* Render: memoized audio detection, bounded stretch, preroll, waveform, cache entries, 3D bounds, solids and selector time
comp_has_audio walked a precomp again for every layer using it; each comp
is answered once. A tiny stored stretch made one audio block fetch a vast
source range; playback speed is capped at 100x. Effect preroll is capped at
10 s. Waveforms stop where the audio ends and summarise at most a day.
Disk cache entries are decompressed only when their stated size matches
what their dimensions need (and at most 1 GiB). Advanced 3D triangle bounds
are clamped on both sides before narrowing. Solids larger than 16 Mpx are
made at a lower scale. A camera-facing layer's parent recursion is bounded.
A layer's Expression Selectors share one time allowance. A disk cache
folder under the temp folder is made private and may not be a link.
* address review: audio detection by breadth-first search, private cache folder resolved before the temp check, oversized layers not written, solid cap after rounding, selector static amount, viewport i32 clamp
* address review: private-folder check skips the temp folder itself; override value in expression fallback; assert alias cache opens
* address review: refuse the shared temp folder itself as a disk cache folder
* address review: shared counted selector budget, 30 s preroll, cache hardening (shared temp roots, matte discard, exact LZ4 length, render version 7), orient nesting guard
A project file, an .aep, a timeline document or a Lottie file could give a
layer a time stretch of 1e-300 %, 0 or 1e300 %. Layer and comp time divide
by it and the audio mix read about usize::MAX source samples for one block,
which overflowed or tried a huge allocation and aborted the app.
- limits::stretch brings a stretch within After Effects' range (±1 % to
±9900 %, Layer.stretch in the scripting guide), sign kept; 0 and NaN are
100 %. Opening a project file corrects each layer and says so; a stretch
written as null (NaN in an auto-save) now opens instead of failing.
- .aep import and Lottie import clamp and report the stretch; timeline
import clamps the clip speed and warns.
- layer.timeStretch refuses a non-number and more than 9900 %; the Time
Stretch dialog's ranges match.
- Layer::layer_time / comp_time and the audio mix use the clamped stretch,
so a value set in memory can't divide by zero or read without bound.
Fixes#3247
A layer with a negative time stretch was skipped by the mix. It now plays
its audio reversed at its stretched speed, through the same linear
resampler as forward stretch: a reversed read starts at the layer time of
the run's end and is played last to first, so comp sample k plays the
source just before layer_time(k) (at -100 % with the start at the out
point, the exact mirror Backwards gives a forward layer). Backwards on a
reversed layer plays it forwards.
Resampled runs start at the exact source position (from the layer time in
ticks) instead of the grid sample a block starts on, so mixing in frame
blocks and in export's one-second blocks gives the same samples, and they
read one more source sample so a block's last sample blends instead of
holding. Audio effects run on comp time since the layer's start, which
joins up across blocks at any stretch and equals the layer time at 100 %.
Reads are made in pieces of at most MAX_READ source frames, so memory
stays bounded for fast layers and nested fast precomps (the work for
nested 1 % precomps is not bounded; follow-up). A stretch faster than 1 %
is silent as a last-resort guard, with a tolerance so exactly 1 % plays.
A zero or non-finite stretch is skipped.
Fixes#3239
The mixdown evaluated levels and effect parameters once per block, and export mixes WAV/AIFF
in one-second blocks, so a level fade became a staircase that changed once a second. Animated
Audio Levels are now evaluated on a 1 ms grid of comp samples and interpolated in dB between,
and animated per-sample effects (Stereo Mixer, Backwards) run in 1 ms pieces with their
parameters at each piece. The grid is aligned to comp sample indices, so the result still does
not depend on the block size.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every undoable edit keeps the previous project for undo, so the comp it touches
is shared and Arc::make_mut deep-copies it: every layer and every keyframe. An
edit therefore costs time in proportion to the comp's size, and a script that
sets many keyframes in a big comp goes quadratic (a 1,600-layer build with
~100k sampled keys took ~45 minutes).
Comp::layers is now a Layers collection backed by Vec<Arc<Layer>>: copying the
stack copies pointers, and only the layer an edit changes is copied. The API
mirrors Vec<Layer>; serialized form is unchanged (a list of layers).
Comp::layer_mut and the iter_mut().filter/find call sites use find_mut /
iter_mut_where so they copy only the layers they actually change.
- prop.addKey in a 400-layer comp: 1.36 ms -> 0.16 ms per key (flat in comp size)
- the 1,600-layer script build: ~45 min -> 47 s, frames pixel-identical
- regression tests: an edit leaves the other layers shared with undo; layers
round-trip as a plain list
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VkpGcKfWyCudfSsQNiWAaG
Stretched layers copied the nearest source sample for each output sample, which adds error: at
48 kHz the worst error on a 440 Hz tone was 0.025, and 0.15 at 8 kHz. They now blend the two
source samples either side. Layers at 100% and 50% speed keep their exact samples.
The render tests cover a plain precomp, which mixes the same audio as its layers, and the stretch
error against the exact tone.
Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>
Comps expose controls (properties typed as text/color/slider/checkbox/point/angle/dropdown,
Media Replacement, groups, comments); precomp layers get an Essential Properties group whose
edits override per instance (Push to Comp / Revert); templates export/import as an open ZIP
(manifest.json + project subset + media + poster). Stretched precomps keep protected regions
at their original speed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
8/16 bpc projects clamp and quantise layer pixels after source/masks and
every effect, and the comp after every layer (16 bpc = 0..32768); 32 bpc
keeps over-range values. Blend modes without HDR support clamp their inputs
in 32 bpc. Working space (sRGB, Rec. 709, Rec. 2020, Display P3; matrices
derived from the published primaries), footage colour profiles from stream
metadata or Interpret Footage, Linearize Working Space and Blend Colors Using
1.0 Gamma. Layer cache keys include the colour settings. Project panel depth
button: click cycles, Alt-click opens Project Settings.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Audio effects (Backwards, Bass & Treble, Delay, Flange & Chorus,
High-Low Pass, Modulator, Parametric EQ, Reverb, Stereo Mixer, Tone)
applied per layer in the comp mixdown, so export and preview share them.
RBJ-cookbook biquads, Schroeder/Moorer reverb, modulated delay lines;
stateless per block via pre-roll. Tone makes any layer audible.
- Real-time audio preview: ui-egui AudioFeed/AudioPlayback (feeder thread
mixing ahead) behind an AudioDevice hook; the desktop app implements it
with cpal (L6). Audio clock drives playback (consumed frames minus device
latency, looped through the work area). Preview panel Include Audio /
playback.muteAudio.
- Audio panel VU meters with peak hold and clip indicators; selected
layer's Audio Levels.
- Timeline Audio > Waveform for footage with audio (L reveals Audio
Levels, LL the Waveform), drawn from a cached per-item peak summary.
- ui.playback reports audio state and meter levels.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>