Since #659 every text or shape layer with an enabled video effect grew its
effect buffer to the comp-sized layer rectangle before the stack ran, so
glow and blur — which only need their own margins — allocated a 1920x1080
RGBA32F texture per layer per frame in flight (~5 GB peak VRAM in parallel
export, measured on an RTX 3090).
Effects that actually place pixels outside the painted content already
grow the buffer themselves by exactly what they need (CC Tiler, CC Smear,
CC Mosaic sharp via layer_cover_pad / Buf::pad), on CPU and GPU. The
blanket target.cover(layer_size) in Renderer::run_effects_on, the
FxTarget::cover trait method with its CPU default, and the GPU
GpuTarget::cover texture allocation are removed; Buf::cover/cover_plan
stay as the effects-crate helpers they were designed as.
The #659 regression tests (tiling across the layer region, centred-shape
edge pinning) and the render/effects/gpu suites pass unchanged.
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.
- verified: 0.9.2 on every page, checked against the v0.9.2 tag and release assets
- install, price-and-platforms: Windows x86 builds, riscv64 deb/rpm, macOS CLI zip
- gpu-not-used: GPU Information now names the adapter or why the GPU is off;
OpenGL means CPU compositing; #613 still open
- import-footage: TGA, HDR, DPX, Cineon sequences; dropped PSDs ask; drop on
an empty viewer makes a comp
- render-a-video: QuickTime (Uncompressed) format
- make-a-composition: Layer > New > Text starts typing; save and delete presets
- CLAUDE.md: regular file instead of a symlink
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Problem: on Windows two tests failed on main although the app is fine; they assumed
macOS/Linux path formats.
Fix: the CLI test strips the verbatim path prefix that canonicalize adds on Windows (a
child's current_dir() never reports it); the host sink test uses a drive-letter path on
Windows (a path rooted without a drive is joined to the home drive by the engine, the test
expected the current drive). Test code only.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Upstream's #1167 fix and this branch both stop marker edits from invalidating
cached frames. Kept this branch's shared Project::reads_markers/same_pixels_with/
pixel_form_with (viewer frames, disk cache, nested comps) and folded upstream's
rules into them: any expression counts as reading markers, protected time
markers with a duration still change the frames.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The startup hold waited for the current frame to be cached before
anything else, so a frame that can never be cached held the first frame
for ever: a zero RAM preview budget, or a single frame larger than the
budget (which the comp-size estimate misses for Extended Viewer frames).
It now gives up at once when the budget can't hold even one comp-sized
frame, and otherwise after STARTUP_WAIT (2 s from pressing play): the
sound starts from the first frame, and the preview goes on as a stalled
one does. frame_bytes() is factored out of frames_that_fit() for it.
Tests (ui_audio_scrub): with a budget of 0 or half a frame the sound
starts on the first step; with an Extended Viewer frame larger than the
whole budget (the estimate says one fits) the first frame holds for about
two seconds, then the sound starts. Both wait for ever on the previous
commit. The Extended Viewer setup is now shared by its two tests.
- Keyboard Shortcuts: Enter is not Delete while the delete confirmation is open (#1113 with #1019).
- Speed graph drag test compares proportions, since the Graph Editor now fits its selection (#1182 with #1109).
- Spanish panel test no longer demands the Chinese catalogs' exact rows (#996 with #1189, #1019).
- Object Matte gets its Simplified Chinese effect-name row (#1199 with #792).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Cryptomatte (#1183): planes shorter than the image read as empty instead of panicking.
- layer.addMarker (#1180): a comp marker no longer replaces one at the same time.
- Desktop (#1005): ignore macOS's -psn_ argument instead of refusing to launch.
- Doc comments put back on their own items (#1153, #1037).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The six PSD PRs merged one by one leave #1131's test calling samples()/to_rgb(), which #1139
replaced; vangel76's integration draft #1149 has the adapted version.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The nightly FreeBSD run measured 119 ms/frame against a 100 ms limit: the mean included
Tracker::new and any scheduling stalls on the two-core VM. The best single step with a
250 ms bound still catches an order-of-magnitude regression.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Vikram Bhaskaran