Crate alpha2_model

Expand description

Versioned, host-independent project model for alpha2.

This crate describes intent only. It contains no decoder handles, GPU resources, filesystem paths, or application-specific types.

Structs§

AudioEffectLoad: The most audio effect state a project holds at once.
AudioProperties: Per-clip audio presentation controls.
AudioStreamShape: An audio stream’s channel count and sample rate.
CaptionCue: Text shown over one timeline interval.
Captions: What a caption track shows: its cues, in one style.
Clip: A reference to a source over independent timeline and source ranges.
ClipComposition: A clip’s use of a Composition, and how its time maps onto it.
ClipEffect: One versioned runtime effect invocation: a composition’s effect or transition node, or a timeline transition.
ClipId: A stable timeline-clip identifier.
ClipTransition: A two-input transition over the overlap of two visual clips.
ColorPipeline: How sources’ colors come into the working space.
Composition: An acyclic graph of image operations with one selected output.
CompositionControl: A node setting a composition exposes to the group nodes that use it: an effect or transition parameter of one of its nodes.
CompositionId: A stable composition-graph identifier.
CompositionNode: One image operation in a Composition.
CompositionReads: What each composition’s output reads: the group inputs and clip nodes a walk from its output reaches, along the inputs drawing each node reads, into the compositions its group nodes draw. A group draws what its nodes would inline, so a group node needs only the ports its composition reads: a port it never reads may be left unconnected, and what a connected one reads is not drawn.
CubicBezier: A CSS cubic-bezier(x1, y1, x2, y2) timing curve.
Ducking: Sidechain ducking authored on the track that is attenuated.
EffectResource: Host-resolved, non-media data consumed by one or more visual effects.
EffectResourceId: A stable effect-resource identifier.
EqBand: A single equalizer section stored independently from its DSP implementation.
GradientStop: One color of a gradient.
ImportedCaptions: Cues read from a subtitle file.
Keyframe: One timestamped control value.
Limiter: Final peak limiter placed at the end of the master bus.
MasterAudio: Project-wide master bus settings.
Matte: Another visual clip whose picture sets where this clip is visible.
MediaFingerprint: Evidence of a source’s media, each part recorded only when it was known.
MediaSource: A host-resolved media resource.
MergeLayer: One layer of a NodeOp::Merge.
NodeId: A stable identifier of a node within one composition graph. It has no /, which joins the IDs of the group and media nodes a step was drawn through.
Project: A complete serializable project document.
ProjectId: A stable project identifier.
ProjectSettings: Output-wide settings independent of a renderer or encoder.
Rgb: An opaque color as a color picker shows it: straight, sRGB-encoded components in the inclusive unit interval.
Rgba: A straight (not premultiplied), sRGB-encoded color with alpha, each component in 0..=1.
ShapeMask: A shape that limits where a clip is visible.
StreamTopology: The streams a media file carries.
Stroke: A shape’s outline paint. The stroke is centred on the outline.
TextBackground: A box behind each line of text.
TextOutline: A band around the glyphs, drawn behind them.
TextStyle: How text looks and where it lies: a caption track’s cues, or a text crate::Generator.
TimeScale: A positive rational rate of one time against another.
Timeline: Timeline tracks ordered from back to front for visual tracks.
Track: A homogeneous visual or audio timeline lane.
TrackAudio: Per-track audio bus controls.
TrackId: A stable timeline-track identifier.
Vec2: A two-dimensional value in normalized output coordinates.
VideoStreamShape: A video stream’s displayed size and frame rate.
VisualProperties: Resolution-independent visual presentation controls for a clip.

Enums§

Animated: A value that is either constant or controlled by sorted keyframes.
AudioEffect: A serializable single-input audio effect applied to a clip or track bus.
BlendSpace: The color space in which blend modes are evaluated.
CacheIntent: Whether a runtime may keep and reuse a clip’s rendered picture.
CaptionImportError: Why a file is not a subtitle file at all.
ClipBlendMode: Artistic blend operation stored independently from a renderer implementation.
Easing: A timing curve from linear progress to eased progress, without a hold.
EffectResourceKind: Portable effect-resource categories supported by the render ABI.
EffectValue: Serializable value supplied to one runtime effect parameter.
EqBandKind: The filter shape of one equalizer band.
FadeShape: Curve applied by a clip’s fade in and fade out.
GeneratedShape: The outline of a Generator::Shape.
Generator: The picture of a generated source.
HdrSources: What a project does with HDR sources.
Interpolation: Interpolation from one keyframe to the next.
LinearPrimaries: Named RGB primaries linear-light media may be in, each with its own white point.
LintEdge: The end of a range a LintIssue is about.
LintIssue: Something in a project that is valid but probably not meant.
LintOwner: What a LintIssue is about.
LoopMode: How each pass of a TimeMap::Loop follows the one before.
MaskMode: How a mask combines with the coverage of the masks before it.
MaskShape: The outline of a ShapeMask.
MatteMode: Which property of the matte clip’s picture is coverage.
MediaKind: The payload domain of a source.
MissingFrames: How an image sequence shows a frame missing from its numbering.
ModelError: A document invariant violation.
NodeInput: One named input of a node, as a node’s inputs are listed.
NodeOp: The operation of a CompositionNode.
Paint: A color or gradient.
ProjectLoadError: Failure while decoding a serialized project document.
ProviderSource: A source whose media an extension provider supplies, rather than the host’s video and image decoders.
Spread: What a linear or radial gradient shows outside its first and last stops.
TextAlign: How lines of text align within it.
TimeMap: How a clip plays its source range over its timeline range.
TrackKind: The media domain accepted by a track.

Constants§

AUDIO_EFFECT_STATE_MARGIN_MS: How long before its start, beyond the lookahead of its track’s and its own compressors, a clip’s effects are counted as holding state, in milliseconds. A mixer builds a clip’s effects for the blocks that mix the clip and drops them for the first that does not; it reads a clip early by the lookahead after it, and one block of the largest size at the lowest project rate, with the rounding of each lookahead to samples, is within this margin. A clip’s effects count until its end.
CURRENT_SCHEMA_VERSION: The newest project-document schema understood by this build.
EQ_FREQUENCY_RANGE_HZ: The frequencies an EQ band may name, in hertz: from 10 Hz to just under the Nyquist frequency of the highest project rate. A band renders at most at 0.49 times the project’s own rate, so a band above that limit of the project it is in sounds at the limit.
EQ_Q_RANGE: The resonances an EQ band renders.
HDR_PEAK_NITS_RANGE: Supported ColorPipeline::hdr_peak_nits: SDR white to PQ’s ceiling.
LINE_HEIGHT_RANGE: The line heights text takes, as multiples of the font size.
LOOP_FADE_SECONDS: How long sound fades out before, and in after, each wrap of a repeating loop.
MAX_AUDIO_DELAY_MS: Most audio a project’s effects hold in delay lines at once, in milliseconds, summed over the effects MAX_AUDIO_EFFECTS counts then: a compressor holds its lookahead_ms, a reverb REVERB_DELAY_MS. At the highest project rate this is under 100 MB of state.
MAX_AUDIO_EFFECTS: Most audio effects a project holds state for at once: its master’s, its tracks’, and those of the clips holding state then (see AUDIO_EFFECT_STATE_MARGIN_MS). An equalizer is one effect, whatever its bands.
MAX_CANVAS_DIMENSION: Largest canvas edge accepted by the portable project model.
MAX_CANVAS_PIXELS: Largest total canvas area accepted before quality scaling.
MAX_CAPTION_FILE_BYTES: The largest subtitle file read, in bytes: a project document’s limit. A cue’s JSON is longer than its subtitle block, so a larger file holds more cues than a project can.
MAX_CAPTION_FILE_CUES: The most cues read from one subtitle file: a day of continuous captions at a cue every 1.3 s, and far more than a feature film’s few thousand.
MAX_CAPTION_FONT_SIZE: Largest caption font size, as a fraction of the frame’s height.
MAX_CHAIN_EFFECTS: Most effects one chain holds: a clip’s, a track’s, or the master’s.
MAX_CLIPS: Maximum number of clips across one project, and of sources in it.
MAX_CLIPS_PER_TRACK: Maximum number of clips on one track.
MAX_CLIP_MASKS: Most shape masks on one clip.
MAX_COMPOSITIONS: Most compositions in one project.
MAX_COMPOSITION_CONTROLS: Most controls a composition exposes as a group.
MAX_COMPOSITION_DEPTH: Deepest nesting of compositions: a composition that draws no other has depth one, and one that draws others, through group nodes or composition-source media nodes, is one deeper than the deepest of them. Drawing a composition recurses once per level.
MAX_COMPOSITION_NODES: Most nodes in one composition.
MAX_COMPRESSOR_RATIO: Largest compressor ratio.
MAX_ENVELOPE_MS: Longest attack or release time, in milliseconds.
MAX_EQ_BANDS: Most bands one equalizer holds.
MAX_EQ_GAIN_DB: Largest peaking gain, in decibels either way, an EQ band renders.
MAX_FONT_FAMILY_BYTES: Most bytes in a text style’s font family name.
MAX_FRAME_INLINED_NODES: Most composition nodes the clips on the timeline at one instant draw together: each clip counts the nodes its composition and its composition source draw once inlined, as MAX_INLINED_NODES counts them, and a clip showing neither counts none. MAX_INLINED_NODES bounds what the compositions compile to however often they are used; this bounds what their uses draw in one frame, so a frame never draws more than one composition of the largest size the project may hold.
MAX_GAIN: Largest linear amplitude accepted by any authored gain.
MAX_GRADIENT_STOPS: Most stops in one gradient.
MAX_GROUP_INPUTS: Most inputs a composition takes as a group.
MAX_INLINED_NODES: Most nodes the project’s compositions draw, summed over every composition: each node counts one, except a group node or a media node reading a composition source, which counts the nodes of the composition it draws. A composition is inlined wherever it is drawn, so this bounds what compiling the compositions makes, however their uses multiply.
MAX_LOOKAHEAD_MS: Longest compressor lookahead, in milliseconds.
MAX_LOTTIE_DOCUMENT_BYTES: Largest Lottie animation a provider reads, in bytes: a Lottie JSON file, or the animation a dotLottie package unpacks to. A Lottie document is parsed whole, to as much as about 45 times its size at its densest (numbers alone), so a larger one is refused before it is parsed, and a host refuses a larger file before reading it. Within it, a document parses to at most about 750 MiB.
MAX_MERGE_LAYERS: Most layers in one merge node.
MAX_PROJECT_JSON_BYTES: Largest project document accepted by Project::from_json.
MAX_PROVIDER_DOCUMENT_BYTES: Largest dotLottie package a provider reads, in bytes. A package is read whole, and only the files a source shows are unpacked from it: its animation, within MAX_LOTTIE_DOCUMENT_BYTES, and that animation’s images. A larger file is refused before it is read.
MAX_SEQUENCE_FRAMES: Most frames an image sequence lists.
MAX_SVG_DOCUMENT_BYTES: Largest SVG document a provider reads, in bytes. An SVG document is parsed whole, to as much as about 30 times its size at its densest (elements alone), so a larger one is refused before it is parsed, and a host refuses a larger file before reading it. Within it, a document parses to at most about 960 MiB.
MAX_TEXT_CONTENT_BYTES: Most bytes of text in one caption cue or text generator.
MAX_TEXT_GENERATOR_FONT_SIZE: Largest font size of a text generator, as a fraction of the canvas height.
MAX_TEXT_OUTLINE_WIDTH: Widest text outline, as a fraction of the font size.
MAX_TEXT_PADDING: Largest text background padding, as a fraction of the font size.
MAX_TRACKS: Maximum number of tracks in one project.
MAX_TRANSITIONS_PER_TRACK: Maximum number of transitions on one track.
MIN_LEVEL_DB: The quietest level a compressor threshold or a ducking depth names, in decibels; both are at most 0 dB.
REVERB_DELAY_MS: What one reverb counts toward MAX_AUDIO_DELAY_MS: its comb and allpass lines hold about 157 ms of audio at any rate.
SAMPLE_RATE_RANGE: Supported project audio sample-rate interval.

Functions§

caption_rows: Splits cues onto rows, each row a caption track’s worth: every cue goes to the first row that has room for another cue (a track holds MAX_CLIPS_PER_TRACK) and whose last cue has ended by its start.
parse_srt: Reads SubRip (.srt) text.
parse_webvtt: Reads WebVTT (.vtt) text. NOTE, STYLE, and REGION blocks are ignored.

Crate alpha2_model

Crate alpha2_model Copy item path

Structs§

Enums§

Constants§

Functions§

Crate alpha2_model