Accessibility and interaction
Semantics and reduced motion
Semantics(role=None, label=None, description=None, reading_order=None, hidden=False)An ARIA-style role such as "graphics-symbol".
The short accessible name.
A longer description.
Rank for linear reading; lower comes first, unranked objects follow in scene order.
Mark a purely decorative object.
Per-node accessibility metadata: role, label, description, reading order and a flag for decorative objects.
AccessibilityInfo(
title=None,
description=None,
alt_text=None,
long_description=None,
lang='en',
reduced_motion=...,
)The title of the animation.
A short description.
The short alternative for GIF and images.
The full text alternative.
Language tag of the texts.
What to show when the viewer asks for reduced motion.
Animation-level title, description, alternative texts, language and the reduced-motion policy.
ReducedMotion(mode='final', time=None)"final", "first" or "time".
Seconds into the animation; only valid with mode="time".
Chooses the representative still state that replaces autoplay when the viewer asks for reduced motion: the final state, the first state, or one explicit time.
ReducedMotion follows the web platform's prefers-reduced-motion preference {W3C} (2026). It is an authored alternative, not an automatic reduction in frame rate. Playback is still available when the viewer asks for it.
from kinetickit import AccessibilityInfo, ReducedMotion, Semantics
semantics = Semantics(role="graphics-symbol", label="Start", reading_order=1)
info = AccessibilityInfo(
title="Reaching the goal",
description="A dot travels to a target",
long_description="A dot starts at the lower left and moves to the upper right.",
reduced_motion=ReducedMotion("time", 1.0),
)
SVG and HTML keep titles, descriptions, stable object ids and per-node semantics. HTML adds keyboard-operable controls and an ordered object list. A GIF cannot carry these runtime semantics, so asking for accessibility output on a GIF needs a text description and can return a representative PNG.
Interaction
An interaction is data: it names an event on a node and the actions to run. KineticKit validates every binding against the node ids and the markers of the scene before it writes anything.
Event(*values)A portable trigger: HOVER_ENTER, HOVER_LEAVE, CLICK, DBLCLICK, FOCUS, SELECT, DESELECT, CAMERA_CHANGE or PLAYBACK_END.
Binding(event, actions, target=None, required=False)Connects one event to data-only actions, and says whether the interaction is required.
InteractionSpec(bindings=(), ranges=...)The validated collection of bindings, plus the named time ranges that PlayRange can play.
from kinetickit.interaction import Binding, Event, InteractionSpec, PlayRange, Select, SetProperty
interactions = InteractionSpec(
(
Binding(Event.CLICK, (PlayRange("run"),), "a"),
Binding(Event.CLICK, (Select(mode="toggle"), SetProperty(None, "opacity", 0.6)), "b"),
),
{"run": (0.0, 2.0)},
)
Action that plays a named time range.
Select(target=None, mode='toggle')Node id, or None for the node that was hit.
"select", "deselect" or "toggle".
Action that selects, deselects or toggles a node.
SetProperty(target=None, property='opacity', value=1.0)Node id, or None for the node that was hit.
Property name.
The new value.
Action that sets a property of a node, such as its opacity.
The actions are Play, Pause, Seek, PlayRange, Select, SetProperty, Toggle, SetSpeed, Reverse and ResetCamera. A binding with required=True fails before output when the chosen format cannot run it. An optional binding is dropped, and the manifest lists it under degraded. SVG and GIF drop interactions by policy; HTML runs them.
The interactive example combines both halves on a two-point scene.