Kaze and the Wild Masks

There’s a specific kind of game that exists because a small studio loved something so much they decided to rebuild it from memory. Kaze and the Wild Masks is that game. Developed by the Brazilian studio PixelHive and published by SOEDESCO on March 26, 2021, it is a 2D side-scrolling action platformer that doesn’t gesture toward the 16-bit era so much as attempt to file paperwork as a resident of it.

The setup is boilerplate on purpose. Kaze, a rabbit girl, is stranded on the Crystal Islands when a cursed artifact turns the local vegetables into an army of angry produce and takes her friend Hogo with it. That’s the whole premise. It exists to justify roughly 30 main stages and more than 50 bonus levels, and PixelHive clearly understands that nobody is here for the narrative architecture.

What they are here for is pedigree. The reference point isn’t subtle — this is Donkey Kong Country, rebuilt with modern animation tools and a rabbit. You run, you spin-attack, you bounce off enemy heads to chain height, you climb into vehicles, you thread pixel-tight jumps over instant-death pits. The Wild Masks are the one genuinely original system: four guardian masks that temporarily rewrite Kaze’s moveset into a pouncing tiger, a gliding eagle, a wall-dashing lizard, and a free-swimming shark.

Critically, it landed where a well-made homage lands: an 82 on Metacritic. That’s a strong showing for a debut from a studio nobody had heard of. It also tells you the ceiling: critics reserve their top marks for games that invent something, and this one deliberately doesn’t.

• The Wild Masks system: Four transformation states — tiger (pounce and heavy attack), eagle (glide and hover), lizard (wall dash and vertical traversal), and shark (unrestricted swimming). They’re temporary, level-gated pickups rather than a permanent progression tree, which makes them level-design tools first and player abilities second.
• Frame-by-frame pixel animation: Not tile-shuffled sprite work. Kaze’s run cycle, landing recovery, and mask transitions are hand-drawn at a frame count most indie platformers can’t afford, and the readability benefit is real.
• Layered collectible economy: Letter collectibles, crystals, and hidden secret exits in every stage, gating the bonus levels and a true ending behind genuine mastery rather than a completion percentage counter.
• Vehicle and gimmick stages: Ride sequences and set-piece levels that break the run-and-jump rhythm at roughly the cadence a 16-bit designer would have used.
• Pattern-based bosses: Each world caps with a memorization fight — learn the tells, punish the window, repeat with escalating phases.

The Gameplay Loop

The core loop is tight, in the engineering sense rather than the marketing one. Kaze has a short list of verbs — run, jump, spin, bounce — and the spin-attack doubles as a horizontal air-extender, which is the single most important mechanic in the game. Every advanced traversal sequence is built on chaining a spin into a bounce into another spin. Once that clicks, the movement gains a rhythm expressive enough to carry stages at speed.

The masks are where the design gets interesting, and also where the ambition tops out. Each mask is a temporary moveset swap, not a persistent unlock, which means PixelHive can build a level around exactly one mask’s physics and tune every gap and hazard to it. The eagle sections become hover-management puzzles about altitude and float duration. The lizard sections turn walls into floors. The shark sections drop the gravity constraint entirely and open the level’s vertical axis.

That’s efficient level design. It also means the masks never really interact. You don’t combine them, chain between them, or make decisions about which to carry into a section. A mask arrives, the level asks the one question that mask can answer, and then it’s gone. Compare that to how Donkey Kong Country 2 used animal buddies as a risk — the tension of whether to cash one in — and you can feel the missing layer. The masks are excellent set dressing on top of a conventional platformer. They are not a system.

Difficulty and Onboarding Friction

World One is a gentle handshake. Everything after it is not.

That’s a legitimate structural complaint rather than a skill-issue dismissal. The difficulty ramp is not a slope; it’s a step. Stages that teach bounce-chaining in a safe sandbox start demanding frame-accurate execution over instant-fail hazards, and the collectible hunt makes it worse. Finding the letters and crystals often requires deliberately taking the harder route through a section already at the edge of survivable, which converts optional content into a difficulty multiplier rather than a reward.

The true ending sits behind that. Locking the narrative payoff behind the hardest content in the game is a defensible old-school choice, but it does mean a meaningful slice of players will finish this game without seeing the actual conclusion — and won’t be told that’s what happened.

There’s a second, smaller problem baked into the design: hitboxes. In a game where a single contact frame resets a two-minute stage attempt, the gap between what the animation shows and what the collision registers becomes the difference between “hard but fair” and “I didn’t touch that.” The controls are built to be responsive — that’s not the issue. The issue is the potential for occasional disagreement between sprite and hurtbox, and precision platformers live or die on that contract.

Pacing and Length

Runtime is modest. A straight critical path through the main stages is a weekend; the completionist route is where the hours live, and PixelHive has clearly built the game assuming you’ll take it. That’s an honest structure, and the 50+ bonus levels are real content rather than remixed geometry. But if the difficulty spike turns you away, the value proposition contracts sharply, because the back half of what you paid for is behind the wall.

The art is the strongest thing here, and it isn’t close.

Frame-by-frame animation is expensive, and PixelHive spent the money in the right places. Kaze’s animation has weight — a landing settles, a spin has wind-up and recovery, a mask transformation reads as a distinct state change even in peripheral vision. That last point is functional, not decorative. In a precision platformer, animation is a UI, and being able to identify the current moveset from a glance at a moving sprite reduces the cognitive load of the hard sections considerably.

The Crystal Islands backgrounds go heavy on parallax depth and saturated color, which is faithful to the reference material and occasionally works against the game. A few of the busier stages let foliage and background detail compete with foreground hazards for attention. It’s not a systemic readability failure, but on some of the later high-speed sections hazards are easy to lose against the scenery.

Performance is a non-story, which is the correct outcome for a 2D sprite game. Nothing in the technical profile suggests this engine is being asked to do anything difficult — a 2D game of this scope is light on hardware, and short load times are what the death-retry loop, the loop that actually matters in this genre, needs.

The platform question here is entirely a d-pad question, and it splits the two versions cleanly.

On Nintendo Switch, this is the more natural home. It’s a pick-up-and-play structure with short, self-contained stages, which is handheld-native design whether or not PixelHive intended it. The catch is the hardware: the Joy-Con’s segmented directional buttons are a poor input device for a game that asks for clean cardinal presses under pressure, and the lizard’s wall-dash and the shark’s free-swimming sections are the ones that punish imprecise input. A Pro Controller isn’t optional here so much as strongly advised. Docked, the pixel art scales cleanly to a TV; handheld, the density flatters the animation work.

On Steam, keyboard support is technically there and practically beside the point — this is a controller game, and WASD is a self-inflicted handicap on the tighter stages. Where PC pulls ahead is input freedom: bring whatever d-pad you trust, and the game’s precision demands stop being an argument with your hardware. PC also gets the usual quality-of-life wins — instant alt-tab, cloud saves, and a short-burst stage structure that suits portable PC play well.

Content is identical across both. This is a hardware-ergonomics decision, not a feature-parity one.

Kaze and the Wild Masks is a good game that has capped its own ceiling on purpose. Most of the craft decisions — the animation budget, the movement tuning, the boss patterns, the collectible layering — land at a level most debut studios don’t reach. PixelHive knows exactly what made those Rare platformers work, down to the frame data, and rebuilt it faithfully.

Faithfully is the operative word. This is a forgery executed with the skill of an original, and forgeries have a value ceiling no amount of technique can raise. The Wild Masks were the one place where the game could have argued for its own existence, and they’re deployed as level-design keys instead of a mechanic with consequences. Ask what this game does that Donkey Kong Country Returns doesn’t, and the honest answer is “it’s Brazilian, it’s cheaper, and it has a rabbit.”

That’s not a dismissal. There is a real, underserved appetite for competent 16-bit-style platformers, and this fills it better than most in its price bracket. We recommend it to that audience specifically: go in expecting a solid cover version rather than a new song, bring a decent d-pad, and brace for the wall that arrives at the end of World One.