After Effects' Liquify shows Distortion ▸ Offset (x, y), and the renderer
already applied distortionMeshOffset — sampling the mesh at layer
coordinates minus the offset — but the parameter was declared hidden, so
users could neither see nor edit it, and #1194 read as 'the effect does
not work and the Offset option is missing'.
It is now a visible 2D Point control (still defaulting to 0,0, still
scriptable by its old path). Regression test: the spec exposes it as a
Point, a [20,0] offset moves the displaced-region centroid right by the
offset (the render differs, and the left of the shifted mesh is back to
untouched pixels), docs/effects.md refreshed via UPDATE_DOCS.
The reported flow — launch to the start screen, Open Demo Project, press
Space — does not reproduce on main: the demo comp activates, the start
screen closes, and playback starts. The behaviour already held (the
#279 menu-focus and #204 start-screen work fixed the neighbouring bugs),
so this lands as a regression test:
space_previews_after_opening_the_demo_from_home walks the exact flow
headlessly and asserts playing=true.
cargo test -p effectcraft-ui-egui --test ui_menus: 8 passed.
Looped *video* footage had a regression (#1045); a grouped PNG sequence
going through the same path — interpret a fresh placeholder to Loop 15,
drag it onto the timeline — did not. The behaviour is already correct on
main (addItem sizes a sequence layer from looped_duration, the pool's
frame_index wraps per cycle), so this is coverage, not a change:
interpret_footage_loop_extends_sequence_layers asserts a 15-loop of a
0.4 s sequence lands a 6 s layer.
cargo test -p effectcraft-engine --lib interpret_footage: 4 passed.
With the comp's and the layer's motion-blur switches on, the renderer
hands effects the shutter (angle, phase, samples) — CC Particle World
now uses it: eight to sixty-four sprite fields sampled across the shutter
are averaged into the transfer, so fast particles draw the motion trails
After Effects shows instead of one sharp frozen field.
The GPU particle pass defers to the CPU while the shutter is open
(catalog::gpu_supported), keeping the oracle identical on both backends.
Tests: sim::particle_world_motion_blur_trails (blurred differs from
sharp, lights more pixels, keeps alpha energy) and a catalog assertion
for the CPU gate; effects 450 and gpu 212 lib tests pass; fmt and
clippy -D warnings clean.
A Merge Paths item appended to a shape layer's Contents cleared every draw
collected so far, including fills and strokes inside groups above it, so
the layer's shapes vanished from the render. AE's rule is per group: only
paint items in the merge's own group above it are dropped. Draws now
record which contents level created them; the merge drops only its own
group's paint, and the group-bubbled draws keep rendering.
Adds a regression test (Merge Shapes and Add modes) that renders nothing
without the fix.
The pinning regression only exercised All Edges (and Left Edge in the CPU
unit test), so the reported 'does not work with Top/Bottom Edge' had no
coverage to catch or clear it. This holds the pinned first/last row in a
CPU test at Direction 90, and adds CPU/GPU parity cases for Top (4),
Bottom (6), diagonal directions, and single-edge pinning on shape layers.
CPU pinning verified correct on main: the guards in dst_wave_pin/wave_pin
agree; the gap was that neither edge path was tested.
A single selected_layers list kept ids from every comp, so the active comp looked selected whenever another comp was. Opening a comp and running a script against a different comp now park and restore that selection.
sanitize_state dropped every selected layer not in the active comp at the
end of each script run, so a layer selected by script in a comp that was not
open was gone by the next script call. Every comp keeps its own selection as
in After Effects; only ids gone from every comp are dropped now. UI and
engine lookups resolve selections per comp and tolerate foreign ids.
setInterpolationTypeAtKey and setTemporalEaseAtKey act on effect parameters
(a Slider Control's slider and a 3D Point Control), not just transform
properties. Verified on current main; lock it in with a script regression
test driving the per-key methods through the property object.
The engine's layer.create has shapesFromText and masksFromText, but the
timeline and Composition context menus only offered Create Shapes from Vector
Layer, so text could not be converted to editable outlines from the UI. Gate
the two text conversions on a text layer (LayerMenuCaps replaces the lone
vector flag), beside the vector one.
A script reading an expression that calls sampleImage on a footage layer got
[0,0,0,0]: the script evaluator built its EvalCtx with EvalCtx::new, which
has no footage source, so the renderer substituted transparent frames (solids
and shapes stayed correct because they are procedural). Pass the session's
footage source, as the app's renderer and diagnostics already do.
setParent re-expressed Position for 3D layers but skipped cameras and lights
entirely, so parenting a camera to a (default) null moved its eye by the
null's world transform and the framing jumped. Cameras and lights now get the
same parent-space Position fix, plus Point of Interest for two-node rigs
(values, keys, spatial tangents and ease speeds). They have no Scale, so
Orientation stays as authored.
Shift-click adds a vertex to the selection and removes it again, a plain
click collapses to one, and a marquee from empty viewer area selects the
enclosed vertices (additive with Shift) without moving the layer. These
behaviors regressed or were never covered; the tests drive them through the
Composition viewer like a user.
anim.reveal keyframes/animation only matched keyframed top-level properties,
so Time Remap vanished from U once its keys were deleted even though enabling
remapping is itself the modification (and freshly enabled remaps, which get
two keys, kept showing). Match the bare timeRemap prop for both kinds, as
modified already did.
The Properties panel kept its own ephemeral per-layer link toggle, so
unlinking Scale there left the Timeline linked and vice versa, and its edit
math differed from the Timeline's. Use the shared per-property unlinked set
and widgets::constrained, with the stable match id instead of the display
name, plus Alt-edits-one-axis like the Timeline.
The viewer.nudge shortcuts bailed out unless Composition, Viewer or Timeline
had focus, so with any other panel focused the arrows silently did nothing.
After Effects nudges from almost anywhere; only the Project panel owns the
arrows for its own keyboard navigation. Typing, dialogs, menus and selected
mask vertices still keep the keys.
Tests: arrow nudge from Effect Controls, no nudge from Project, Shift+arrow
10x, time untouched, click-row path.
Right-clicking selected layers and choosing Pre-compose grouped only the
right-clicked layer in 0.8.0 while Layer > Pre-compose grouped all of them.
The menu plumbing was fixed since; add a regression test driving the dialog
from both the outline row and the layer bar and asserting both selected
layers end up nested in the new composition.
The modern curve used g = 2^(2c), so Contrast 100 crushed 0.25 to about 8/255 while AE is about 40/255. Brightness is now a pin-black-and-white power toward white or black, then a milder S-curve g = 1 + 0.53 c, fitted to kasper9n's AE grey ramps. Legacy linear mode is unchanged. The GPU kernel matches the CPU.
Time > Pixel Motion Blur fetched the layer at its shutter times and averaged the decoded
frames as they were, so fast motion rendered as doubled copies and the accepted Vector
Detail parameter changed nothing. Build a TimewarpPlan instead: every shutter instant is
decoded between the two source frames it falls between along a measured motion field — the
same block-matching and in-between synthesis Time > Timewarp uses, with a search radius
wide enough for full-frame displacements and the vector grid tightened from Vector Detail
0 (one vector per 32 px) to 100 (one per 2 px) — and the instants average into the blur.
CPU and GPU share the plan (fx_time renders it with its Timewarp step loop, declining to
the CPU when there is no host), so the existing GPU parity cases for the effect compare
the new path. Falling back to the old average is kept for layers without a time host.
A render integration test moves a textured block 12 px between source frames, renders
the straddling shutter, and asserts the output is neither the still nor the plain frame
overlay, lays trail coverage along the block's path, and changes with Vector Detail;
without the synthesis it fails at the overlay assertion.
An authored layer's effect buffer spanned only its painted content — the glyph
bounding box for text, the drawn bounds for shapes — while the size the stack
measures from (effect_bounds) is the whole comp-sized layer rectangle. Tiling
and displacement effects computed correct output for pixels outside the painted
rectangle and then wrote them into the buffer that was too small, so CC Tiler
repeated only within the text box and CC Smear lost its liquid trail at the
glyph edge (#658). Buf::cover now grows the stack's input to the layer's
rectangle with transparent margins — what a sized source's buffer already is —
applied by run_effect_stack whenever an enabled video effect will run, and
natively on the GPU by reallocating and copying the texture. The placement and
the sized-source paths are unchanged; two regression tests (CC Tiler spans the
region, CC Smear trails past the text bounds) fail without the growth.
Importing an animated GIF as a movie (bfbdadd) made it preview and export
again, but nothing pinned it: the item must probe as a GIF movie and its
rendered frames must follow the file's own per-frame delays, not hold one
still. Add a host integration test that imports a red/green/blue GIF and
asserts each third of its duration renders the matching colour, so a
regression to still-frame import is caught.
The Scrub helper and the real-properties test discarded FullOutput, so the
first pass' font-atlas texture delta panicked on drop in debug builds —
which is how the FreeBSD CI gate failed. Consume the delta the way the
rest of the crate's egui tests do.
Pulse and PipeWire were opening a large default period and playing silence until the feeder caught up, so the picture ran while the sound started late. The stream now waits for queued samples and requests about 10 ms of buffering.
Timecode and Frames entries were parsed as absolute counts without the composition Start Timecode or Start Numbering Frames At offset, so committing the displayed text jumped to another frame.
Those edits offset the combined Position property, which a separated layer does not render. X and Y Position, including their keys, now move by the same amount.
The renderer follows X, Y and Z Position once those channels exist. The Position command was writing only the combined property, so the layer stayed where it was.
Setting the start past the current end, or shortening the composition, stored an inverted interval. Trimming that interval then saved a negative duration.
A same-named file is used only when kind, size, codec, and for moving pictures the rate and duration agree. Symlinks are not followed, and duplicate names are left alone.
Vikram Bhaskaran