UYVY backend

src/README.md

@@ -13,8 +13,9 @@ RenderThread
   owns a hidden OpenGL context
   polls the oldest ready input frame without waiting
   uploads input frames into a render-owned GL texture
-  renders simple BGRA8 motion at selected cadence
-  queues async PBO readback
+  renders at selected cadence
+  optionally GPU-packs the output target into UYVY8
+  queues async PBO readback for the selected system-frame format
   publishes completed frames into SystemFrameExchange
 
 InputFrameMailbox
@@ -47,17 +48,17 @@ Startup builds a small output preroll reserve before DeckLink scheduled playback
 
 Included now:
 
-- generic video output edge contract with DeckLink as the current implementation
+- generic video output edge contract with DeckLink and NDI output implementations
 - optional DeckLink input edge with BGRA8 capture or raw UYVY8 capture decoded on the render thread
 - non-blocking startup when DeckLink output is unavailable
 - hidden render-thread-owned OpenGL context
 - simple smooth-motion renderer
-- BGRA8-only output
+- BGRA8 and render-thread-packed UYVY8 system-memory output
 - non-blocking three-frame FIFO input mailbox for render
 - fast contiguous mailbox copy path for matching input row strides
 - bounded three-frame input warmup before render cadence starts
 - render-thread-owned input texture upload
-- async PBO readback
+- async PBO readback for the configured system-frame format
 - bounded FIFO system-memory frame exchange
 - bounded completed-frame output preroll reserve before DeckLink playback, with DeckLink scheduled depth still targeted at four
 - conservative DeckLink schedule-lead telemetry and recovery
@@ -103,7 +104,7 @@ Those features should be ported only after the cadence spine is stable.
 This tracks parity with `apps/LoopThroughWithOpenGLCompositing`.
 
 - [x] Stable DeckLink output cadence
-- [x] BGRA8 system-memory output path
+- [x] BGRA8 and UYVY8 system-memory output paths
 - [x] Render thread owns its primary GL context
 - [x] Output startup warmup before scheduled playback
 - [x] Non-blocking startup when DeckLink output is unavailable
@@ -208,6 +209,7 @@ Currently consumed fields:
 - `output.device`
 - `output.resolution`
 - `output.frameRate`
+- `output.pixelFormat` (`auto`, `bgra8`, or `uyvy8`)
 - `output.keying.external`
 - `output.keying.alphaRequired`
 - `autoReload`
@@ -215,9 +217,11 @@ Currently consumed fields:
 - `previewEnabled`
 - `previewFps`
 
-`input.backend` and `output.backend` currently support `decklink` and `none`. Backend creation is routed through the app-side video backend factory, so new concrete backends can be added without making `main` or the render cadence path own their startup details.
+`input.backend` and `output.backend` currently support `decklink`, `ndi`, and `none`. Backend creation is routed through the app-side video backend factory, so new concrete backends can be added without making `main` or the render cadence path own their startup details.
 
-When `previewEnabled` is true, the preview window runs on `PreviewWindowThread`. It paints BGRA8 system-memory frames with Win32/GDI after render readback has already completed, so it does not bind GL and does not consume frames from DeckLink output. `previewFps` controls the preview repaint cadence; the default is 60 fps and `config/runtime-host.json` tracks the shipped 59.94 output cadence.
+`output.pixelFormat=auto` chooses UYVY8 for DeckLink/NDI output unless alpha output is required, in which case it uses BGRA8. Explicit `uyvy8` requests are rejected back to BGRA8 when alpha is required. V210/YUVA remain explicit unsupported render-readback states until matching render-thread packers exist.
+
+When `previewEnabled` is true, the preview window runs on `PreviewWindowThread`. It paints BGRA8 system-memory frames directly and decodes UYVY8 frames to BGRA for Win32/GDI after render readback has already completed, so it does not bind GL and does not consume frames from video output. `previewFps` controls the preview repaint cadence; the default is 60 fps and `config/runtime-host.json` tracks the shipped 59.94 output cadence.
 
 The loaded config is treated as a read-only startup snapshot. Subsystems that need config should receive this snapshot or a narrowed config struct from app orchestration; they should not reload files independently.
 
@@ -305,7 +309,7 @@ Input telemetry:
 - `renderFrameMs`: most recent render-thread draw duration, excluding completed-readback copy and readback queue work
 - `renderFrameBudgetUsedPercent`: most recent render draw time as a percentage of the selected frame budget
 - `renderFrameMaxMs`: maximum observed render-thread draw duration for this process
-- `readbackQueueMs`: time spent queueing the most recent async BGRA8 PBO readback
+- `readbackQueueMs`: time spent queueing the most recent async PBO readback for the selected system-frame format
 - `completedReadbackCopyMs`: time spent mapping/copying the most recent completed readback into system-memory frame storage
 - `completedDrops`: oldest completed unscheduled system-memory frames dropped because the bounded completed reserve overflowed; this is an app-side reserve drop, not a DeckLink dropped-frame report
 - `acquireMisses`: times render/readback could not acquire a writable system-memory frame slot; completed frames waiting for playout are preserved instead of being displaced
@@ -440,7 +444,7 @@ This app keeps the same core behavior but splits it into modules that can grow:
 - `render/`: cadence thread, clock, and simple renderer
 - `frames/InputFrameMailbox`: non-blocking bounded FIFO CPU input handoff with contiguous-copy fast path for matching row strides
 - `render/InputFrameTexture`: render-thread-owned upload of the currently acquired CPU input frame into GL, including raw UYVY8 decode into the shader-visible input texture
-- `render/readback/`: PBO-backed BGRA8 readback and completed-frame publication
+- `render/readback/`: PBO-backed BGRA8/UYVY8 readback and completed-frame publication
 - `render/thread/`: render thread lifecycle, GL startup/cadence loop, metrics, and runtime layer commit mailbox
 - `render/runtime/RuntimeRenderScene`: render-thread-owned GL scene for ready runtime shader layers
 - `render/runtime/RuntimeShaderPrepareWorker`: shared-context runtime shader program compile/link worker