreadsheet analysis of PvZ's reward progression was particularly useful.
Instead of saying:
"I'll make four levels however I feel like."
the original game's structure was examined and a repeatable pattern was discovered.
That pattern became a constraint for the new world.
This is a stronger design approach than simply reproducing individual content.
The most important questions were increasingly framed around what the player would actually do.
For example:
Does Lights Out create tension or annoyance?
Does Shovel Zombie create panic or invalidate terrain strategies?
Does Tumbleweed create meaningful terrain decisions?
Does Slingshot Zombie create a vulnerability in Night's cheap-plant strategy?
Does concentrating dirt piles with Garlic represent a legitimate strategy or an exploit?
These are player-experience questions rather than implementation questions.
Several concepts were abandoned:
Concrete/dirt floor
Lights Out as the preferred mechanic
Excavator Zombie as the first zombie
Crawler Zombie
Shovel Zombie
This is important.
A weaker design process often becomes attached to ideas because they sound cool.
Here, the concepts were repeatedly tested against the broader system and removed when they created problems.
Shovel Zombie is probably the strongest example.
It was thematically appropriate and mechanically interesting, but the analysis suggested that its existence could discourage players from engaging with the world's central mechanic.
It was removed anyway.
The realization that the plants and zombies could help define the world mechanic was particularly valuable.
Tumbleweed changed the understanding of terrain manipulation.
Slingshot Zombie helped define what a useful 6-1 plant might look like.
Cheer Zombie led naturally toward Spikeshroom.
The world mechanic was therefore treated as part of a larger system rather than a specification that everything else had to obey.
The biggest weakness in the process was not game-design ability; it was efficiency and structure.
The process spent substantially longer than the original planned two hours.
That happened partly because individual ideas repeatedly expanded into many hypothetical interactions.
For example, the Crawler Zombie accumulated multiple possible mechanics before being discarded.
Similarly, Shovel Zombie generated several questions about counters and terrain creation before ultimately being removed.
This is not inherently bad—the exploration produced useful information—but a future process could converge faster by establishing a simple rule earlier:
If an idea requires several additional mechanics to explain why it is fun, temporarily remove it and continue with simpler candidates.
Another improvement would be to distinguish earlier between:
questions that require design decisions
questions that require balance
questions that require playtesting
For example, "Can Tumbleweed move a plant?" is a design question.
"Should its cooldown be 12 or 16 seconds?" is primarily a tuning question.
"Does the Garlic + concentrated-terrain strategy dominate?" is a playtesting question.
Recognizing those categories earlier would reduce unnecessary paper-design time.
The strongest part of this process is not the individual plant ideas.
Someone can invent a Tumbleweed or a Slingshot Zombie without demonstrating much design ability.
The stronger evidence is the process behind them.
The design demonstrates several useful habits:
identifying the intended player experience
comparing new mechanics against existing game conventions
using constraints rather than fighting them
considering counterplay
considering player strategy rather than only enemy behavior
recognizing when a mechanic risks invalidating another mechanic
considering onboarding and progression
separating theme from mechanical role
thinking about interaction patterns and UX
deliberately leaving balance questions for playtesting
discarding ideas when they stop serving the broader design
using existing game data to identify structural patterns
considering how individual pieces combine into a larger system
The most impressive example is the evolution of the terrain mechanic.
It began as:
"Dirt piles block projectiles and maybe slow zombies."
It developed into a system involving:
projectile interaction
terrain positioning
plant positioning
terrain manipulation
plant-based interaction
zombie behavior
zombie spawning
Garlic strategy
Tumbleweed
Slingshot Zombie
and potentially other plant/zombie interactions
That is a substantially richer result.
The paper design itself is not proof that the world is good.
The next step is therefore important: implementation and playtesting.
The strongest portfolio story is not:
"I designed a really clever Basement world."
It is:
"I designed an initial hypothesis for a new world, explicitly identified what I could and couldn't determine on paper, had the content implemented through my API, playtested it, and then evaluated which of my assumptions were actually correct."
That turns the design process into an engineering/design experiment.
It also makes your willingness to change your mind a strength rather than a weakness.
The initial design is sufficiently complete to move on.
There are unresolved questions, but they are primarily experimental questions, not missing pieces of the game's concept.
The world currently has:
a clear theme
a distinct world mechanic
a progression structure
four plant rewards
two new zombies
a world-specific final-wave mechanic
intended interactions between the new content
known risks and strategies worth testing
explicitly identified tuning questions
At this point, continuing to design on paper would mostly mean trying to predict the results of playtesting.
The more valuable next step is therefore to implement the first version, let other developers interact with the API, and see what happens when the design meets both real players and real code.
That is where the next layer of the portfolio story begins.