- Portal IconSelect popups to document.body with position:fixed to prevent
clipping by modal overflow-y:auto
- Replace custom scene selectors in automation editor with EntitySelect
command-palette pickers (main scene + fallback scene)
- Add IconSelect grid for automation deactivation mode (none/revert/fallback)
- Add IconSelect grid for automation condition type and match type
- Replace mapped zone source dropdowns with EntitySelect pickers
- Replace scene target selector with EntityPalette.pick() pattern
- Remove effect palette preview bar from CSS editor
- Remove sensitivity badge from audio color strip source cards
- Clean up unused scene-selector CSS and scene-target-add-row CSS
- Add locale keys for all new UI elements across en/ru/zh
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Decouple processing loop from blocking TCP writes via single-slot buffer sender thread
- Build raw UpdateZoneLeds packets with struct.pack instead of RGBColor objects (reduces GC pressure)
- Add change-threshold frame dedup to minimize GPU I2C/SMBus writes that cause system stalls
- Set TCP_NODELAY and reduce socket timeout for lower latency
- Cache zone IDs for direct packet construction
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Rename all Python modules, classes, API endpoints, config keys, frontend
fetch URLs, and Home Assistant integration URLs from picture-targets to
output-targets. Store loads both new and legacy JSON keys for backward
compatibility with existing data files.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Autorestore fixes:
- Snapshot WLED state before connect() mutates it (lor, AudioReactive)
- Gate restore on auto_shutdown setting (was unconditional)
- Remove misleading auto_restore capability from serial provider
- Default auto_shutdown to false for all new devices
Protocol badge fixes:
- Show correct protocol per device type (OpenRGB SDK, MQTT, WebSocket)
- Was showing "Serial" for all non-WLED devices
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Extract shared process picker module (core/process-picker.js) used by
both automation conditions and notification CSS app filter
- Remove led_count property from notification CSS source (backend + frontend)
- Replace comma-separated app filter with newline-separated textarea + browse
- Inline color cycle add button (+) into the color row
- Fix notification app color layout to horizontal rows
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
These types always auto-size from the connected device — the explicit
led_count override was unused clutter. Streams now use getattr fallback.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
New source_type "notification" fires one-shot visual effects (flash, pulse, sweep)
triggered via POST webhook. Designed as a composite layer for overlay on persistent
sources. Includes app color mapping, whitelist/blacklist filtering, and auto-sizing.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add `startup` automation condition type that activates on server boot,
replacing the per-target `auto_start` flag
- Remove `auto_start` field from targets, scene snapshots, and all API layers
- Remove auto-start UI section and star buttons from dashboard and target cards
- Remove `color` field from scene presets (backend, API, modal, frontend)
- Add card color support to scene preset cards (color picker + border style)
- Show localStorage-backed card colors on all dashboard cards (targets,
automations, sync clocks, scene presets)
- Fix card color picker updating wrong card when duplicate data attributes
exist by using closest() from picker wrapper instead of global querySelector
- Add sync clocks step to Sources tab tutorial
- Bump SW cache v9 → v10
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Speed is now exclusively controlled via sync clocks — CSS sources no longer
carry their own speed/cycle_speed fields. Streams default to 1.0× when no
clock is assigned. Also fixes false-positive dirty check on the device
settings modal (array reference comparison) and converts several frontend
modules to use DataCache for consistent API response caching.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Introduces Synchronization Clocks — shared, controllable time bases
that CSS sources can optionally reference for synchronized animation.
Backend:
- New SyncClock dataclass, JSON store, Pydantic schemas, REST API
- Runtime clock with thread-safe pause/resume/reset and speed control
- Ref-counted runtime pool with eager creation for API control
- clock_id field on all ColorStripSource types
- Stream integration: clock time/speed replaces source-local values
- Paused clock skips rendering (saves CPU + stops frame pushes)
- Included in backup/restore via STORE_MAP
Frontend:
- Sync Clocks tab in Streams section with cards and controls
- Clock dropdown in CSS editor (hidden speed slider when clock set)
- Clock crosslink badge on CSS source cards (replaces speed badge)
- Targets tab uses DataCache for picture/audio sources and sync clocks
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Zone picker UI in device add/settings modals with per-zone checkbox selection
- Combined mode: pixels distributed sequentially across zones
- Separate mode: full effect resampled independently to each zone via linear interpolation
- Per-zone LED preview in target cards: one canvas strip per zone with hover overlay labels
- Zone badges on device cards enriched with actual LED counts from OpenRGB API
- Fix stale led_count by using device_led_count discovered at connect time
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Per-automation webhook URL with auto-generated 128-bit hex token.
External services (Home Assistant, IFTTT, curl) can POST to
/api/v1/webhooks/{token} with {"action": "activate"|"deactivate"}
to control automation state — no API key required (token is auth).
Backend: WebhookCondition model, engine state tracking with
immediate evaluation, webhook endpoint, schema/route updates.
Frontend: webhook option in condition editor, URL display with
copy button, card badge, i18n for en/ru/zh.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Cache camera enumeration results for 30s and limit probe range using
WMI camera count on Windows. Open source modal instantly with a loading
spinner while dropdowns are populated asynchronously.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- New CameraEngine using OpenCV VideoCapture for webcam capture
- HAS_OWN_DISPLAYS class attribute on CaptureEngine base to distinguish
engines with their own device lists from desktop monitor engines
- Display picker renders device list for cameras/scrcpy, spatial layout
for desktop monitors
- Engine-aware display label formatting (camera name vs monitor index)
- Stream modal properly loads engine-specific displays on template change,
edit, and clone
- Camera backend config rendered as dropdown (auto/dshow/msmf/v4l2)
- Remove offline label from device cards (healthcheck indicator suffices)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
New device type enabling control of keyboards, mice, RAM, GPU, and fans
via the OpenRGB SDK server (TCP port 6742). Includes auto-discovery,
health monitoring, state snapshot/restore, and fast synchronous pixel
send with brightness scaling. Also updates TODO.md with complexity notes.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add noise gate filter: suppresses per-pixel color flicker below threshold
using stateful frame comparison with pre-allocated int16 buffers
- Add palette quantization filter: maps pixels to nearest color in preset
or custom hex palette, using chunked processing for memory efficiency
- Add "string" option type to filter schema system (base, API, frontend)
- Replace up/down buttons with pointer-event drag-and-drop in PP template
filter list, with clone/placeholder feedback and modal auto-scroll
- Add frame_interpolation locale keys (was missing from all 3 locales)
- Update TODO.md: mark completed processing pipeline items
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- New WS device type: broadcaster singleton + LEDClient that sends binary
frames to connected WebSocket clients during processing
- FastAPI WS endpoint at /api/v1/devices/{device_id}/ws with token auth
- Frontend: add/edit WS devices, connection URL with copy button in settings
- Add health_check and auto_restore capabilities to WLED and Serial providers;
hide health interval and auto-restore toggle for devices without them
- Skip health check loop for virtual devices (Mock, MQTT, WS) — set always-online
- Copy buttons and labels for API CSS push endpoints (REST POST / WebSocket)
- Hide mock:// and ws:// URLs in target device dropdown
- Hide filter textbox when card section is collapsed (cs-collapsed CSS class)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Remove DeviceBrightnessSnapshot and AutomationSnapshot from scene data model
- Simplify capture_current_snapshot and apply_scene_state to targets only
- Remove device/automation dependencies from scene preset API routes
- Add target selector (combobox + add/remove) to scene capture modal
- Fix stale profiles reference bug in scene_preset_store recapture
- Update automation engine call sites for simplified scene functions
- Sync scene presets cache between automations and scene-presets modules
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Rename the "profiles" entity to "automations" throughout the entire
codebase for clarity. Updates Python models, storage, API routes/schemas,
engine, frontend JS modules, HTML templates, CSS classes, i18n keys
(en/ru/zh), dashboard, tutorials, and command palette.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Profiles now activate scene presets instead of individual targets, with
configurable deactivation behavior (none/revert/fallback scene). The
target checklist UI is replaced by a searchable combobox for scene
selection that scales well with many scenes.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Feature 1 — Profile Conditions: time-of-day, system idle (Win32
GetLastInputInfo), and display state (GUID_CONSOLE_DISPLAY_STATE)
condition types for automatic profile activation.
Feature 2 — Scene Presets: snapshot/restore system that captures target
running states, device brightness, and profile enables. Server-side
capture with 5-step activation order. Dedicated Scenes tab with
CardSection-based card grid, command palette integration, and dashboard
quick-activate section.
Feature 3 — MQTT Integration: MQTTService singleton with aiomqtt,
MQTTLEDClient device provider for pixel output, MQTT profile condition
type with topic/payload matching, and frontend support for MQTT device
type and condition editor.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Stability:
- Fix race condition: set _is_running before create_task in target processors
- Await probe task after cancel in wled_target_processor
- Replace raw fetch() with fetchWithAuth() across devices, kc-targets, pattern-templates
- Add try/catch to showTestTemplateModal in streams.js
- Wrap blocking I/O in asyncio.to_thread (picture_targets, system restore)
- Fix dashboardStopAll to filter only running targets with ok guard
Performance:
- Vectorize fire effect spark loop with numpy in effect_stream
- Vectorize FFT band binning with cumulative sum in analysis.py
- Rewrite pixel_processor with vectorized numpy (accept ndarray or list)
- Add httpx.AsyncClient connection pooling with lock in wled_provider
- Optimize _send_pixels_http to avoid np.hstack allocation in wled_client
- Mutate chart arrays in-place in dashboard, perf-charts, targets
- Merge dashboard 2-batch fetch into single Promise.all
- Hoist frame_time outside loop in mapped_stream
Usability:
- Fix health check interval load/save in device settings
- Swap confirm modal button classes (No=secondary, Yes=danger)
- Add aria-modal to audio/value source editors, fix close button aria-labels
- Add modal footer close button to settings modal
- Add dedicated calibration LED count validation error keys
i18n:
- Replace ~50 hardcoded English strings with t() calls across 12 JS files
- Add 50 new keys to en.json, ru.json, zh.json
- Localize inline toasts in index.html with window.t fallback
- Add data-i18n to command palette footer
- Add localization policy to CLAUDE.md
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add enumerate_devices_by_engine() returning per-engine device lists
without cross-engine dedup so frontend can filter correctly
- API /audio-devices now includes by_engine dict alongside flat list
- Frontend caches per-engine data, filters device dropdown by selected
template's engine_type, refreshes on template change
- Reorder getting-started tutorial: add API docs and accent color steps
- Fix tutorial trigger button focus outline persisting on step 2
- Use accent color variable for tutorial pulse ring animation
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
When brightness source (e.g. Audio Volume) returned 0 during silence,
the zero-brightness suppression path skipped all frames without sending
DDP keepalive packets, causing WLED to exit live mode after ~2.5s.
Now sends periodic keepalive even when suppressing zero-brightness frames.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Reduces map_leds_ms timing spikes from 4ms to ~1.5ms by eliminating
~540KB/frame of numpy temporary allocations:
- Pre-allocate _led_buf (reused instead of np.zeros per call)
- Pre-compute offset-adjusted segment indices (eliminates np.roll copy)
- Lazy-cache per-edge cumsum and mean buffers with np.mean/cumsum out=
- Pre-compute Phase 3 skip resampling arrays in __init__
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add protocol field (ddp/http) to storage, API schemas, routes, processor
- WledTargetProcessor passes protocol to create_led_client(use_ddp=...)
- Target editor: protocol dropdown + keepalive in collapsible Specific Settings
- FPS, brightness threshold, adaptive FPS moved to main form area
- Hide Specific Settings section for serial devices (protocol is WLED-only)
- Card badge: show DDP/HTTP for WLED devices, Serial for serial devices
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
DDP uses fire-and-forget UDP, so when a WiFi device becomes overwhelmed
by sustained traffic, sends appear successful while the device is
actually unreachable. This adds:
- HTTP liveness probe (GET /json/info, 2s timeout) every 10s during
streaming, exposed as device_streaming_reachable in target state
- Adaptive FPS (opt-in): exponential backoff when device is unreachable,
gradual recovery when it stabilizes — finds sustainable send rate
- Honest health checks: removed the lie that forced device_online=true
during streaming; now runs actual health checks regardless
- Target editor toggle, FPS display shows effective rate when throttled,
health dot reflects streaming reachability, red highlight when
unreachable
- Auto-backup scheduling support in settings modal
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Stability: Add outer try/except/finally with _running=False cleanup to all 6
processing loop methods (live, color_strip, effect, audio, composite, mapped).
Add exponential backoff on consecutive capture errors in live_stream. Move
audio stream.stop() outside lock scope.
Performance: Replace per-pixel Python loop with np.array().tobytes() in
ddp_client. Vectorize pixelate filter with cv2.resize down+up. Vectorize
gradient rendering with np.searchsorted.
Frontend: Add lockBody/unlockBody re-entrancy counter. Add {once:true} to
fetchWithAuth abort listener. Null ws.onclose before ws.close() in LED preview.
Backend: Remove auth token prefix from log messages. Add atomic_write_json
helper (tempfile + os.replace) and update all 10 stores. Add name uniqueness
checks to all update methods. Fix DELETE status codes to 204 in audio_sources
and value_sources. Fix get_source() silent bug in color_strip_sources.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Audio value source cards link to the referenced audio source
- Adaptive scene cards link to the referenced picture source
- Fix SceneValueStream starting at 0.5 regardless of actual scene;
first frame now skips smoothing to avoid artificial bias
- Add crosslinks guidance to CLAUDE.md card appearance section
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add real-time value source test: WebSocket endpoint streams get_value() at
~20Hz, frontend renders scrolling time-series chart with min/max/current stats
- Add auto-gain for audio value sources: rolling peak normalization with slow
decay, sensitivity range increased to 0.1-20.0
- Always show brightness overlay on LED preview when brightness source is set
- Refactor ValueStreamManager to shared ref-counted streams (value streams
produce scalars, not LED-count-dependent, so sharing is correct)
- Simplify acquire/release API: remove consumer_id parameter since streams
are no longer consumer-dependent
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
New per-target property: when effective output brightness
(max pixel value × device/source brightness) falls below
the threshold, LEDs turn off completely. Useful for cutting
dim flicker in audio-reactive and ambient setups.
Threshold slider (0–254) in target editor, badge on card,
hot-swap to running processors, persisted in storage.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Show effective brightness percentage on the LED preview when a
value source dims below 100%. Prepend a brightness byte to the
preview WebSocket wire format. Increase preview height to 32px.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add WebSocket endpoints for live audio spectrum streaming at ~20Hz
- Audio source test: resolves device/channel, shares stream via ref-counting
- Audio template test: includes device picker dropdown for selecting input
- Canvas-based 64-band spectrum visualizer with falling peaks and beat flash
- Channel-aware: mono sources show left/right/mixed spectrum correctly
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Introduces an engine+template abstraction for audio capture, mirroring the
existing screen capture engine pattern. This enables multiple audio backends
(WASAPI for Windows, sounddevice for cross-platform) with per-source
engine configuration via reusable templates.
Backend:
- AudioCaptureEngine ABC with WasapiEngine and SounddeviceEngine implementations
- AudioEngineRegistry for engine discovery and factory creation
- AudioAnalyzer class decouples FFT/RMS/beat analysis from engine-specific capture
- ManagedAudioStream wraps engine stream + analyzer in background thread
- AudioCaptureTemplate model and AudioTemplateStore with JSON CRUD
- AudioCaptureManager keyed by (engine_type, device_index, is_loopback)
- Auto-migration: default template created on startup, assigned to existing sources
- Full REST API: CRUD for audio templates + engine listing with availability flags
- audio_template_id added to MultichannelAudioSource model and API schemas
Frontend:
- Audio template cards in Streams > Audio tab with engine badge and config details
- Audio template editor modal with engine selector and dynamic config fields
- Audio template dropdown in multichannel audio source editor
- Template name crosslink badge on multichannel audio source cards
- Confirm modal z-index fix (always stacks above editor modals)
- i18n keys for EN and RU
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Charts: blue filled area for fps_actual (rolling avg), green line for
fps_current (real-time sends/sec)
- Labels: fps_current/fps_target as primary, avg fps_actual as secondary
- Track fps_current in metrics history for dashboard chart preload
- Applied to both LED targets page and dashboard
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Remove light.py platform (static color control via HA light entity)
- Add select.py with CSS Source and Brightness Source dropdowns for LED targets
- Coordinator now fetches color-strip-sources and value-sources lists
- Add generic update_target() method for partial target updates
- Clean up stale device registry entries on integration reload
- Skip frame sends when effective brightness is ~0 (suppresses unnecessary
UDP/HTTP traffic while LEDs are dark)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Fix inverted gamma formula: use (i/255)^gamma instead of (i/255)^(1/gamma)
so gamma>1 correctly darkens midtones (standard LED gamma correction)
- Fix frame interpolation flicker: move interp buffer update after temporal
smoothing so idle-tick output is consistent with new-frame output
- Fix target card hover/animation reset: use stable placeholder values in
card HTML for volatile metrics (data-tm attributes), patch real values
in-place after reconcile instead of replacing entire card DOM element
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add AlwaysCondition model and evaluation (always returns true)
- Add condition type selector (Always/Application) in profile editor
- Show condition type pill on profile cards
- Fix misleading empty-conditions text (was "never activate", actually always active)
- Add i18n keys for Always condition (en + ru)
- Add CSS for condition type selector and description
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Fall back to stored device rgbw field when health check doesn't report
it (mock devices have no hardware to query). Add emoji icons to all
property badges on audio source and value source cards.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add nonlocal declarations for _u16_a, _u16_b, _i32 in nested functions
_blend_u16 and _apply_corrections — Python treats augmented assignments
(*=, +=, >>=) as local variable bindings, causing UnboundLocalError
that prevented any frames from being sent to devices.
Move 🔢 emoji from brightness source label to dropdown option items.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
filter_template references were silently ignored in PP template test,
picture source test, and KC target test routes — they created a no-op
FilterTemplateFilter instead of expanding into the referenced template's
filters. Centralized expansion logic into PostprocessingTemplateStore.
resolve_filter_instances() and use it in all test routes + live stream
manager.
Also fixed empty template_id when adding filter_template filters: the
select dropdown showed the first template visually but onchange never
fired, saving "" instead. Now initializes with first choice's value and
auto-corrects stale/empty values at render time.
Other fixes: ScreenCapture dimensions now use actual image shape after
filter processing; brightness source label emoji updates.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Use i/(count-1) fraction (matching calibration dialog) so LEDs span
the full edge, and apply seg.reverse flag for correct numbering.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
audio_capture.py:
- Move _fft_bands from inner function to method (avoid per-frame closure)
- Pre-allocate channel split buffers and RMS scratch arrays
- Use in-place numpy ops (np.copyto, np.multiply) instead of copies
- In-place FFT smoothing instead of temp array allocation
- Cache loop-invariant values as locals
- Fix energy index to wrap-around instead of unbounded increment
audio_stream.py:
- Pre-compute interpolation arrays (band_x, led_x, full_amp, indices_buf,
vu_gradient) once on LED count change instead of every frame
- Pre-compute VU meter base/peak float arrays in _update_from_source
- Reuse full_amp and indices_buf buffers across frames
- In-place spectrum smoothing to avoid temp allocations
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The processing loop skipped resending when the CSS frame was
unchanged (static source). With a dynamic brightness value source
(e.g. audio), brightness can change every frame while colors stay
the same. Now compare effective brightness against previous value
and resend when it differs.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add resolve_audio_source() that accepts both MultichannelAudioSource
(defaults to mono mix) and MonoAudioSource. Update CSS and brightness
value source dropdowns to show all audio sources with type badges.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Pop send_latency_ms and rgbw from kwargs in WLED, Adalight, and
AmbiLED providers so mock-only fields don't leak through to
non-mock client constructors.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Instrument AudioCaptureStream with read/FFT timing and
AudioColorStripStream with render timing. Display audio-specific
timing segments (read/fft/render/send) in the target card
breakdown bar when an audio source is active.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Virtual device with configurable LED count, RGB/RGBW mode, and simulated
send latency. Includes full provider/client implementation, API schema
support, and frontend add/settings modal integration.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
