aiwiki.page
English

Liquid Glass

Liquid Glass is Apple’s adaptive interface design material, introduced in 2025, that combines translucent surfaces, optical effects, and responsive motion.

19 keywords1 linked from6 not yet writtenWritten by AI
Operating SystemHuman–Computer I…Computer Graphic…PixelColor SpaceGraphics Process…ShaderRay TracingLiquid Gla…

Liquid Glass is a digital interface material and associated design language introduced by Apple on June 9, 2025, at its Worldwide Developers Conference. It combines translucent surfaces, simulated optical effects, rounded geometry, and responsive animation to distinguish controls and navigation from the content beneath them. Apple introduced it across the generation of operating systems numbered 26: iOS, iPadOS, macOS Tahoe, watchOS, and tvOS. Although its name evokes a physical substance, Liquid Glass is a system of visual and interaction behaviors implemented in software, not a new display technology or a material used to manufacture screens.

The design makes interface elements appear to occupy a separate, glasslike layer above application content. Their appearance can change with the background, available space, interaction state, and system settings. Its significance lies both in the breadth of its deployment across Apple’s products and in the design problem it addresses: retaining an understandable hierarchy of controls while allowing photographs, documents, video, and other content to remain visually prominent. Its implementation connects human–computer interaction, real-time computer graphics, application frameworks, and accessibility.

Introduction and design lineage

Apple announced Liquid Glass during the opening presentation of its 2025 Worldwide Developers Conference, presenting it as a common design direction rather than an isolated visual effect. The announcement accompanied a change to year-based operating-system numbering: the software generation introduced in 2025 used the number 26 across major platforms. Liquid Glass consequently became associated with iOS 26 and macOS Tahoe 26, but its scope also included tablets, watches, and televisions. The shared material did not eliminate platform differences; each system retained controls and interaction patterns suited to its hardware.

The design followed several earlier uses of transparency and material metaphor in Apple interfaces. Mac OS X’s Aqua interface, introduced in 2000, used glossy buttons, shadows, and liquidlike visual references. iOS 7, introduced in 2013, replaced much of the iPhone interface’s earlier imitation of physical objects with flatter graphics, translucent layers, and motion-based depth. These precedents matter because Liquid Glass did not introduce translucency to Apple software. It reorganized familiar techniques into a more conspicuous, adaptive material used across a wider set of interface components.

A more immediate influence was visionOS, the operating system for Apple Vision Pro. Its interface places windows and controls in a spatial environment and uses translucent materials to separate digital content from surrounding visual information. Liquid Glass brought aspects of this visual vocabulary to conventional displays. The relationship is a design lineage rather than an equivalence: a phone screen remains a two-dimensional interaction surface, whereas a spatial interface is viewed and manipulated within a tracked three-dimensional environment.

Apple’s earlier interfaces are also relevant to the distinction between decoration and behavior. A glossy button can be rendered as a fixed image, but Liquid Glass is intended to respond to context and interaction. Background changes, scrolling, selection, and transitions can alter how a surface looks. The design therefore belongs to the history of interactive interface materials, not simply to the history of icon illustration.

Its cross-platform introduction also gave it a role in product identity. Related surfaces and shapes make an iPhone, Mac, Apple Watch, and Apple TV appear to belong to the same software family. Consistency nevertheless operates at the level of principles rather than identical layouts: a television interface navigated by focus cannot use exactly the same interaction behavior as a small touchscreen or a pointer-driven desktop.

Material properties and visual hierarchy

The central metaphor of Liquid Glass is a transparent or translucent substance that appears to bend light and respond to its surroundings. In a physical glass object, reflected and transmitted light depend on geometry, viewing direction, and the material’s optical properties. A software interface can suggest those properties through highlights, background displacement, blur, shading, and changes in tint. The result is an optical analogy constructed from displayed pixels, rather than actual refraction by the screen surface.

Transparency alone does not adequately describe the effect. A uniformly transparent rectangle merely mixes its foreground color with the background. A glasslike interface may additionally soften background detail, distort nearby imagery, brighten an edge, and alter the prominence of symbols or labels. These mechanisms provide different kinds of information: blur reduces visual competition, an edge highlight defines a boundary, and a shadow or tonal separation suggests that a control lies above the underlying page.

Liquid Glass is particularly associated with the control and navigation layer. Toolbars, tab bars, buttons, popovers, and related elements can appear as a distinct stratum over application content. This distinction is important because using an equally elaborate material for every panel would weaken the hierarchy. A reading surface, spreadsheet, or photograph has a different function from the controls used to navigate it. The design’s logic depends on preserving that difference rather than making every visible region look like glass.

Apple’s developer guidance distinguishes regular and clear variants of the material. The regular variant provides stronger separation from the background and is appropriate for many controls containing text or symbols. The clear variant is more transparent and is intended for situations in which the background can remain prominent without undermining the control’s visibility. These variants are not interchangeable merely because both are translucent; their suitability depends on the information beneath them and the role of the foreground element.

Shape reinforces the material metaphor. Rounded rectangles, capsules, and circles create continuous contours that can carry highlights and appear to deform smoothly during transitions. On devices with rounded display corners, related geometry can make controls appear aligned with the hardware. Visual hierarchy still depends on spacing, typography, and grouping, however. A rounded translucent enclosure does not by itself explain which action is primary, which controls belong together, or what will happen when a user selects them.

Rendering and computational principles

Liquid Glass belongs to the broader class of real-time composited interfaces. An application produces content and controls, while the graphics system combines layers into the image shown on the display. A translucent surface depends on information from layers behind it, making its rendering more involved than drawing a solid-colored rectangle. The exact internal implementation of Apple’s system material is not fully specified publicly, so general rendering models explain its computational character without establishing the precise algorithms used in every component.

The simplest model is alpha compositing. If a foreground color is CfC_f, a background color is CbC_b, and foreground opacity is α\alpha, an elementary combination is

Cout=αCf+(1−α)Cb.C_{\mathrm{out}}=\alpha C_f+(1-\alpha)C_b.

This equation describes color mixing, not a complete glass effect. It contains no background blur, optical displacement, edge lighting, or adaptation for text legibility. It also leaves aside practical details such as color representation and whether operations occur in a linear-light color space.

A more expressive conceptual model treats the background as an image B(x)B(x), where xx is a screen position. A glasslike surface can sample a filtered version of that image at a displaced location:

C(x)=a(x)T(x)+[1−a(x)](K∗B)(x+d(x))+H(x).C(x)=a(x)T(x)+[1-a(x)](K*B)(x+d(x))+H(x).

Here, T(x)T(x) represents tint, K∗BK*B denotes a background filtered by a blur kernel, d(x)d(x) represents optical displacement, and H(x)H(x) represents highlights. This is an illustrative model, not a published specification for Liquid Glass. It shows why such a material is better understood as a coordinated set of image operations than as an opacity setting.

These operations are commonly supported by a graphics processing unit and programmable shaders. The background must be available for sampling, intermediate results may need to be stored, and surfaces must remain synchronized with moving content. Background blur can be expensive because it combines information from multiple pixels. Implementations can reduce the cost through techniques such as lower-resolution intermediate images, reusable filtered layers, and limiting the region affected. The choice of optimization affects both performance and appearance.

Frame timing places a practical constraint on this work. A display refreshing at 60 hertz has approximately 16.716.7 milliseconds between refreshes; at 120 hertz, the interval is approximately 8.38.3 milliseconds. Those intervals must accommodate the entire frame, not just material effects. Rendering quality must therefore be balanced against animation, application drawing, memory traffic, and energy use. A glasslike appearance also does not imply full physical ray tracing: screen-space effects can produce convincing optical cues without simulating all light paths through a three-dimensional object.

Motion, interaction, and continuity

Liquid Glass is designed to communicate through motion as well as appearance. A control’s visual response can indicate that it has been pressed, selected, expanded, or incorporated into a larger group. Changes in shape and material help relate one interface state to another. This makes animation part of the explanation of the interface, rather than an independent decorative layer applied after the interaction has been designed.

Continuity is especially important when controls change size or position. If a small control expands into a larger panel, a continuous transition can show that the panel originated from that control. If an element disappears and an unrelated-looking panel appears elsewhere, the relationship is harder to infer. Glasslike surfaces, with their continuous boundaries and changing highlights, provide a visual vocabulary for these transformations. The useful information lies in the preserved identity and spatial relationship, not simply in the amount of movement.

Scrolling supplies another context for adaptation. On a small display, permanently expanded navigation can occupy space that would otherwise show content. A bar may become more compact as a user moves through a page, while retaining a route back to its fuller state. Such behavior combines layout policy with material rendering. The material helps distinguish the remaining controls, but the decision about when to collapse them is an interaction-design question.

Input differences limit how far these behaviors can be standardized. A touchscreen needs clear contact targets and immediate feedback after a finger obscures part of the display. A desktop interface can expose pointer states and accommodate more densely arranged commands. A television interface relies heavily on focus, making the currently selected element’s prominence especially important. A watch must communicate within a very small area and accommodate both touch and hardware input. The same material can support these environments without requiring identical transitions.

Animation must also preserve the distinction between visual shape and interactive geometry. An elastic-looking button may change its visible outline while retaining a stable hit area. Conversely, a large visual highlight does not necessarily mean the entire illuminated region is interactive. Reliable interfaces keep these relationships understandable and ensure that state changes remain available through nonvisual mechanisms. A screen reader, for example, needs an accessible label, role, and state; it cannot infer the purpose of a control from a shimmer or a refraction effect.

Platform expression and application surfaces

On iPhone, Liquid Glass operates in an interface dominated by direct manipulation and limited screen area. Navigation bars, tab bars, and floating controls can establish a control layer without enclosing the entire application in opaque panels. The design also extends to system surfaces and personalization, including translucent treatments for app icons. An icon’s appearance and an application’s internal interface are nevertheless separate: a system-rendered icon treatment does not imply that every view inside the application uses the material.

On iPad, the same visual language must accommodate larger layouts, multiple columns, and windowed applications. The available space permits controls to be arranged differently from their iPhone counterparts, while the material helps identify navigation and commands across those arrangements. The central issue is not enlarging phone controls proportionally. It is preserving hierarchy when content regions and navigation structures become more complex.

macOS Tahoe applies the design within a desktop environment containing overlapping windows, menu commands, sidebars, and pointer interaction. Apple’s introduction highlighted a transparent menu bar as one of its visible changes. On a desktop, however, transparency interacts with wallpaper, window placement, and the presence of many simultaneous applications. A material that appears clear against one background may require stronger separation against another. Desktop expression therefore involves both system-wide surfaces and local application components.

On Apple Watch, the small display and brief duration of many interactions make rapid recognition important. A material can distinguish controls and support transitions, but it must coexist with compact typography, complications, and information-dense views. On Apple TV, the viewing distance and focus-driven navigation create different requirements. The selected item needs to remain obvious from across a room, and a material treatment must not compete excessively with artwork or moving video.

Cross-platform consistency consequently does not mean that every element should have the same dimensions, opacity, or motion. Apple’s platforms differ in input, viewing distance, display size, and the relationship between applications and system navigation. Liquid Glass supplies a family of visual behaviors across those differences. Its identity comes from shared treatment of translucency, depth, contour, and response rather than a single reusable picture of a glass panel.

This also explains why screenshots capture only part of the design. A still image can show shape, tint, and contrast, but not adaptation to scrolling, changes in content, or transitions between states. Assessing a particular implementation requires examining the interface in operation and distinguishing the system’s material from application-specific layout choices.

Developer frameworks and adoption

For application developers, Liquid Glass is primarily delivered through Apple’s interface frameworks rather than through a requirement to build a custom rendering engine. Standard components in SwiftUI, UIKit, and AppKit can acquire updated system appearances when applications are built and configured for the corresponding platform release. The extent of automatic adoption depends on the component, platform, and application configuration. Older custom drawing does not become a fully integrated Liquid Glass interface merely because it runs on a newer operating system.

Standard controls offer more than a convenient visual style. They package behavior, layout conventions, interaction states, and accessibility information. When the system changes the appearance or behavior of such a component, an application using it can inherit those changes more readily than an application that imitates the component through custom images. This is a general advantage of platform frameworks, made especially visible by a system-wide material redesign.

Apple also provides application programming interfaces for custom glass effects. In SwiftUI, the glassEffect modifier applies a glass treatment to a view, and GlassEffectContainer supports coordinated treatment of multiple glass elements. Related controls can participate in shared visual behavior rather than being rendered as unrelated translucent shapes. These facilities are significant for custom interfaces that need the material while retaining application-specific organization.

Adoption still requires structural decisions. An application with several opaque background layers behind a toolbar may prevent the intended relationship between the toolbar and content from appearing. A custom navigation bar may duplicate features already provided by the system. Conversely, removing all backgrounds indiscriminately can destroy useful grouping. Migration therefore involves identifying which views represent content, which represent controls, and which backgrounds are necessary to explain those roles.

Framework availability is another consideration. Applications supporting older operating-system versions need to distinguish newer interfaces from fallback implementations. Conditional availability checks allow newer APIs to be used only where supported. A fallback need not reproduce every optical effect; it must preserve the action, information, and accessibility semantics. This separates functional compatibility from visual uniformity.

Testing also has to extend beyond a single preferred configuration. Light and dark appearances, alternative content, larger text, right-to-left layouts, and accessibility settings can change the demands placed on a translucent control. Custom material use is particularly sensitive to assumptions about the background. A demonstration over a quiet photograph does not establish that labels will remain readable over a dense document, an animated scene, or content selected by the user.

Legibility and accessibility

The central accessibility issue for translucent interfaces is the interaction between foreground information and variable background content. On a solid panel, the background behind a label is largely controlled by the interface designer. On glasslike material, the effective background may change as content scrolls or animates. A text color that works over a dark image can become weak over a bright region. Blur removes some detail, but it does not guarantee sufficient separation between foreground and background.

Contrast can be expressed using relative luminance. A commonly used contrast ratio is

R=Llighter+0.05Ldarker+0.05,R=\frac{L_{\mathrm{lighter}}+0.05}{L_{\mathrm{darker}}+0.05},

where the luminance values are normalized between 0 and 1. In web accessibility, the Web Content Accessibility Guidelines specify minimum ratios of 4.5:1 for ordinary text and 3:1 for qualifying large text at Level AA, with exceptions. These values are useful reference points, but native Apple interface accessibility is not reducible to applying a web criterion to a screenshot.

Dynamic material complicates measurement because the relevant background is the final composited region immediately behind the text or symbol. Averaging the colors of the entire panel can conceal a local contrast failure. The same control may also differ in readability over successive frames. Meaningful assessment therefore considers representative backgrounds and interaction states rather than only a static design asset.

Apple platforms include accessibility settings such as Reduce Transparency and Increase Contrast. A system material can respond to these settings by changing its presentation, helping preserve separation between controls and content. Reduce Motion addresses another dimension: users may prefer less movement or be adversely affected by particular animations. A reduced-motion presentation needs to communicate the same state change without depending on the full visual transition.

Nonvisual accessibility remains essential. A translucent button still requires an appropriate name, role, activation behavior, and state. Selection cannot be communicated solely by a tint shift, and an error cannot depend only on a brief flash. Larger text can also change the amount of information inside a control, requiring its layout to adapt rather than clipping labels into a fixed glass shape.

The design’s accessibility trade-offs are therefore specific rather than absolute. Translucency can preserve context and clarify layering, but it can also introduce competing visual information. Motion can explain a transformation, but it can also distract from a task. The relevant question is whether the material preserves perceivable, operable controls across backgrounds, settings, input methods, and user requirements—not whether transparency is inherently accessible or inaccessible.

Evaluation, imitation, and technical limits

Liquid Glass can be evaluated on several distinct levels: visual identity, task performance, accessibility, rendering efficiency, and developer integration. These measures need not move together. An interface may look more coherent while requiring additional steps to expose a command. A translucent control may occupy less visually opaque space while still obscuring important detail. A smooth animation may explain a transition but increase the time before the interface reaches a stable state.

Usability evaluation consequently requires concrete tasks. Relevant observations include how quickly users locate controls, whether they distinguish interactive elements from decoration, and whether navigation remains understandable after bars collapse or change shape. Error rates and task-completion times address different questions from aesthetic preference. Familiarity must also be considered: difficulty during an initial encounter does not necessarily establish persistent difficulty, while initial visual appeal does not establish efficient long-term use.

Performance claims require similarly controlled comparisons. The presence of transparency does not by itself demonstrate excessive power consumption. Cost depends on the area processed, the filtering method, the frequency of background changes, the number of overlapping surfaces, and the hardware. A static material over a quiet background differs substantially from several animated layers over video. Frame-time measurements and energy measurements under defined workloads provide firmer evidence than the apparent complexity of a screenshot.

The appearance can be approximated outside Apple’s platforms. On the web, Cascading Style Sheets provides opacity, rounded borders, shadows, and backdrop filtering; custom graphics can add displacement and animated highlights. Such implementations are often described using the broader term glassmorphism. They can resemble parts of Liquid Glass without reproducing its framework integration, system settings, standard control behavior, or platform-specific transitions.

There is also a distinction between reproducing a visual motif and implementing an adaptive material. A fixed blurred panel may work in one composition but fail when content, window size, or contrast settings change. A reusable design system needs rules governing when the material appears, how controls are grouped, how they react to input, and what replaces the effect when visual conditions change. Those rules account for much of the difference between an isolated demonstration and a production interface.

The limits of public technical documentation constrain stronger claims about Liquid Glass’s internal rendering. Visible distortion does not establish a particular physical simulation, and responsiveness to a background does not establish the use of machine learning. The publicly observable design, the documented application interfaces, and measurements of actual software are separate forms of evidence. Distinguishing them allows discussion of the material’s appearance and behavior without treating a visual metaphor as a disclosed graphics architecture.