Planet Racer v2 — reimagined for iOS

Planet Racer began as a Flash racing game with its own artwork, audio, physics, economy, and progression. Rebuilding it for iPhone meant understanding how those pieces worked together, starting from the original SWF file.

The challenge was to preserve the handling and visual identity that made the game recognizable while adapting its controls and interface to a different device.

My work spans source and asset investigation, the Swift and SpriteKit implementation, touch controls, interface design, app lifecycle handling, persistence, and validation tools.

The original game design, artwork, and audio provide the reference. My contribution is the native adaptation and the work to understand, adapt, and verify it.

The playable build connects menus, the garage, travel, opponent selection, racing, results, and saved progression. Vehicle behavior, terrain generation, opponents, damage, and race rewards follow the original game’s rules.

Comparisons with the original runtime, simulator journeys, and selected device checks document the adaptation. The captures below show native development builds; the hero image is a separate visual concept.

The vehicles, planets, typography, yellow-and-black controls, and audiovisual identity remain the starting point. The interface work focuses on what needs to change for iPhone: clearer hierarchy, grouped actions, readable equipment states, and touch controls.

The garage comparison uses the original Flash game running in Ruffle. The other before images show an early iOS port; the after images show the reworked interface in the iOS Simulator. Game states and devices vary, so balances, equipment, and selections differ between captures.

The Modern direction grew from vehicle studies and lunar references: recognizable silhouettes, broad matte shading, dark glass, metallic suspension, and a quieter blue and lavender environment. The aim is illustrated depth while keeping the original driving model.

That direction had to work as game artwork. Bodies, tires, hubs, suspension, and scenery needed separate layers, with the original exports retained as references for shape and placement. AI-generated revisions became sprite candidates; earlier masters and prompts were preserved.

A preliminary Modern Preview now runs in the native app. Full menus, the race display, and artwork for all planets and equipment remain in development. The interface studies below show proposed next steps.

Early studies used strong glossy edges. Later revisions moved toward broad matte shading and a small white reflection near the front of the glass, echoing the original designs. Race and King provided a reference for Casual and Sports without making their silhouettes identical.

Generated edits could shift a body’s shape or transparent bounds, so each candidate also needed to fit the existing pivots, wheel centers, layers, and suspension. An appealing image still had to work inside the renderer.

Color needed similar care. Separate paint and stripe regions allow recoloring while preserving metal, glass, and reflections. King keeps its fixed yellow-and-gold livery; shared geometry avoids creating a different body for every color.

The interface study extends the visual direction with matte midnight-blue surfaces, warm yellow actions, rounded controls, larger vehicle artwork, and quieter supporting information. Home centers the vehicle and Continue action; Garage becomes a workshop; the race view groups driving actions within thumb reach.

These are interactive browser studies. Balances, wear, and race values are illustrative, and vehicle selectors preview artwork rather than change owned equipment. The proposed pause flow keeps controls accessible and asks for confirmation before leaving a race.

The design must also preserve the existing control choices, touch targets, cancellation, and input resets. Modern and Nostalgia already share saved preferences; changing the visual style should preserve those choices.

Small numerical differences changed the handling

The Swift simulation could look correct while gradually drifting from the reference during a longer replay. Small differences in math functions and arithmetic grouping accumulated through the spring and lean calculations.

Recording the first divergent update made it possible to compare individual operations with the reference Ruffle runtime. Matching its math behavior, friction grouping, and multiplication-based squaring corrected documented drift. Continuous replay then checked the integrated simulation. Some strict timer comparisons still differ and remain recorded for review.

A wider view needed a longer terrain history

The wider landscape view exposed missing ground when the vehicle reversed: the simulation had already discarded segments outside the original viewport. The renderer now retains terrain that was actually generated until it leaves the wider view. Collision and progression rules stay unchanged, and additional sky decoration uses no gameplay random values.

Original audio needed its original timing

Copying the MP3 files was not enough. Native playback durations differed from the sound windows declared in the SWF, affecting engine intros, loops, and boost cues. A preparation tool verifies the original payloads and selects the source-defined sample windows, preserving the MP3 masters alongside WAV playback assets. Duration and callback checks cover part of the behavior; audible transitions still need device listening.

Shared layout rules made the interface easier to adapt

Early screens positioned elements with independent offsets, while some text and borders remained baked into artwork. Common safe-area frames and grouped controls gave panels, labels, and actions a shared layout. Original artwork retained the game’s identity; native layout handled the elements that needed to respond to the screen. Simulator and geometry checks support the changes, with physical reach and multi-touch comfort still to be assessed.

Faithfulness required judgment

An instrumented Ruffle run exposed a non-finite value that made an arriving opponent invisible. The native adaptation kept its working arrival and documented the difference, since one runtime’s failure does not establish how every historical Flash version behaved.

The priority is recognizable handling, original audiovisual identity, correct progression, and a complete playable journey. Smaller timing and presentation differences remain visible in the backlog; lost progress, incorrect rewards, broken controls, and freezes remain defects.

The playable MVP connects the garage, travel, racing, results, and saved progression. Equipment trading, local challenge scores, audio settings, and selectable controls are implemented. Optional Modern Preview adds matte vehicles and lunar scenery using the same gameplay and saves.

The project remains in development. Full Modern menus, the race display, and artwork coverage are unfinished; the interface studies are proposals. Complete content coverage, sustained device performance, distribution, and multiplayer remain separate work.

Preserving a game means understanding how its rules, controls, presentation, and saved progress work together. Source comparisons explain behavior; simulator journeys check the flow; physical-device testing establishes how it feels to play. Each contributes a different kind of evidence.

The work also informed a proposed Flash-to-iOS workflow. Its extraction and validation tools have been used on this game; applying them to other games remains untested.