dissolve

Project Overview

dissolve is a sleep induction tool built around one idea: the hypnagogic state. Hypnagogia is the threshold between waking and sleep - where thoughts loosen, imagery drifts in, and the mind begins to let go. Most sleep apps try to distract you into unconsciousness. dissolve tries to take you there directly, through a five-minute sequence of visual and audio stimulation designed to guide the brain toward that edge.

The session is structured, intentional, and has no settings to configure. You open the app, press begin, and follow it.

Session Structure

• Minutes 0-2 (visual phase) - A full-screen particle field of slow, morphing light blobs. Colors drift across a cool-to-warm spectrum. Shapes breathe and dissolve. The display stays on; the user watches until their eyes grow heavy.
• Handoff - A brief crossfade to a dark screen with a single instruction: put the phone down. Close your eyes.
• Minutes 2-5 (audio-only phase) - The screen goes dark. Ambient music continues, gradually gaining subtle distortion and reverb - designed to follow the listener into the hypnagogic and go quiet once they're under.
• Complete - Audio fades to silence. Session ends.

Visual Design

The particle system is rendered via a custom UIView subclass driven by CADisplayLink for smooth, frame-rate-locked animation outside of SwiftUI's rendering pipeline. Each blob is an ellipse with dynamic corner radius morphing, three-layer glow rendering using additive (.screen) blend mode, and independent phase clocks for scale, rotation, and color drift. The result is a field of luminous, organic light that avoids any mechanical or digital quality.

• 12 concurrent blobs - each with independent scale, speed, rotation, and color phase
• Additive blending - overlapping blobs produce luminous, natural color mixing
• Cool-to-warm color gradient - cyan → blue → violet → pink → amber, with each blob drifting slowly through the spectrum
• Intensity-driven animation - blob speed and opacity wind down over the two-minute visual window, easing the viewer toward stillness

Audio Design

Audio is handled via AVAudioEngine with a custom signal chain, allowing real-time parameter manipulation during playback. Ambient music runs through a distortion node and a reverb node, both gradually increasing in wetness during the audio-only phase - creating a subtle sense of the sound becoming less anchored to physical space.

• Looping ambient music - scheduled and looped via AVAudioPlayerNode
• Progressive distortion - AVAudioUnitDistortion ramps from 0-18% wetness over the audio-only window
• Progressive reverb - AVAudioUnitReverb (large chamber) ramps from 0-40% wetness
• Scheduled yawns - a soft yawn sound fires at four points in the session (1:00, 2:07, 3:15, 4:30), a known hypnagogic induction technique leveraging the body's contagious yawn response
• Volume fade - master volume fades to silence in the final 30 seconds

Technical Implementation

• Platform: Native iOS (portrait)
• Framework: SwiftUI with UIKit interop for the particle canvas
• Animation: CADisplayLink-driven rendering at display frame rate
• Audio: AVAudioEngine with real-time distortion and reverb nodes
• State management: @MainActor ObservableObject ViewModel with phase-driven session state machine
• Session timing: Combine timer publisher at 50ms intervals driving phase transitions and audio parameter interpolation
• Persistence: AppStorage for onboarding state
• Screen management: Idle timer disabled during visual phase only; status bar hidden throughout