feat(engine): add HDR video output pipeline (#265)

## Summary

Adds the ability to render HDR video output (H.265 10-bit, BT.2020) from HyperFrames compositions. When the renderer detects HDR source video, it automatically switches to the HDR output pipeline — no flags needed.

## What it does

- **Auto-detection** — Probes each video source with `ffprobe`. If any has bt2020/PQ/HLG color metadata, the output switches to H.265 10-bit with correct color tags. SDR-only compositions are unaffected (H.264, bt709).
- **HLG pass-through** — Native HLG pixels from FFmpeg extraction are piped directly to the encoder without conversion. This avoids brightness loss from HLG→linear→PQ conversion (which requires an OOTF system gamma we can't reliably apply).
- **Encoder HDR support** — Both chunk and streaming encoders accept HDR presets: `libx265`, `yuv420p10le`, BT.2020 color primaries, `hvc1` codec tag (required for Apple playback).
- **WebGPU HDR capture (gated)** — A complete WebGPU float16 readback pipeline is implemented and tested but gated behind headed Chrome (headless doesn't expose WebGPU). Ready for future use with WebGPU canvas content.
- **HDR utilities** — `detectTransfer()` (PQ vs HLG), `getHdrEncoderColorParams()`, `analyzeCompositionHdr()`. 15 unit tests.

## Key design decisions

| Decision | Why |
|----------|-----|
| No `--hdr` flag | SDR content encoded as HDR causes orange shift in browsers. Auto-detect eliminates this. |
| HLG pass-through (not HLG→PQ) | Conversion loses brightness without OOTF. Pass-through matches source exactly. |
| `hvc1` codec tag | Apple QuickTime requires `hvc1` (not `hev1`) for HEVC playback. |
| 1-hour streaming timeout | HDR capture at ~6fps needs more time than the default 10-minute FFmpeg timeout. |

## Files changed

| File | What changed |
|------|-------------|
| `packages/engine/src/utils/hdr.ts` | **NEW** — HDR detection, transfer types, encoder params (15 tests) |
| `packages/engine/src/services/hdrCapture.ts` | **NEW** — WebGPU readback, HLG conversion, PQ encode |
| `packages/engine/src/services/streamingEncoder.ts` | HDR presets, raw rgb48le input, color tags |
| `packages/engine/src/services/chunkEncoder.ts` | HDR presets, conditional color tags |
| `packages/producer/src/services/renderOrchestrator.ts` | Auto-detection loop, HDR pass-through capture path |

## How to test

Render a composition with an HDR video source. The output should be H.265 10-bit with HDR metadata visible in `ffprobe` (bt2020, arib-std-b67 or smpte2084). Plays correctly in QuickTime and on HDR displays.

## Stack position

**2 of 6** — Stacked on #258 (SDR/HDR normalization). Provides the encoder infrastructure that phases 1-5 build on.

🤖 Generated with [Claude Code](https://claude.com/claude-code)
This commit is contained in:
Vance Ingalls
2026-04-19 15:10:59 -07:00
committed by GitHub
parent d1f992570a
commit 5a3fde19d4
19 changed files with 1794 additions and 318 deletions
+11 -7
View File
@@ -21,7 +21,7 @@
},
"packages/cli": {
"name": "@hyperframes/cli",
"version": "0.3.0",
"version": "0.4.5",
"bin": {
"hyperframes": "./dist/cli.js",
},
@@ -61,7 +61,7 @@
},
"packages/core": {
"name": "@hyperframes/core",
"version": "0.3.0",
"version": "0.4.5",
"dependencies": {
"@chenglou/pretext": "^0.0.5",
},
@@ -87,7 +87,7 @@
},
"packages/engine": {
"name": "@hyperframes/engine",
"version": "0.3.0",
"version": "0.4.5",
"dependencies": {
"@hono/node-server": "^1.13.0",
"@hyperframes/core": "workspace:^",
@@ -98,13 +98,14 @@
},
"devDependencies": {
"@types/node": "^22.10.1",
"@webgpu/types": "^0.1.69",
"typescript": "^5.7.2",
"vitest": "^3.2.4",
},
},
"packages/player": {
"name": "@hyperframes/player",
"version": "0.3.0",
"version": "0.4.5",
"devDependencies": {
"tsup": "^8.0.0",
"typescript": "^5.0.0",
@@ -113,7 +114,7 @@
},
"packages/producer": {
"name": "@hyperframes/producer",
"version": "0.3.0",
"version": "0.4.5",
"dependencies": {
"@fontsource/archivo-black": "^5.2.8",
"@fontsource/eb-garamond": "^5.2.7",
@@ -144,6 +145,7 @@
"@fontsource/roboto": "^5.2.10",
"@fontsource/source-code-pro": "^5.2.7",
"@types/node": "^22.10.1",
"@webgpu/types": "^0.1.69",
"esbuild": "^0.27.2",
"tsx": "^4.7.0",
"typescript": "^5.7.2",
@@ -151,7 +153,7 @@
},
"packages/shader-transitions": {
"name": "@hyperframes/shader-transitions",
"version": "0.3.0",
"version": "0.4.5",
"dependencies": {
"html2canvas": "^1.4.1",
},
@@ -163,7 +165,7 @@
},
"packages/studio": {
"name": "@hyperframes/studio",
"version": "0.3.0",
"version": "0.4.5",
"dependencies": {
"@codemirror/autocomplete": "^6.20.1",
"@codemirror/commands": "^6.10.3",
@@ -781,6 +783,8 @@
"@vitest/utils": ["@vitest/utils@3.2.4", "", { "dependencies": { "@vitest/pretty-format": "3.2.4", "loupe": "3.2.1", "tinyrainbow": "2.0.0" } }, "sha512-fB2V0JFrQSMsCo9HiSq3Ezpdv4iYaXRG1Sx8edX3MwxfyNn83mKiGzOcH+Fkxt4MHxr3y42fQi1oeAInqgX2QA=="],
"@webgpu/types": ["@webgpu/types@0.1.69", "", {}, "sha512-RPmm6kgRbI8e98zSD3RVACvnuktIja5+yLgDAkTmxLr90BEwdTXRQWNLF3ETTTyH/8mKhznZuN5AveXYFEsMGQ=="],
"acorn": ["acorn@8.16.0", "", { "bin": { "acorn": "bin/acorn" } }, "sha512-UVJyE9MttOsBQIDKw1skb9nAwQuR5wuGD3+82K6JgJlm/Y+KI92oNsMNGZCYdDsVtRHSak0pcV5Dno5+4jh9sw=="],
"adm-zip": ["adm-zip@0.5.16", "", {}, "sha512-TGw5yVi4saajsSEgz25grObGHEUaDrniwvA2qwSC060KfqGPdglhvPMA2lPIoxs3PQIItj2iag35fONcQqgUaQ=="],
+1
View File
@@ -28,6 +28,7 @@
},
"devDependencies": {
"@types/node": "^22.10.1",
"@webgpu/types": "^0.1.69",
"typescript": "^5.7.2",
"vitest": "^3.2.4"
},
+16
View File
@@ -52,6 +52,12 @@ export interface EngineConfig {
/** Timeout for FFmpeg streaming encode (ms). Default: 600_000 */
ffmpegStreamingTimeout: number;
// ── HDR ──────────────────────────────────────────────────────────────
/** HDR output transfer function. false = SDR output (default). */
hdr: { transfer: "hlg" | "pq" } | false;
/** Auto-detect HDR from video sources when hdr is not explicitly set. */
hdrAutoDetect: boolean;
// ── Media ────────────────────────────────────────────────────────────
audioGain: number;
frameDataUriCacheLimit: number;
@@ -96,6 +102,9 @@ export const DEFAULT_CONFIG: EngineConfig = {
ffmpegProcessTimeout: 300_000,
ffmpegStreamingTimeout: 600_000,
hdr: false,
hdrAutoDetect: true,
audioGain: 1.35,
frameDataUriCacheLimit: 256,
@@ -170,6 +179,13 @@ export function resolveConfig(overrides?: Partial<EngineConfig>): EngineConfig {
DEFAULT_CONFIG.ffmpegStreamingTimeout,
),
hdr: (() => {
const raw = env("PRODUCER_HDR_TRANSFER");
if (raw === "hlg" || raw === "pq") return { transfer: raw };
return undefined;
})(),
hdrAutoDetect: envBool("PRODUCER_HDR_AUTO_DETECT", DEFAULT_CONFIG.hdrAutoDetect),
audioGain: envNum("PRODUCER_AUDIO_GAIN", DEFAULT_CONFIG.audioGain),
frameDataUriCacheLimit: Math.max(
32,
+20
View File
@@ -153,3 +153,23 @@ export {
} from "./utils/ffprobe.js";
export { downloadToTemp, isHttpUrl } from "./utils/urlDownloader.js";
export {
initHdrReadback,
uploadAndReadbackHdrFrame,
convertHdrFrameToRgb48le,
float16ToPqRgb,
buildHdrChromeArgs,
launchHdrBrowser,
} from "./services/hdrCapture.js";
export {
isHdrColorSpace,
detectTransfer,
getHdrEncoderColorParams,
analyzeCompositionHdr,
type HdrTransfer,
type HdrEncoderColorParams,
type CompositionHdrInfo,
} from "./utils/hdr.js";
export type { VideoColorSpace } from "./utils/ffprobe.js";
@@ -247,3 +247,91 @@ describe("buildEncoderArgs color space", () => {
expect(args).not.toContain("-video_track_timescale");
});
});
describe("getEncoderPreset HDR", () => {
it("returns h265 with 10-bit for HDR HLG", () => {
const preset = getEncoderPreset("standard", "mp4", { transfer: "hlg" });
expect(preset.codec).toBe("h265");
expect(preset.pixelFormat).toBe("yuv420p10le");
expect(preset.hdr).toEqual({ transfer: "hlg" });
});
it("returns h265 with 10-bit for HDR PQ", () => {
const preset = getEncoderPreset("high", "mp4", { transfer: "pq" });
expect(preset.codec).toBe("h265");
expect(preset.pixelFormat).toBe("yuv420p10le");
expect(preset.hdr).toEqual({ transfer: "pq" });
});
it("avoids ultrafast preset for HDR (upgrades to fast)", () => {
const preset = getEncoderPreset("draft", "mp4", { transfer: "hlg" });
expect(preset.preset).toBe("fast");
});
it("ignores HDR for webm format", () => {
const preset = getEncoderPreset("standard", "webm", { transfer: "hlg" });
expect(preset.codec).toBe("vp9");
expect(preset.hdr).toBeUndefined();
});
it("ignores HDR for mov format", () => {
const preset = getEncoderPreset("standard", "mov", { transfer: "pq" });
expect(preset.codec).toBe("prores");
expect(preset.hdr).toBeUndefined();
});
});
describe("buildEncoderArgs HDR color space", () => {
const baseOptions = { fps: 30, width: 1920, height: 1080 };
const inputArgs = ["-framerate", "30", "-i", "frames/%04d.png"];
it("keeps bt709 color tags when HDR flag is set but frames are still Chrome sRGB captures", () => {
// HDR flag gives H.265 + 10-bit encoding but pixels are still sRGB/bt709.
// Tagging as bt2020 causes orange shift — so we tag truthfully as bt709.
const args = buildEncoderArgs(
{ ...baseOptions, codec: "h265", preset: "medium", quality: 23, hdr: { transfer: "hlg" } },
inputArgs,
"out.mp4",
);
expect(args[args.indexOf("-colorspace:v") + 1]).toBe("bt709");
expect(args[args.indexOf("-color_primaries:v") + 1]).toBe("bt709");
expect(args[args.indexOf("-color_trc:v") + 1]).toBe("bt709");
const paramIdx = args.indexOf("-x265-params");
expect(args[paramIdx + 1]).toContain("colorprim=bt709");
expect(args[paramIdx + 1]).toContain("transfer=bt709");
});
it("uses bt709 when HDR is not set", () => {
const args = buildEncoderArgs(
{ ...baseOptions, codec: "h265", preset: "medium", quality: 23 },
inputArgs,
"out.mp4",
);
expect(args[args.indexOf("-colorspace:v") + 1]).toBe("bt709");
expect(args[args.indexOf("-color_trc:v") + 1]).toBe("bt709");
});
it("uses range conversion (not colorspace) for HDR CPU encoding", () => {
// Chrome screenshots are sRGB — we don't convert primaries (causes color shifts).
// Just range-convert and let the bt2020 container metadata + 10-bit handle the rest.
const args = buildEncoderArgs(
{ ...baseOptions, codec: "h265", preset: "medium", quality: 23, hdr: { transfer: "hlg" } },
inputArgs,
"out.mp4",
);
const vfIdx = args.indexOf("-vf");
expect(vfIdx).toBeGreaterThan(-1);
expect(args[vfIdx + 1]).toContain("scale=in_range=pc:out_range=tv");
expect(args[vfIdx + 1]).not.toContain("colorspace");
});
it("uses same range conversion for SDR CPU encoding", () => {
const args = buildEncoderArgs(
{ ...baseOptions, codec: "h264", preset: "medium", quality: 23 },
inputArgs,
"out.mp4",
);
const vfIdx = args.indexOf("-vf");
expect(args[vfIdx + 1]).toContain("scale=in_range=pc:out_range=tv");
});
});
+35 -9
View File
@@ -10,6 +10,7 @@ import { copyFileSync, existsSync, mkdirSync, readdirSync, statSync, writeFileSy
import { join, dirname } from "path";
import { DEFAULT_CONFIG, type EngineConfig } from "../config.js";
import { type GpuEncoder, getCachedGpuEncoder, getGpuEncoderName } from "../utils/gpuEncoder.js";
import { type HdrTransfer } from "../utils/hdr.js";
import { runFfmpeg } from "../utils/runFfmpeg.js";
import type { EncoderOptions, EncodeResult, MuxResult } from "./chunkEncoder.types.js";
@@ -21,15 +22,24 @@ export const ENCODER_PRESETS = {
high: { preset: "slow", quality: 15, codec: "h264" as const },
};
export interface EncoderPreset {
preset: string;
quality: number;
codec: "h264" | "h265" | "vp9" | "prores";
pixelFormat: string;
hdr?: { transfer: HdrTransfer };
}
/**
* Get encoder preset for a given quality and output format.
* WebM uses VP9 with alpha-capable pixel format; MP4 uses h264;
* WebM uses VP9 with alpha-capable pixel format; MP4 uses h264 (or h265 for HDR);
* MOV uses ProRes 4444 with alpha for editor-compatible transparency.
*/
export function getEncoderPreset(
quality: "draft" | "standard" | "high",
format: "mp4" | "webm" | "mov" = "mp4",
): { preset: string; quality: number; codec: "h264" | "vp9" | "prores"; pixelFormat: string } {
hdr?: { transfer: HdrTransfer },
): EncoderPreset {
const base = ENCODER_PRESETS[quality];
if (format === "webm") {
return {
@@ -47,6 +57,15 @@ export function getEncoderPreset(
pixelFormat: "yuva444p10le",
};
}
if (hdr) {
return {
preset: base.preset === "ultrafast" ? "fast" : base.preset,
quality: base.quality,
codec: "h265",
pixelFormat: "yuv420p10le",
hdr,
};
}
return { ...base, pixelFormat: "yuv420p" };
}
@@ -109,9 +128,8 @@ export function buildEncoderArgs(
if (bitrate) args.push("-b:v", bitrate);
else args.push("-crf", String(quality));
// Encoder-specific params: anti-banding + bt709 color space.
// Encoder-specific params: anti-banding + color space tagging.
// aq-mode=3 redistributes bits to dark flat areas (gradients).
// colorprim/transfer/colormatrix embed bt709 in the H.264/H.265 VUI.
const xParamsFlag = codec === "h264" ? "-x264-params" : "-x265-params";
const colorParams = "colorprim=bt709:transfer=bt709:colormatrix=bt709";
if (preset === "ultrafast") {
@@ -120,6 +138,10 @@ export function buildEncoderArgs(
args.push(xParamsFlag, `aq-mode=3:aq-strength=0.8:deblock=1,1:${colorParams}`);
}
}
// Apple devices require hvc1 tag for HEVC playback (default hev1 won't open in QuickTime)
if (codec === "h265") {
args.push("-tag:v", "hvc1");
}
} else if (codec === "vp9") {
args.push("-c:v", "libvpx-vp9", "-b:v", bitrate || "0", "-crf", String(quality));
args.push("-deadline", preset === "ultrafast" ? "realtime" : "good");
@@ -134,8 +156,12 @@ export function buildEncoderArgs(
return [...args, "-y", outputPath];
}
// BT.709 color space metadata — Chrome screenshots are sRGB which maps to bt709.
// Tags the output so players interpret colors correctly across devices.
// Color space metadata — tags the output so players interpret colors correctly.
// Chrome screenshots are always sRGB/bt709 pixels regardless of --hdr flag.
// We tag truthfully as bt709 even for HDR output — the --hdr flag gives
// H.265 + 10-bit encoding (better quality/compression) without lying about
// the color space. Tagging as bt2020 when pixels are bt709 causes browsers
// to apply the wrong color transform, producing visible orange/warm shifts.
if (codec === "h264" || codec === "h265") {
args.push(
"-colorspace:v",
@@ -148,15 +174,15 @@ export function buildEncoderArgs(
"tv",
);
// Convert full-range RGB input (Chrome screenshots) to limited/TV range for H.264.
// VAAPI already has a -vf chain for hwupload; prepend range conversion to it.
// Range conversion: Chrome's full-range RGB → limited/TV range.
if (gpuEncoder === "vaapi") {
// Replace the existing VAAPI -vf with one that includes range conversion
const vfIdx = args.indexOf("-vf");
if (vfIdx !== -1) {
args[vfIdx + 1] = `scale=in_range=pc:out_range=tv,${args[vfIdx + 1]}`;
}
} else if (!shouldUseGpu) {
// Range conversion: Chrome screenshots are full-range RGB.
// The scale filter handles both 8-bit and 10-bit correctly.
args.push("-vf", "scale=in_range=pc:out_range=tv");
}
@@ -1,3 +1,5 @@
import type { HdrTransfer } from "../utils/hdr.js";
export interface EncoderOptions {
fps: number;
width: number;
@@ -8,6 +10,7 @@ export interface EncoderOptions {
bitrate?: string;
pixelFormat?: string;
useGpu?: boolean;
hdr?: { transfer: HdrTransfer };
}
export interface EncodeResult {
@@ -0,0 +1,159 @@
import { describe, it, expect } from "vitest";
import { float16ToPqRgb } from "./hdrCapture.js";
// IEEE 754 half-precision (float16) bit patterns used to feed
// `float16ToPqRgb`. Encoding rule: sign(1) | exp(5) | frac(10).
const F16_ZERO = 0x0000; // +0.0
const F16_HALF = 0x3800; // +0.5 (exp=14, frac=0 → 2^-1)
const F16_ONE = 0x3c00; // +1.0 (exp=15, frac=0 → 2^0 — SDR white)
// PQ caps at 10000 nits and SDR_NITS = 203, so the linear input must exceed
// ~58x SDR white before linearToPQ(L) clips at 1.0. 1024 is well above that.
const F16_OVERBRIGHT = 0x6400; // +1024.0 (exp=25, frac=0 → 2^10)
function makeFloat16Frame(
width: number,
height: number,
pixel: { r: number; g: number; b: number; a: number },
bytesPerRow: number = width * 8,
): Buffer {
// Row-padded layout matches WebGPU readback: bytesPerRow ≥ width * 8 (4
// channels × 2 bytes), with garbage bytes after each row's pixel data.
const buf = Buffer.alloc(height * bytesPerRow);
for (let y = 0; y < height; y++) {
for (let x = 0; x < width; x++) {
const idx = y * bytesPerRow + x * 8;
buf.writeUInt16LE(pixel.r, idx);
buf.writeUInt16LE(pixel.g, idx + 2);
buf.writeUInt16LE(pixel.b, idx + 4);
buf.writeUInt16LE(pixel.a, idx + 6);
}
}
return buf;
}
describe("float16ToPqRgb", () => {
it("returns a buffer of width * height * 6 bytes (rgb48le)", () => {
const frame = makeFloat16Frame(4, 3, { r: 0, g: 0, b: 0, a: 0 });
const out = float16ToPqRgb(frame, 32, 4, 3);
expect(out.length).toBe(4 * 3 * 6);
});
it("encodes float16 black to PQ zero (linearToPQ(0) ≈ 0 after uint16 quantization)", () => {
const frame = makeFloat16Frame(2, 2, {
r: F16_ZERO,
g: F16_ZERO,
b: F16_ZERO,
a: F16_ZERO,
});
const out = float16ToPqRgb(frame, 16, 2, 2);
for (let i = 0; i < out.length; i += 2) {
expect(out.readUInt16LE(i)).toBe(0);
}
});
it("clamps overbright float16 input to PQ 65535 (linearToPQ(>>1.0) → 1.0)", () => {
// ~1024 linear is well past the 58x-SDR PQ saturation point; output caps
// at 1.0 → 65535 in uint16.
const frame = makeFloat16Frame(2, 2, {
r: F16_OVERBRIGHT,
g: F16_OVERBRIGHT,
b: F16_OVERBRIGHT,
a: F16_ZERO,
});
const out = float16ToPqRgb(frame, 16, 2, 2);
for (let pixel = 0; pixel < 4; pixel++) {
const dst = pixel * 6;
expect(out.readUInt16LE(dst)).toBe(65535);
expect(out.readUInt16LE(dst + 2)).toBe(65535);
expect(out.readUInt16LE(dst + 4)).toBe(65535);
}
});
it("preserves channel ordering R, G, B (alpha is discarded)", () => {
// Distinct float16 values per channel verify the function doesn't
// mix them up. Alpha is set high but should not appear in the output.
const frame = makeFloat16Frame(1, 1, {
r: F16_ONE,
g: F16_HALF,
b: F16_ZERO,
a: F16_ONE,
});
const out = float16ToPqRgb(frame, 8, 1, 1);
const r = out.readUInt16LE(0);
const g = out.readUInt16LE(2);
const b = out.readUInt16LE(4);
expect(r).toBeGreaterThan(g);
expect(g).toBeGreaterThan(b);
expect(b).toBe(0);
});
it("is monotonic: higher float16 input produces higher PQ output", () => {
const dark = makeFloat16Frame(1, 1, { r: F16_ZERO, g: 0, b: 0, a: 0 });
const mid = makeFloat16Frame(1, 1, { r: F16_HALF, g: 0, b: 0, a: 0 });
const bright = makeFloat16Frame(1, 1, { r: F16_ONE, g: 0, b: 0, a: 0 });
const r0 = float16ToPqRgb(dark, 8, 1, 1).readUInt16LE(0);
const r1 = float16ToPqRgb(mid, 8, 1, 1).readUInt16LE(0);
const r2 = float16ToPqRgb(bright, 8, 1, 1).readUInt16LE(0);
expect(r0).toBe(0);
expect(r1).toBeGreaterThan(r0);
expect(r2).toBeGreaterThan(r1);
});
it("is deterministic across calls with the same input", () => {
const frame = makeFloat16Frame(3, 2, {
r: F16_HALF,
g: F16_ONE,
b: F16_ZERO,
a: F16_ONE,
});
const a = float16ToPqRgb(frame, 24, 3, 2);
const b = float16ToPqRgb(frame, 24, 3, 2);
expect(a.equals(b)).toBe(true);
});
it("handles padded bytesPerRow (WebGPU 256-byte alignment)", () => {
// WebGPU readback pads rows to 256-byte multiples. For a 4-pixel-wide
// frame the actual pixel data is 32 bytes but bytesPerRow is 256.
const width = 4;
const height = 2;
const bytesPerRow = 256;
const frame = makeFloat16Frame(
width,
height,
{ r: F16_HALF, g: F16_HALF, b: F16_HALF, a: 0 },
bytesPerRow,
);
const out = float16ToPqRgb(frame, bytesPerRow, width, height);
expect(out.length).toBe(width * height * 6);
// Every R component should be the same non-zero value (uniform input).
const expected = out.readUInt16LE(0);
expect(expected).toBeGreaterThan(0);
for (let pixel = 0; pixel < width * height; pixel++) {
expect(out.readUInt16LE(pixel * 6)).toBe(expected);
}
});
it("ignores garbage bytes in the row padding region", () => {
// Stuff junk into the trailing padding to make sure the PQ encoder
// walks via bytesPerRow stride and not via raw buffer position.
const width = 2;
const height = 2;
const bytesPerRow = 64;
const frame = makeFloat16Frame(
width,
height,
{ r: F16_ZERO, g: F16_ZERO, b: F16_ZERO, a: F16_ZERO },
bytesPerRow,
);
for (let y = 0; y < height; y++) {
const padStart = y * bytesPerRow + width * 8;
for (let i = padStart; i < (y + 1) * bytesPerRow; i++) {
frame[i] = 0xff;
}
}
const out = float16ToPqRgb(frame, bytesPerRow, width, height);
for (let i = 0; i < out.length; i += 2) {
expect(out.readUInt16LE(i)).toBe(0);
}
});
});
+354
View File
@@ -0,0 +1,354 @@
/// <reference types="@webgpu/types" />
/**
* HDR Capture Service
*
* Captures HDR video frames via WebGPU float16 readback.
*
* The pipeline:
* 1. FFmpeg extracts raw HDR pixels (rgba64le) from video sources
* 2. Node converts HLG/PQ signal linear light float16
* 3. writeTexture uploads float16 data to WebGPU rgba16float texture
* 4. (Optional) WebGPU shader applies GSAP CSS transform
* 5. readback extracts float16 RGBA via base64 transfer
* 6. Node converts linear float16 PQ signal pipe to FFmpeg H.265
*
* Requirements:
* - Headed Chrome (not headless) WebGPU unavailable in headless mode
* - GPU access (Metal on macOS, Vulkan+NVIDIA on Linux)
*
* Performance: ~6 fps at 1080x1920 via base64 transfer.
*/
import type { Page, Browser, PuppeteerNode } from "puppeteer-core";
import { existsSync, readdirSync } from "fs";
import { join } from "path";
import { homedir } from "os";
// ── PQ (SMPTE 2084) OETF ─────────────────────────────────────────────────────
const PQ_M1 = 0.1593017578125;
const PQ_M2 = 78.84375;
const PQ_C1 = 0.8359375;
const PQ_C2 = 18.8515625;
const PQ_C3 = 18.6875;
const PQ_MAX_NITS = 10000.0;
const SDR_NITS = 203.0;
function linearToPQ(L: number): number {
const Lp = Math.max(0, (L * SDR_NITS) / PQ_MAX_NITS);
const Lm1 = Math.pow(Lp, PQ_M1);
return Math.pow((PQ_C1 + PQ_C2 * Lm1) / (1.0 + PQ_C3 * Lm1), PQ_M2);
}
function float16Decode(h: number): number {
const sign = (h >> 15) & 1;
const exp = (h >> 10) & 0x1f;
const frac = h & 0x3ff;
if (exp === 0) return (sign ? -1 : 1) * Math.pow(2, -14) * (frac / 1024);
if (exp === 31) return frac ? NaN : sign ? -Infinity : Infinity;
return (sign ? -1 : 1) * Math.pow(2, exp - 15) * (1 + frac / 1024);
}
// ── Browser-side interface ────────────────────────────────────────────────────
interface HdrCaptureRuntime {
uploadAndReadback(float16Base64: string): Promise<{ base64: string; bytesPerRow: number }>;
}
// ── Initialization ────────────────────────────────────────────────────────────
/**
* Inject the WebGPU HDR readback runtime into the page.
*
* Creates an rgba16float render texture that accepts writeTexture uploads
* and provides readback via base64 transfer.
*/
export async function initHdrReadback(page: Page, width: number, height: number): Promise<boolean> {
return page.evaluate(
async (w: number, h: number): Promise<boolean> => {
if (!navigator.gpu) return false;
const adapter = await navigator.gpu.requestAdapter();
if (!adapter) return false;
const device = await adapter.requestDevice();
const bytesPerPixel = 8; // rgba16float = 4 channels × 2 bytes
const bytesPerRow = Math.ceil((w * bytesPerPixel) / 256) * 256;
// Render texture — includes COPY_DST for writeTexture uploads
const renderTexture = device.createTexture({
size: [w, h],
format: "rgba16float",
usage:
GPUTextureUsage.RENDER_ATTACHMENT |
GPUTextureUsage.COPY_SRC |
GPUTextureUsage.COPY_DST |
GPUTextureUsage.TEXTURE_BINDING,
});
const readBuffer = device.createBuffer({
size: bytesPerRow * h,
usage: GPUBufferUsage.COPY_DST | GPUBufferUsage.MAP_READ,
});
const captureRuntime = {
device,
renderTexture,
readBuffer,
bytesPerRow,
width: w,
height: h,
/**
* Upload pre-converted float16 RGBA data and read it back.
* The float16 data must be row-aligned to bytesPerRow.
*
* Input: base64-encoded Uint16Array (float16 RGBA, row-padded)
* Output: base64-encoded readback of the same texture
*/
async uploadAndReadback(
float16Base64: string,
): Promise<{ base64: string; bytesPerRow: number }> {
// Decode base64 → Uint8Array
const binary = atob(float16Base64);
const bytes = new Uint8Array(binary.length);
for (let i = 0; i < binary.length; i++) bytes[i] = binary.charCodeAt(i);
// Upload to texture
device.queue.writeTexture(
{ texture: renderTexture },
bytes.buffer,
{ bytesPerRow, rowsPerImage: h },
[w, h],
);
// Readback
const encoder = device.createCommandEncoder();
encoder.copyTextureToBuffer(
{ texture: renderTexture },
{ buffer: readBuffer, bytesPerRow },
[w, h],
);
device.queue.submit([encoder.finish()]);
await readBuffer.mapAsync(GPUMapMode.READ);
const readBytes = new Uint8Array(readBuffer.getMappedRange().slice(0));
readBuffer.unmap();
// Base64 encode in chunks
let b64 = "";
const chunkSize = 32768;
for (let i = 0; i < readBytes.length; i += chunkSize) {
const slice = readBytes.subarray(i, Math.min(i + chunkSize, readBytes.length));
b64 += String.fromCharCode(...slice);
}
return { base64: btoa(b64), bytesPerRow };
},
};
(window as unknown as Record<string, unknown>).__hfHdrCapture = captureRuntime;
return true;
},
width,
height,
);
}
// ── HDR frame conversion ──────────────────────────────────────────────────────
/**
* Convert raw rgba64le pixels (from FFmpeg) to a base64 string for FFmpeg encoding.
*
* For HLG sources: the pixel values are already HLG-encoded. We pass them through
* as-is (normalized to 16-bit) and tag the output as HLG. No OETF conversion needed
* the HLG signal values ARE the correct encoding. Converting to linear and back to
* PQ produces worse results because every viewer's PQdisplay tone-mapping differs
* from its HLGdisplay tone-mapping.
*
* The WebGPU round-trip is skipped for pass-through the pixels go directly from
* FFmpeg extraction to FFmpeg encoding. WebGPU is only needed when transforms
* (scale, rotate, opacity from GSAP) must be applied to the HDR pixels.
*/
export function convertHdrFrameToRgb48le(
rawRgba64le: Buffer,
width: number,
height: number,
): Buffer {
const input = new Uint16Array(
rawRgba64le.buffer,
rawRgba64le.byteOffset,
rawRgba64le.byteLength / 2,
);
// Convert RGBA → RGB (drop alpha) for rgb48le output
const output = Buffer.alloc(width * height * 6);
for (let y = 0; y < height; y++) {
for (let x = 0; x < width; x++) {
const srcIdx = (y * width + x) * 4;
const dstIdx = (y * width + x) * 6;
output.writeUInt16LE(input[srcIdx] ?? 0, dstIdx);
output.writeUInt16LE(input[srcIdx + 1] ?? 0, dstIdx + 2);
output.writeUInt16LE(input[srcIdx + 2] ?? 0, dstIdx + 4);
}
}
return output;
}
// ── Frame upload + readback ───────────────────────────────────────────────────
/**
* Upload a float16 frame to WebGPU and read it back.
* Call after converting with convertHdrFrameToFloat16Base64.
*/
export async function uploadAndReadbackHdrFrame(
page: Page,
float16Base64: string,
): Promise<{ rawBuffer: Buffer; bytesPerRow: number }> {
const result = await page.evaluate(
async (b64: string): Promise<{ base64: string; bytesPerRow: number }> => {
const hdr = (window as unknown as Record<string, unknown>).__hfHdrCapture as
| HdrCaptureRuntime
| undefined;
if (!hdr) throw new Error("HDR capture not initialized");
return hdr.uploadAndReadback(b64);
},
float16Base64,
);
return {
rawBuffer: Buffer.from(result.base64, "base64"),
bytesPerRow: result.bytesPerRow,
};
}
// ── PQ conversion ─────────────────────────────────────────────────────────────
/**
* Convert float16 RGBA readback to PQ-encoded rgb48le for FFmpeg.
*/
export function float16ToPqRgb(
rawBuffer: Buffer,
bytesPerRow: number,
width: number,
height: number,
): Buffer {
const data = new Uint16Array(rawBuffer.buffer, rawBuffer.byteOffset, rawBuffer.byteLength / 2);
const channelsPerRow = bytesPerRow / 2;
const output = Buffer.alloc(width * height * 6);
for (let y = 0; y < height; y++) {
for (let x = 0; x < width; x++) {
const srcIdx = y * channelsPerRow + x * 4;
const r = float16Decode(data[srcIdx] ?? 0);
const g = float16Decode(data[srcIdx + 1] ?? 0);
const b = float16Decode(data[srcIdx + 2] ?? 0);
const dstIdx = (y * width + x) * 6;
output.writeUInt16LE(Math.round(Math.min(1.0, linearToPQ(r)) * 65535), dstIdx);
output.writeUInt16LE(Math.round(Math.min(1.0, linearToPQ(g)) * 65535), dstIdx + 2);
output.writeUInt16LE(Math.round(Math.min(1.0, linearToPQ(b)) * 65535), dstIdx + 4);
}
}
return output;
}
// ── Chrome launch ─────────────────────────────────────────────────────────────
function resolveHeadedChromePath(): string | undefined {
const baseDir = join(homedir(), ".cache", "puppeteer", "chrome");
if (!existsSync(baseDir)) return undefined;
const versions = readdirSync(baseDir).sort().reverse();
for (const version of versions) {
const candidates = [
join(
baseDir,
version,
"chrome-mac-arm64",
"Google Chrome for Testing.app",
"Contents",
"MacOS",
"Google Chrome for Testing",
),
join(
baseDir,
version,
"chrome-mac-x64",
"Google Chrome for Testing.app",
"Contents",
"MacOS",
"Google Chrome for Testing",
),
join(baseDir, version, "chrome-linux64", "chrome"),
join(baseDir, version, "chrome-win64", "chrome.exe"),
];
for (const binary of candidates) {
if (existsSync(binary)) return binary;
}
}
return undefined;
}
/**
* Launch a headed Chrome browser with WebGPU enabled.
*/
export async function launchHdrBrowser(
width: number,
height: number,
): Promise<{ browser: Browser; page: Page }> {
let ppt: PuppeteerNode | undefined;
try {
const mod = await import("puppeteer" as string);
ppt = mod.default;
} catch (err) {
const code = (err as NodeJS.ErrnoException | undefined)?.code;
if (code !== "ERR_MODULE_NOT_FOUND" && code !== "MODULE_NOT_FOUND") {
throw err;
}
const mod = await import("puppeteer-core");
ppt = mod.default;
}
if (!ppt) throw new Error("Neither puppeteer nor puppeteer-core found");
const chromePath = resolveHeadedChromePath();
if (!chromePath) {
throw new Error(
"[HDR] No Chrome binary found. Install: npx @puppeteer/browsers install chrome@stable",
);
}
const browser = await ppt.launch({
headless: false,
executablePath: chromePath,
args: buildHdrChromeArgs(width, height),
});
const page = await browser.newPage();
await page.setViewport({ width, height });
return { browser, page };
}
export function buildHdrChromeArgs(width: number, height: number): string[] {
return [
"--enable-unsafe-webgpu",
"--no-sandbox",
"--disable-setuid-sandbox",
"--window-position=-10000,-10000",
`--window-size=${width},${height}`,
"--disable-background-timer-throttling",
"--disable-backgrounding-occluded-windows",
"--disable-renderer-backgrounding",
"--disable-background-media-suspend",
"--disable-extensions",
"--disable-component-update",
"--disable-default-apps",
"--disable-sync",
"--no-zygote",
"--force-gpu-mem-available-mb=4096",
];
}
@@ -0,0 +1,160 @@
/**
* buildStreamingArgs unit tests.
*
* These tests focus on the FFmpeg CLI shape rather than spawning the encoder
* they're the cheap regression net for the HDR static-metadata bug
* (side_data=[none] in the encoded MP4) reproduced by
* packages/producer/scripts/hdr-smoke.ts. Without these assertions, future
* refactors of the x265-params string can silently strip
* master-display / max-cll and ship as SDR BT.2020 again.
*/
import { describe, expect, it } from "vitest";
import { buildStreamingArgs, type StreamingEncoderOptions } from "./streamingEncoder.js";
import { DEFAULT_HDR10_MASTERING } from "../utils/hdr.js";
const baseHdrPq: StreamingEncoderOptions = {
fps: 30,
width: 1920,
height: 1080,
codec: "h265",
preset: "medium",
quality: 23,
pixelFormat: "yuv420p10le",
useGpu: false,
rawInputFormat: "rgb48le",
hdr: { transfer: "pq" },
};
const baseHdrHlg: StreamingEncoderOptions = {
...baseHdrPq,
hdr: { transfer: "hlg" },
};
const baseSdr: StreamingEncoderOptions = {
fps: 30,
width: 1920,
height: 1080,
codec: "h264",
preset: "medium",
quality: 23,
useGpu: false,
};
function getX265ParamsValue(args: string[]): string | undefined {
const idx = args.indexOf("-x265-params");
return idx === -1 ? undefined : args[idx + 1];
}
describe("buildStreamingArgs", () => {
describe("HDR PQ (libx265)", () => {
it("emits master-display and max-cll in -x265-params", () => {
const args = buildStreamingArgs(baseHdrPq, "/tmp/out.mp4");
const x265 = getX265ParamsValue(args);
expect(x265).toBeDefined();
expect(x265).toContain(`master-display=${DEFAULT_HDR10_MASTERING.masterDisplay}`);
expect(x265).toContain(`max-cll=${DEFAULT_HDR10_MASTERING.maxCll}`);
expect(x265).toContain("colorprim=bt2020");
expect(x265).toContain("transfer=smpte2084");
expect(x265).toContain("colormatrix=bt2020nc");
});
it("tags the output stream with bt2020 / smpte2084 / tv range", () => {
const args = buildStreamingArgs(baseHdrPq, "/tmp/out.mp4");
expect(args).toContain("-colorspace:v");
expect(args[args.indexOf("-colorspace:v") + 1]).toBe("bt2020nc");
expect(args[args.indexOf("-color_primaries:v") + 1]).toBe("bt2020");
expect(args[args.indexOf("-color_trc:v") + 1]).toBe("smpte2084");
expect(args[args.indexOf("-color_range") + 1]).toBe("tv");
});
it("uses libx265 with -tag:v hvc1 for QuickTime compatibility", () => {
const args = buildStreamingArgs(baseHdrPq, "/tmp/out.mp4");
const cvIdx = args.indexOf("-c:v");
expect(cvIdx).toBeGreaterThan(-1);
expect(args[cvIdx + 1]).toBe("libx265");
expect(args).toContain("-tag:v");
expect(args[args.indexOf("-tag:v") + 1]).toBe("hvc1");
});
it("keeps the aq-mode prefix even with master-display present", () => {
const args = buildStreamingArgs(baseHdrPq, "/tmp/out.mp4");
const x265 = getX265ParamsValue(args);
expect(x265?.startsWith("aq-mode=3")).toBe(true);
});
it("uses the simpler aq-mode prefix on ultrafast preset", () => {
const args = buildStreamingArgs({ ...baseHdrPq, preset: "ultrafast" }, "/tmp/out.mp4");
const x265 = getX265ParamsValue(args);
expect(x265?.startsWith("aq-mode=3:")).toBe(true);
expect(x265).not.toContain("aq-strength");
expect(x265).toContain(`master-display=${DEFAULT_HDR10_MASTERING.masterDisplay}`);
});
});
describe("HDR HLG (libx265)", () => {
it("emits master-display, max-cll, and the HLG transfer", () => {
const args = buildStreamingArgs(baseHdrHlg, "/tmp/out.mp4");
const x265 = getX265ParamsValue(args);
expect(x265).toContain("transfer=arib-std-b67");
expect(x265).toContain(`master-display=${DEFAULT_HDR10_MASTERING.masterDisplay}`);
expect(x265).toContain(`max-cll=${DEFAULT_HDR10_MASTERING.maxCll}`);
});
it("tags the output stream with arib-std-b67", () => {
const args = buildStreamingArgs(baseHdrHlg, "/tmp/out.mp4");
expect(args[args.indexOf("-color_trc:v") + 1]).toBe("arib-std-b67");
});
});
describe("HDR raw input tagging", () => {
it("tags the rawvideo input with the matching color metadata", () => {
const args = buildStreamingArgs(baseHdrPq, "/tmp/out.mp4");
const inputColorTrcIdx = args.indexOf("-color_trc");
expect(inputColorTrcIdx).toBeGreaterThan(-1);
expect(args[inputColorTrcIdx + 1]).toBe("smpte2084");
const inputPrimariesIdx = args.indexOf("-color_primaries");
expect(inputPrimariesIdx).toBeGreaterThan(-1);
expect(args[inputPrimariesIdx + 1]).toBe("bt2020");
// Pix_fmt of the raw input must match the buffer we hand FFmpeg.
expect(args.indexOf("rgb48le")).toBeGreaterThan(-1);
});
it("does not strip the input color tags when bitrate is set instead of CRF", () => {
const args = buildStreamingArgs({ ...baseHdrPq, bitrate: "20M" }, "/tmp/out.mp4");
const x265 = getX265ParamsValue(args);
expect(x265).toContain(`master-display=${DEFAULT_HDR10_MASTERING.masterDisplay}`);
expect(args).toContain("-b:v");
expect(args[args.indexOf("-b:v") + 1]).toBe("20M");
});
});
describe("SDR fallback", () => {
it("does NOT emit HDR mastering metadata for SDR encodes", () => {
const args = buildStreamingArgs(baseSdr, "/tmp/out.mp4");
const x264 = args[args.indexOf("-x264-params") + 1];
expect(x264).toContain("colorprim=bt709");
expect(x264).toContain("transfer=bt709");
expect(x264).toContain("colormatrix=bt709");
expect(x264).not.toContain("master-display");
expect(x264).not.toContain("max-cll");
});
it("tags SDR output with bt709 and tv range", () => {
const args = buildStreamingArgs(baseSdr, "/tmp/out.mp4");
expect(args[args.indexOf("-color_trc:v") + 1]).toBe("bt709");
expect(args[args.indexOf("-color_primaries:v") + 1]).toBe("bt709");
expect(args[args.indexOf("-colorspace:v") + 1]).toBe("bt709");
expect(args[args.indexOf("-color_range") + 1]).toBe("tv");
});
});
describe("output path", () => {
it("places the output path last after -y", () => {
const args = buildStreamingArgs(baseHdrPq, "/tmp/some-output.mp4");
expect(args[args.length - 2]).toBe("-y");
expect(args[args.length - 1]).toBe("/tmp/some-output.mp4");
});
});
});
+102 -35
View File
@@ -17,6 +17,7 @@ import { existsSync, mkdirSync, statSync } from "fs";
import { dirname } from "path";
import { type GpuEncoder, getCachedGpuEncoder, getGpuEncoderName } from "../utils/gpuEncoder.js";
import { getHdrEncoderColorParams } from "../utils/hdr.js";
import { type EncoderOptions } from "./chunkEncoder.types.js";
import { DEFAULT_CONFIG, type EngineConfig } from "../config.js";
@@ -79,6 +80,9 @@ export interface StreamingEncoderOptions {
pixelFormat?: string;
useGpu?: boolean;
imageFormat?: "jpeg" | "png";
hdr?: { transfer: import("../utils/hdr.js").HdrTransfer };
/** When set, use rawvideo input instead of image2pipe. For HDR PQ-encoded frames. */
rawInputFormat?: "rgb48le";
}
export interface StreamingEncoderResult {
@@ -98,8 +102,11 @@ export interface StreamingEncoder {
* Build FFmpeg args for streaming (image2pipe) input.
* Reuses the same codec/quality/GPU logic as chunkEncoder's buildEncoderArgs
* but with `-f image2pipe` instead of `-i <pattern>`.
*
* Exported so unit tests can assert on the constructed CLI without spawning
* FFmpeg see streamingEncoder.test.ts.
*/
function buildStreamingArgs(
export function buildStreamingArgs(
options: StreamingEncoderOptions,
outputPath: string,
gpuEncoder: GpuEncoder = null,
@@ -116,19 +123,41 @@ function buildStreamingArgs(
} = options;
// Input args: pipe from stdin
const inputCodec = imageFormat === "png" ? "png" : "mjpeg";
const args: string[] = [
"-f",
"image2pipe",
"-vcodec",
inputCodec,
"-framerate",
String(fps),
"-i",
"-",
"-r",
String(fps),
];
const args: string[] = [];
if (options.rawInputFormat) {
// Raw pixel input (HLG/PQ-encoded rgb48le from FFmpeg extraction).
// Tag the input with the correct color space so FFmpeg uses the right
// YUV matrix when converting rgb48le → yuv420p10le for encoding.
// Without these tags FFmpeg assumes bt709 and applies the wrong matrix.
const hdrTransfer = options.hdr?.transfer;
const inputColorTrc =
hdrTransfer === "pq" ? "smpte2084" : hdrTransfer === "hlg" ? "arib-std-b67" : undefined;
args.push(
"-f",
"rawvideo",
"-pix_fmt",
options.rawInputFormat,
"-s",
`${options.width}x${options.height}`,
"-framerate",
String(fps),
);
if (inputColorTrc) {
args.push(
"-color_primaries",
"bt2020",
"-color_trc",
inputColorTrc,
"-colorspace",
"bt2020nc",
);
}
args.push("-i", "-");
} else {
const inputCodec = imageFormat === "png" ? "png" : "mjpeg";
args.push("-f", "image2pipe", "-vcodec", inputCodec, "-framerate", String(fps), "-i", "-");
}
args.push("-r", String(fps));
const shouldUseGpu = useGpu && gpuEncoder !== null;
@@ -169,16 +198,25 @@ function buildStreamingArgs(
if (bitrate) args.push("-b:v", bitrate);
else args.push("-crf", String(quality));
// Encoder-specific params: anti-banding + bt709 color space.
// aq-mode=3 redistributes bits to dark flat areas (gradients).
// colorprim/transfer/colormatrix embed bt709 in the H.264/H.265 VUI.
// Encoder-specific params: anti-banding + color space tagging.
// For HDR, getHdrEncoderColorParams also emits the SMPTE ST 2086
// mastering-display and CTA-861.3 MaxCLL/MaxFALL SEI messages —
// without them, players (Apple, YouTube, HDR TVs) treat the file
// as SDR BT.2020 and tone-map incorrectly.
const xParamsFlag = codec === "h264" ? "-x264-params" : "-x265-params";
const colorParams = "colorprim=bt709:transfer=bt709:colormatrix=bt709";
const colorParams =
options.rawInputFormat && options.hdr
? getHdrEncoderColorParams(options.hdr.transfer).x265ColorParams
: "colorprim=bt709:transfer=bt709:colormatrix=bt709";
if (preset === "ultrafast") {
args.push(xParamsFlag, `aq-mode=3:${colorParams}`);
} else {
args.push(xParamsFlag, `aq-mode=3:aq-strength=0.8:deblock=1,1:${colorParams}`);
}
// Apple devices require hvc1 tag for HEVC playback (default hev1 won't open in QuickTime)
if (codec === "h265") {
args.push("-tag:v", "hvc1");
}
}
} else if (codec === "vp9") {
args.push("-c:v", "libvpx-vp9", "-b:v", bitrate || "0", "-crf", String(quality));
@@ -194,27 +232,47 @@ function buildStreamingArgs(
return [...args, "-y", outputPath];
}
// BT.709 color space metadata — Chrome screenshots are sRGB which maps to bt709.
// Tags the output so players interpret colors correctly across devices.
// Color space metadata.
// When rawInputFormat is set, data comes from the WebGPU HDR pipeline
// (PQ-encoded) — tag with bt2020/PQ truthfully.
// Otherwise, Chrome captures sRGB — tag as bt709.
if (codec === "h264" || codec === "h265") {
args.push(
"-colorspace:v",
"bt709",
"-color_primaries:v",
"bt709",
"-color_trc:v",
"bt709",
"-color_range",
"tv",
);
if (options.rawInputFormat && options.hdr) {
args.push(
"-colorspace:v",
"bt2020nc",
"-color_primaries:v",
"bt2020",
"-color_trc:v",
options.hdr.transfer === "pq" ? "smpte2084" : "arib-std-b67",
"-color_range",
"tv",
);
} else {
args.push(
"-colorspace:v",
"bt709",
"-color_primaries:v",
"bt709",
"-color_trc:v",
"bt709",
"-color_range",
"tv",
);
}
// Convert full-range RGB input (Chrome screenshots) to limited/TV range for H.264.
if (gpuEncoder === "vaapi") {
// Video filter for range/color conversion.
// Raw HDR input (from WebGPU pipeline) is already PQ-encoded — no conversion needed.
// Chrome screenshots need full→TV range conversion.
if (options.rawInputFormat) {
// No filter needed — PQ data goes straight to encoder
} else if (gpuEncoder === "vaapi") {
const vfIdx = args.indexOf("-vf");
if (vfIdx !== -1) {
args[vfIdx + 1] = `scale=in_range=pc:out_range=tv,${args[vfIdx + 1]}`;
}
} else if (!shouldUseGpu) {
// Range conversion: Chrome screenshots are full-range RGB.
args.push("-vf", "scale=in_range=pc:out_range=tv");
}
@@ -304,7 +362,15 @@ export async function spawnStreamingEncoder(
if (exitStatus !== "running" || !ffmpeg.stdin || ffmpeg.stdin.destroyed) {
return false;
}
return ffmpeg.stdin.write(buffer);
// Copy the buffer before writing — Node streams hold a reference to the
// provided buffer and drain it asynchronously. The HDR path's compositor
// reuses pre-allocated transOutput/normalCanvas buffers across frames,
// so without this copy the pipe would read partially-overwritten data
// and flicker. The SDR path doesn't invoke writeFrame at all (it pipes
// PNG files via encodeFramesFromDir), so the memcpy here is HDR-only
// and justified by correctness.
const copy = Buffer.from(buffer);
return ffmpeg.stdin.write(copy);
},
close: async (): Promise<StreamingEncoderResult> => {
@@ -312,9 +378,10 @@ export async function spawnStreamingEncoder(
if (signal) signal.removeEventListener("abort", onAbort);
// Close stdin to signal end of input
if (ffmpeg.stdin && !ffmpeg.stdin.destroyed) {
const stdin = ffmpeg.stdin;
if (stdin && !stdin.destroyed) {
await new Promise<void>((resolve) => {
ffmpeg.stdin!.end(() => resolve());
stdin.end(() => resolve());
});
}
@@ -10,7 +10,9 @@ import { existsSync, mkdirSync, readdirSync, rmSync } from "fs";
import { join } from "path";
import { parseHTML } from "linkedom";
import { extractVideoMetadata, type VideoMetadata } from "../utils/ffprobe.js";
import { isHdrColorSpace as isHdrColorSpaceUtil } from "../utils/hdr.js";
import { downloadToTemp, isHttpUrl } from "../utils/urlDownloader.js";
import { runFfmpeg } from "../utils/runFfmpeg.js";
import { DEFAULT_CONFIG, type EngineConfig } from "../config.js";
export interface VideoElement {
@@ -114,18 +116,28 @@ export async function extractVideoFramesRange(
const framePattern = `frame_%05d.${format}`;
const outputPattern = join(videoOutputDir, framePattern);
const args: string[] = [
"-ss",
String(startTime),
"-i",
videoPath,
"-t",
String(duration),
"-vf",
`fps=${fps}`,
"-q:v",
format === "jpg" ? String(Math.ceil((100 - quality) / 3)) : "0",
];
// When extracting from HDR source, tone-map to SDR in FFmpeg rather than
// letting Chrome's uncontrollable tone-mapper handle it (which washes out).
// macOS: VideoToolbox hardware decoder does HDR→SDR natively on Apple Silicon.
// Linux: zscale filter (when available) or colorspace filter as fallback.
const isHdr = isHdrColorSpaceUtil(metadata.colorSpace);
const isMacOS = process.platform === "darwin";
const args: string[] = [];
if (isHdr && isMacOS) {
args.push("-hwaccel", "videotoolbox");
}
args.push("-ss", String(startTime), "-i", videoPath, "-t", String(duration));
const vfFilters: string[] = [];
if (isHdr && isMacOS) {
// VideoToolbox tone-maps during decode; force output to bt709 SDR format
vfFilters.push("format=nv12");
}
vfFilters.push(`fps=${fps}`);
args.push("-vf", vfFilters.join(","));
args.push("-q:v", format === "jpg" ? String(Math.ceil((100 - quality) / 3)) : "0");
if (format === "png") args.push("-compression_level", "6");
args.push("-y", outputPattern);
@@ -195,6 +207,53 @@ export async function extractVideoFramesRange(
});
}
/**
* Convert an SDR video to HDR color space (HLG / BT.2020) so it can be
* composited alongside HDR content without looking washed out.
*
* Uses zscale for color space conversion with a nominal peak luminance of
* 600 nits high enough that SDR content doesn't appear too dark next to
* HDR, matching the approach used by HeyGen's Rio pipeline.
*/
async function convertSdrToHdr(
inputPath: string,
outputPath: string,
signal?: AbortSignal,
config?: Partial<Pick<EngineConfig, "ffmpegProcessTimeout">>,
): Promise<void> {
const timeout = config?.ffmpegProcessTimeout ?? DEFAULT_CONFIG.ffmpegProcessTimeout;
const args = [
"-i",
inputPath,
"-vf",
"colorspace=all=bt2020:iall=bt709:range=tv",
"-color_primaries",
"bt2020",
"-color_trc",
"arib-std-b67",
"-colorspace",
"bt2020nc",
"-c:v",
"libx264",
"-preset",
"fast",
"-crf",
"16",
"-c:a",
"copy",
"-y",
outputPath,
];
const result = await runFfmpeg(args, { signal, timeout });
if (!result.success) {
throw new Error(
`SDR→HDR conversion failed (exit ${result.exitCode}): ${result.stderr.slice(-300)}`,
);
}
}
export async function extractAllVideoFrames(
videos: VideoElement[],
baseDir: string,
@@ -208,30 +267,75 @@ export async function extractAllVideoFrames(
const errors: Array<{ videoId: string; error: string }> = [];
let totalFramesExtracted = 0;
// Process videos in parallel for better performance
// Phase 1: Resolve paths and download remote videos
const resolvedVideos: Array<{ video: VideoElement; videoPath: string }> = [];
for (const video of videos) {
if (signal?.aborted) break;
try {
let videoPath = video.src;
if (!videoPath.startsWith("/") && !isHttpUrl(videoPath)) {
const fromCompiled = compiledDir ? join(compiledDir, videoPath) : null;
videoPath =
fromCompiled && existsSync(fromCompiled) ? fromCompiled : join(baseDir, videoPath);
}
if (isHttpUrl(videoPath)) {
const downloadDir = join(options.outputDir, "_downloads");
mkdirSync(downloadDir, { recursive: true });
videoPath = await downloadToTemp(videoPath, downloadDir);
}
if (!existsSync(videoPath)) {
errors.push({ videoId: video.id, error: `Video file not found: ${videoPath}` });
continue;
}
resolvedVideos.push({ video, videoPath });
} catch (err) {
errors.push({ videoId: video.id, error: err instanceof Error ? err.message : String(err) });
}
}
// Phase 2: Probe color spaces and normalize if mixed HDR/SDR
const videoColorSpaces = await Promise.all(
resolvedVideos.map(async ({ videoPath }) => {
const metadata = await extractVideoMetadata(videoPath);
return metadata.colorSpace;
}),
);
const hasAnyHdr = videoColorSpaces.some(isHdrColorSpaceUtil);
if (hasAnyHdr) {
const convertDir = join(options.outputDir, "_hdr_normalized");
mkdirSync(convertDir, { recursive: true });
for (let i = 0; i < resolvedVideos.length; i++) {
if (signal?.aborted) break;
const cs = videoColorSpaces[i] ?? null;
if (!isHdrColorSpaceUtil(cs)) {
// SDR video in a mixed timeline — convert to HDR color space
const entry = resolvedVideos[i];
if (!entry) continue;
const convertedPath = join(convertDir, `${entry.video.id}_hdr.mp4`);
try {
await convertSdrToHdr(entry.videoPath, convertedPath, signal, config);
entry.videoPath = convertedPath;
} catch (err) {
errors.push({
videoId: entry.video.id,
error: `SDR→HDR conversion failed: ${err instanceof Error ? err.message : String(err)}`,
});
}
}
}
}
// Phase 3: Extract frames (parallel)
const results = await Promise.all(
videos.map(async (video) => {
resolvedVideos.map(async ({ video, videoPath }) => {
if (signal?.aborted) {
throw new Error("Video frame extraction cancelled");
}
try {
let videoPath = video.src;
if (!videoPath.startsWith("/") && !isHttpUrl(videoPath)) {
const fromCompiled = compiledDir ? join(compiledDir, videoPath) : null;
videoPath =
fromCompiled && existsSync(fromCompiled) ? fromCompiled : join(baseDir, videoPath);
}
if (isHttpUrl(videoPath)) {
const downloadDir = join(options.outputDir, "_downloads");
mkdirSync(downloadDir, { recursive: true });
videoPath = await downloadToTemp(videoPath, downloadDir);
}
if (!existsSync(videoPath)) {
return { error: { videoId: video.id, error: `Video file not found: ${videoPath}` } };
}
let videoDuration = video.end - video.start;
// Fallback: if no data-duration/data-end was specified (end is Infinity or 0),
+22
View File
@@ -42,6 +42,15 @@ function parseProbeJson(stdout: string): FFProbeOutput {
const videoMetadataCache = new Map<string, Promise<VideoMetadata>>();
const audioMetadataCache = new Map<string, Promise<AudioMetadata>>();
export interface VideoColorSpace {
/** Color transfer characteristics, e.g. "bt709", "smpte2084", "arib-std-b67" */
colorTransfer: string;
/** Color primaries, e.g. "bt709", "bt2020" */
colorPrimaries: string;
/** Color matrix/space, e.g. "bt709", "bt2020nc" */
colorSpace: string;
}
export interface VideoMetadata {
durationSeconds: number;
width: number;
@@ -51,6 +60,8 @@ export interface VideoMetadata {
hasAudio: boolean;
/** True when r_frame_rate and avg_frame_rate differ significantly (>10%), indicating variable frame rate. */
isVFR: boolean;
/** Color space info from the video stream. Null if ffprobe didn't report it. */
colorSpace: VideoColorSpace | null;
}
export interface AudioMetadata {
@@ -70,6 +81,9 @@ interface FFProbeStream {
avg_frame_rate?: string;
sample_rate?: string;
channels?: number;
color_transfer?: string;
color_primaries?: string;
color_space?: string;
}
interface FFProbeFormat {
@@ -117,6 +131,11 @@ export async function extractVideoMetadata(filePath: string): Promise<VideoMetad
// VFR: r_frame_rate (max/nominal) differs from avg_frame_rate (actual average) by >10%
const isVFR = rFps > 0 && avgFps > 0 && Math.abs(rFps - avgFps) / Math.max(rFps, avgFps) > 0.1;
const colorTransfer = videoStream.color_transfer || "";
const colorPrimaries = videoStream.color_primaries || "";
const colorSpaceVal = videoStream.color_space || "";
const hasColorInfo = !!(colorTransfer || colorPrimaries || colorSpaceVal);
return {
durationSeconds: output.format.duration ? parseFloat(output.format.duration) : 0,
width: videoStream.width || 0,
@@ -125,6 +144,9 @@ export async function extractVideoMetadata(filePath: string): Promise<VideoMetad
videoCodec: videoStream.codec_name || "unknown",
hasAudio: output.streams.some((s) => s.codec_type === "audio"),
isVFR,
colorSpace: hasColorInfo
? { colorTransfer, colorPrimaries, colorSpace: colorSpaceVal }
: null,
};
})();
+191
View File
@@ -0,0 +1,191 @@
import { describe, expect, it } from "vitest";
import {
isHdrColorSpace,
detectTransfer,
getHdrEncoderColorParams,
analyzeCompositionHdr,
DEFAULT_HDR10_MASTERING,
} from "./hdr.js";
import type { VideoColorSpace } from "./ffprobe.js";
describe("isHdrColorSpace", () => {
it("returns false for null", () => {
expect(isHdrColorSpace(null)).toBe(false);
});
it("returns false for bt709 SDR", () => {
expect(
isHdrColorSpace({ colorTransfer: "bt709", colorPrimaries: "bt709", colorSpace: "bt709" }),
).toBe(false);
});
it("detects bt2020 primaries", () => {
expect(
isHdrColorSpace({ colorTransfer: "bt709", colorPrimaries: "bt2020", colorSpace: "bt709" }),
).toBe(true);
});
it("detects smpte2084 (PQ)", () => {
expect(
isHdrColorSpace({
colorTransfer: "smpte2084",
colorPrimaries: "bt2020",
colorSpace: "bt2020nc",
}),
).toBe(true);
});
it("detects arib-std-b67 (HLG)", () => {
expect(
isHdrColorSpace({
colorTransfer: "arib-std-b67",
colorPrimaries: "bt2020",
colorSpace: "bt2020nc",
}),
).toBe(true);
});
});
describe("detectTransfer", () => {
it("returns hlg for null", () => {
expect(detectTransfer(null)).toBe("hlg");
});
it("returns pq for smpte2084", () => {
expect(
detectTransfer({
colorTransfer: "smpte2084",
colorPrimaries: "bt2020",
colorSpace: "bt2020nc",
}),
).toBe("pq");
});
it("returns hlg for arib-std-b67", () => {
expect(
detectTransfer({
colorTransfer: "arib-std-b67",
colorPrimaries: "bt2020",
colorSpace: "bt2020nc",
}),
).toBe("hlg");
});
it("returns hlg for bt709 (fallback)", () => {
expect(
detectTransfer({ colorTransfer: "bt709", colorPrimaries: "bt709", colorSpace: "bt709" }),
).toBe("hlg");
});
});
describe("getHdrEncoderColorParams", () => {
it("returns PQ params with mastering metadata", () => {
const params = getHdrEncoderColorParams("pq");
expect(params.colorTrc).toBe("smpte2084");
expect(params.colorPrimaries).toBe("bt2020");
expect(params.colorspace).toBe("bt2020nc");
expect(params.pixelFormat).toBe("yuv420p10le");
expect(params.x265ColorParams).toContain("colorprim=bt2020");
expect(params.x265ColorParams).toContain("transfer=smpte2084");
expect(params.x265ColorParams).toContain("colormatrix=bt2020nc");
expect(params.mastering).toEqual(DEFAULT_HDR10_MASTERING);
});
it("returns HLG params with mastering metadata", () => {
const params = getHdrEncoderColorParams("hlg");
expect(params.colorTrc).toBe("arib-std-b67");
expect(params.colorPrimaries).toBe("bt2020");
expect(params.pixelFormat).toBe("yuv420p10le");
expect(params.x265ColorParams).toContain("transfer=arib-std-b67");
expect(params.mastering).toEqual(DEFAULT_HDR10_MASTERING);
});
// Regression guard for the side_data=[none] bug. See
// packages/producer/scripts/hdr-smoke.ts and the bug-1 entry in
// hdr-deferred-followups.md. Without master-display + max-cll in the
// x265-params, downstream players (Apple QuickTime, YouTube, HDR TVs) treat
// the file as SDR BT.2020 and tone-map incorrectly.
it("emits master-display and max-cll for PQ", () => {
const params = getHdrEncoderColorParams("pq");
expect(params.x265ColorParams).toContain(
`master-display=${DEFAULT_HDR10_MASTERING.masterDisplay}`,
);
expect(params.x265ColorParams).toContain(`max-cll=${DEFAULT_HDR10_MASTERING.maxCll}`);
});
it("emits master-display and max-cll for HLG", () => {
const params = getHdrEncoderColorParams("hlg");
expect(params.x265ColorParams).toContain(
`master-display=${DEFAULT_HDR10_MASTERING.masterDisplay}`,
);
expect(params.x265ColorParams).toContain(`max-cll=${DEFAULT_HDR10_MASTERING.maxCll}`);
});
it("respects an explicit mastering override", () => {
const custom = {
masterDisplay: "G(1,2)B(3,4)R(5,6)WP(7,8)L(9,10)",
maxCll: "500,200",
};
const params = getHdrEncoderColorParams("pq", custom);
expect(params.mastering).toBe(custom);
expect(params.x265ColorParams).toContain("master-display=G(1,2)B(3,4)R(5,6)WP(7,8)L(9,10)");
expect(params.x265ColorParams).toContain("max-cll=500,200");
});
// The DEFAULT_HDR10_MASTERING values are tagged as "P3-D65 inside BT.2020,
// 0.0001-1000 nits, MaxCLL 1000 / MaxFALL 400". If anyone tweaks these
// numbers without updating the docstring or the deferred-followups doc,
// this test will fail and force a deliberate review.
it("DEFAULT_HDR10_MASTERING matches the documented HDR10 reference", () => {
expect(DEFAULT_HDR10_MASTERING.masterDisplay).toBe(
"G(13250,34500)B(7500,3000)R(34000,16000)WP(15635,16450)L(10000000,1)",
);
expect(DEFAULT_HDR10_MASTERING.maxCll).toBe("1000,400");
});
});
describe("analyzeCompositionHdr", () => {
const sdr: VideoColorSpace = {
colorTransfer: "bt709",
colorPrimaries: "bt709",
colorSpace: "bt709",
};
const hlg: VideoColorSpace = {
colorTransfer: "arib-std-b67",
colorPrimaries: "bt2020",
colorSpace: "bt2020nc",
};
const pq: VideoColorSpace = {
colorTransfer: "smpte2084",
colorPrimaries: "bt2020",
colorSpace: "bt2020nc",
};
it("returns no HDR for all SDR", () => {
expect(analyzeCompositionHdr([sdr, sdr, null])).toEqual({
hasHdr: false,
dominantTransfer: null,
});
});
it("detects HLG", () => {
expect(analyzeCompositionHdr([sdr, hlg])).toEqual({
hasHdr: true,
dominantTransfer: "hlg",
});
});
it("detects PQ", () => {
expect(analyzeCompositionHdr([sdr, pq])).toEqual({
hasHdr: true,
dominantTransfer: "pq",
});
});
it("PQ takes priority over HLG in mixed HDR", () => {
expect(analyzeCompositionHdr([hlg, pq])).toEqual({
hasHdr: true,
dominantTransfer: "pq",
});
});
});
+137
View File
@@ -0,0 +1,137 @@
/**
* HDR Color Space Utilities
*
* Centralized HDR detection, transfer type handling, and FFmpeg color
* parameter generation for the HDR rendering pipeline.
*/
import type { VideoColorSpace } from "./ffprobe.js";
export type HdrTransfer = "hlg" | "pq";
/**
* Check if a video's color space indicates HDR content.
* Re-exported from videoFrameExtractor for backward compatibility.
*/
export function isHdrColorSpace(cs: VideoColorSpace | null): boolean {
if (!cs) return false;
return (
cs.colorPrimaries.includes("bt2020") ||
cs.colorSpace.includes("bt2020") ||
cs.colorTransfer === "smpte2084" ||
cs.colorTransfer === "arib-std-b67"
);
}
/**
* Determine the HDR transfer function from a video's color space metadata.
*
* IMPORTANT: Callers must gate on `isHdrColorSpace(cs)` first. This function
* assumes the input has already been classified as HDR and defaults ambiguous
* inputs to "hlg" calling it with an SDR color space silently returns "hlg",
* which is wrong for SDR.
*
* Returns "pq" for SMPTE 2084, "hlg" for ARIB STD-B67, defaults to "hlg".
*/
export function detectTransfer(cs: VideoColorSpace | null): HdrTransfer {
if (cs?.colorTransfer === "smpte2084") return "pq";
return "hlg";
}
/**
* HDR static metadata for the encoded stream.
*
* `masterDisplay` is the SMPTE ST 2086 mastering-display color volume string
* accepted by x265 (`G(Gx,Gy)B(Bx,By)R(Rx,Ry)WP(WPx,WPy)L(Lmax,Lmin)`).
* Chromaticity values are scaled by 50000 (0.00002 cd/m² per unit) and
* luminance values by 10000 (0.0001 cd/m² per unit).
*
* `maxCll` is the CTA-861.3 Content Light Level pair `MaxCLL,MaxFALL` in
* cd/m². Without these SEI messages, downstream players (Apple QuickTime,
* YouTube, HDR TVs) treat the stream as SDR BT.2020 and tone-map incorrectly
* see packages/producer/scripts/hdr-smoke.ts for the regression assertion.
*/
export interface HdrMasteringMetadata {
masterDisplay: string;
maxCll: string;
}
/**
* Default HDR10 mastering metadata: P3-D65 primaries inside a BT.2020
* container, mastered for 0.00011000 cd/m² with MaxCLL=1000, MaxFALL=400.
*
* These are conservative defaults that match how most HDR10 grading suites
* (Premiere, DaVinci Resolve) tag content when per-frame measured values
* aren't available. A future PR can plumb measured MaxCLL through `--hdr-opt`.
*/
export const DEFAULT_HDR10_MASTERING: HdrMasteringMetadata = {
masterDisplay: "G(13250,34500)B(7500,3000)R(34000,16000)WP(15635,16450)L(10000000,1)",
maxCll: "1000,400",
};
export interface HdrEncoderColorParams {
colorPrimaries: string;
colorTrc: string;
colorspace: string;
pixelFormat: string;
/**
* Full x265-params string including color tagging and HDR static metadata.
* Pass directly to `-x265-params` (concatenate with other options via `:`).
*/
x265ColorParams: string;
/** The mastering metadata that was baked into `x265ColorParams`. */
mastering: HdrMasteringMetadata;
}
/**
* Get FFmpeg encoder color parameters for a given HDR transfer function.
*
* The returned `x265ColorParams` includes both color tagging
* (`colorprim`/`transfer`/`colormatrix`) and HDR static metadata
* (`master-display`/`max-cll`). Without the static metadata the encoded
* stream is rejected as SDR by most HDR-aware players and CDNs.
*/
export function getHdrEncoderColorParams(
transfer: HdrTransfer,
mastering: HdrMasteringMetadata = DEFAULT_HDR10_MASTERING,
): HdrEncoderColorParams {
const colorTrc = transfer === "pq" ? "smpte2084" : "arib-std-b67";
const tagging = `colorprim=bt2020:transfer=${colorTrc}:colormatrix=bt2020nc`;
const metadata = `master-display=${mastering.masterDisplay}:max-cll=${mastering.maxCll}`;
return {
colorPrimaries: "bt2020",
colorTrc,
colorspace: "bt2020nc",
pixelFormat: "yuv420p10le",
x265ColorParams: `${tagging}:${metadata}`,
mastering,
};
}
export interface CompositionHdrInfo {
hasHdr: boolean;
dominantTransfer: HdrTransfer | null;
}
/**
* Analyze a set of video color spaces to determine if the composition
* contains HDR content and what the dominant transfer function is.
*/
export function analyzeCompositionHdr(
colorSpaces: Array<VideoColorSpace | null>,
): CompositionHdrInfo {
let hasPq = false;
let hasHdr = false;
for (const cs of colorSpaces) {
if (!isHdrColorSpace(cs)) continue;
hasHdr = true;
if (cs?.colorTransfer === "smpte2084") hasPq = true;
}
if (!hasHdr) return { hasHdr: false, dominantTransfer: null };
// PQ takes priority — it's the more common HDR10 format
const dominantTransfer: HdrTransfer = hasPq ? "pq" : "hlg";
return { hasHdr: true, dominantTransfer };
}
+2 -1
View File
@@ -11,7 +11,8 @@
"rootDir": "./src",
"declaration": true,
"declarationMap": true,
"sourceMap": true
"sourceMap": true,
"types": ["@webgpu/types"]
},
"include": ["src/**/*"],
"exclude": ["node_modules", "dist", "src/**/*.test.ts"]
+1
View File
@@ -74,6 +74,7 @@
"@fontsource/roboto": "^5.2.10",
"@fontsource/source-code-pro": "^5.2.7",
"@types/node": "^22.10.1",
"@webgpu/types": "^0.1.69",
"esbuild": "^0.27.2",
"tsx": "^4.7.0",
"typescript": "^5.7.2"
@@ -30,6 +30,7 @@ import {
createFrameLookupTable,
type VideoElement,
FrameLookupTable,
type HdrTransfer,
createCaptureSession,
initializeSession,
closeCaptureSession,
@@ -54,6 +55,8 @@ import {
spawnStreamingEncoder,
createFrameReorderBuffer,
type StreamingEncoder,
convertHdrFrameToRgb48le,
analyzeCompositionHdr,
} from "@hyperframes/engine";
import { join, dirname, resolve } from "path";
import { randomUUID } from "crypto";
@@ -111,18 +114,9 @@ export interface RenderConfig {
producerConfig?: EngineConfig;
/** Custom logger. Defaults to console-based defaultLogger. */
logger?: ProducerLogger;
/**
* Override CRF (Constant Rate Factor) for the video encoder.
* Lower values = higher quality / larger files. Range: 051 for H.264.
* When set, overrides the CRF from the quality preset.
* Mutually exclusive with `videoBitrate`.
*/
/** Override CRF for the video encoder. Mutually exclusive with `videoBitrate`. */
crf?: number;
/**
* Target video bitrate (e.g. "10M", "5000k").
* When set, uses bitrate-based encoding instead of CRF.
* Mutually exclusive with `crf`.
*/
/** Target video bitrate (e.g. "10M"). Mutually exclusive with `crf`. */
videoBitrate?: string;
}
@@ -730,9 +724,10 @@ export async function executeRenderJob(
let frameLookup: FrameLookupTable | null = null;
const compiledDir = join(workDir, "compiled");
let extractionResult: Awaited<ReturnType<typeof extractAllVideoFrames>> | null = null;
if (composition.videos.length > 0) {
const extractionResult = await extractAllVideoFrames(
extractionResult = await extractAllVideoFrames(
composition.videos,
projectDir,
{ fps: job.config.fps, outputDir: join(workDir, "video-frames") },
@@ -772,6 +767,28 @@ export async function executeRenderJob(
perfStages.videoExtractMs = Date.now() - stage2Start;
}
// ── HDR auto-detection ──────────────────────────────────────────────
// If any extracted video source has an HDR color space, the output
// automatically uses H.265 10-bit with the dominant transfer (PQ if any
// PQ source is present, otherwise HLG). No flag needed.
let effectiveHdr: { transfer: HdrTransfer } | undefined;
if (frameLookup) {
const colorSpaces = (extractionResult?.extracted ?? []).map((ext) => ext.metadata.colorSpace);
const info = analyzeCompositionHdr(colorSpaces);
if (info.hasHdr && info.dominantTransfer) {
effectiveHdr = { transfer: info.dominantTransfer };
}
}
if (effectiveHdr && outputFormat !== "mp4") {
log.info(`[Render] HDR source detected but format is ${outputFormat} — using SDR`);
effectiveHdr = undefined;
}
if (effectiveHdr) {
log.info(
`[Render] HDR source detected — output: ${effectiveHdr.transfer.toUpperCase()} (BT.2020, 10-bit H.265)`,
);
}
// ── Stage 3: Audio processing ───────────────────────────────────────
const stage3Start = Date.now();
updateJobStatus(job, "preprocessing", "Processing audio tracks", 20, onProgress);
@@ -829,275 +846,379 @@ export async function executeRenderJob(
const FORMAT_EXT: Record<string, string> = { mp4: ".mp4", webm: ".webm", mov: ".mov" };
const videoExt = FORMAT_EXT[outputFormat] ?? ".mp4";
const videoOnlyPath = join(workDir, `video-only${videoExt}`);
const preset = getEncoderPreset(job.config.quality, outputFormat);
// User-level CRF/bitrate overrides take precedence over presets.
const effectiveQuality = job.config.crf ?? preset.quality;
const effectiveBitrate = job.config.videoBitrate;
// Shared encoder options used by both streaming and chunk encode paths.
const baseEncoderOpts = {
fps: job.config.fps,
width,
height,
codec: preset.codec,
preset: preset.preset,
quality: effectiveQuality,
bitrate: effectiveBitrate,
pixelFormat: preset.pixelFormat,
useGpu: job.config.useGpu,
};
// Only use the HDR encoder preset when there's HDR video to pass through.
// For SDR-only compositions, --hdr is a no-op — H.265 10-bit causes browser
// color management issues (orange shift) with no quality benefit.
const hasHdrVideo = effectiveHdr && composition.videos.length > 0 && frameLookup;
const encoderHdr = hasHdrVideo ? effectiveHdr : undefined;
const preset = getEncoderPreset(job.config.quality, outputFormat, encoderHdr);
job.framesRendered = 0;
// Streaming encode mode: pipe frame buffers directly to FFmpeg stdin,
// skipping disk writes and the separate Stage 5 encode step.
let streamingEncoder: StreamingEncoder | null = null;
// ── HDR pass-through path ────────────────────────────────────────────
// When HDR output is requested AND the composition has HDR video sources,
// extract raw HLG frames and pass them directly to FFmpeg — no conversion.
// The HLG pixel values ARE the correct encoding. The output is tagged as
// HLG/BT.2020 so HDR displays render it correctly.
//
// No WebGPU or Chrome needed for this path — it's a pure FFmpeg pipeline.
// GSAP transforms are NOT applied (future work: WebGPU shader transforms).
if (hasHdrVideo) {
log.info("[Render] HDR pass-through: extracting native HLG frames from video sources");
if (enableStreamingEncode) {
streamingEncoder = await spawnStreamingEncoder(
const hdrEncoder = await spawnStreamingEncoder(
videoOnlyPath,
{
...baseEncoderOpts,
imageFormat: captureOptions.format || "jpeg",
fps: job.config.fps,
width,
height,
codec: preset.codec,
preset: preset.preset,
quality: preset.quality,
pixelFormat: preset.pixelFormat,
hdr: preset.hdr,
rawInputFormat: "rgb48le",
},
abortSignal,
{ ffmpegStreamingTimeout: 3_600_000 },
);
assertNotAborted();
}
if (enableStreamingEncode && streamingEncoder) {
// ── Streaming capture + encode (Stage 4 absorbs Stage 5) ──────────
const reorderBuffer = createFrameReorderBuffer(0, job.totalFrames!);
const currentEncoder = streamingEncoder;
const { execSync } = await import("child_process");
if (workerCount > 1) {
// Parallel capture → streaming encode
const tasks = distributeFrames(job.totalFrames, workerCount, workDir);
const onFrameBuffer = async (frameIndex: number, buffer: Buffer): Promise<void> => {
await reorderBuffer.waitForFrame(frameIndex);
currentEncoder.writeFrame(buffer);
reorderBuffer.advanceTo(frameIndex + 1);
};
await executeParallelCapture(
fileServer.url,
workDir,
tasks,
captureOptions,
() => createVideoFrameInjector(frameLookup),
abortSignal,
(progress) => {
job.framesRendered = progress.capturedFrames;
const frameProgress = progress.capturedFrames / progress.totalFrames;
const progressPct = 25 + frameProgress * 55;
if (
progress.capturedFrames % 30 === 0 ||
progress.capturedFrames === progress.totalFrames
) {
updateJobStatus(
job,
"rendering",
`Streaming frame ${progress.capturedFrames}/${progress.totalFrames} (${workerCount} workers)`,
Math.round(progressPct),
onProgress,
);
}
},
onFrameBuffer,
cfg,
);
if (probeSession) {
lastBrowserConsole = probeSession.browserConsoleBuffer;
await closeCaptureSession(probeSession);
probeSession = null;
}
} else {
// Sequential capture → streaming encode
const videoInjector = createVideoFrameInjector(frameLookup);
const session =
probeSession ??
(await createCaptureSession(
fileServer.url,
framesDir,
captureOptions,
videoInjector,
cfg,
));
if (probeSession) {
prepareCaptureSessionForReuse(session, framesDir, videoInjector);
probeSession = null;
}
try {
if (!session.isInitialized) {
await initializeSession(session);
}
try {
for (let i = 0; i < job.totalFrames!; i++) {
assertNotAborted();
lastBrowserConsole = session.browserConsoleBuffer;
const time = i / job.config.fps;
for (let i = 0; i < job.totalFrames!; i++) {
assertNotAborted();
const time = i / job.config.fps;
const { buffer } = await captureFrameToBuffer(session, i, time);
await reorderBuffer.waitForFrame(i);
currentEncoder.writeFrame(buffer);
reorderBuffer.advanceTo(i + 1);
job.framesRendered = i + 1;
const activeFrames = frameLookup!.getActiveFramePayloads(time);
if (activeFrames.size > 0) {
const [videoId] = activeFrames.keys();
const video = composition.videos.find((v) => v.id === videoId);
if (video) {
const localTime = time - video.start + video.mediaStart;
let srcPath = video.src;
const compiledDir = join(workDir, "compiled");
if (!srcPath.startsWith("/")) {
const fromCompiled = join(compiledDir, srcPath);
srcPath = existsSync(fromCompiled) ? fromCompiled : join(projectDir, srcPath);
}
let rawFrame: Buffer;
try {
rawFrame = execSync(
`ffmpeg -ss ${localTime} -i "${srcPath}" -vframes 1 -f rawvideo -pix_fmt rgba64le -`,
{ maxBuffer: width * height * 8 + 1024 * 1024, stdio: ["pipe", "pipe", "pipe"] },
);
} catch {
rawFrame = Buffer.alloc(width * height * 8);
}
// Pass through HLG pixels as-is (RGBA → RGB, no color conversion)
const rgb48Frame = convertHdrFrameToRgb48le(rawFrame, width, height);
hdrEncoder.writeFrame(rgb48Frame);
} else {
hdrEncoder.writeFrame(Buffer.alloc(width * height * 6));
}
} else {
hdrEncoder.writeFrame(Buffer.alloc(width * height * 6));
}
job.framesRendered = i + 1;
if ((i + 1) % 10 === 0 || i + 1 === job.totalFrames!) {
const frameProgress = (i + 1) / job.totalFrames!;
const progress = 25 + frameProgress * 55;
updateJobStatus(
job,
"rendering",
`Streaming frame ${i + 1}/${job.totalFrames}`,
Math.round(progress),
`HDR frame ${i + 1}/${job.totalFrames}`,
Math.round(25 + frameProgress * 55),
onProgress,
);
}
} finally {
lastBrowserConsole = session.browserConsoleBuffer;
await closeCaptureSession(session);
}
} finally {
// No browser to close — pure FFmpeg path
}
// Close encoder and get result
const encodeResult = await currentEncoder.close();
const hdrEncodeResult = await hdrEncoder.close();
assertNotAborted();
if (!encodeResult.success) {
throw new Error(`Streaming encode failed: ${encodeResult.error}`);
if (!hdrEncodeResult.success) {
throw new Error(`HDR encode failed: ${hdrEncodeResult.error}`);
}
perfStages.captureMs = Date.now() - stage4Start;
perfStages.encodeMs = encodeResult.durationMs; // Overlapped with capture
} else {
// ── Disk-based capture (original flow) ────────────────────────────
if (workerCount > 1) {
// Parallel capture
const tasks = distributeFrames(job.totalFrames, workerCount, workDir);
perfStages.encodeMs = hdrEncodeResult.durationMs;
} else // ── Standard capture paths (SDR or DOM-only HDR) ──────────────────
// Streaming encode mode: pipe frame buffers directly to FFmpeg stdin,
// skipping disk writes and the separate Stage 5 encode step.
{
let streamingEncoder: StreamingEncoder | null = null;
await executeParallelCapture(
fileServer.url,
workDir,
tasks,
captureOptions,
() => createVideoFrameInjector(frameLookup),
if (enableStreamingEncode) {
streamingEncoder = await spawnStreamingEncoder(
videoOnlyPath,
{
fps: job.config.fps,
width,
height,
codec: preset.codec,
preset: preset.preset,
quality: preset.quality,
pixelFormat: preset.pixelFormat,
useGpu: job.config.useGpu,
imageFormat: captureOptions.format || "jpeg",
hdr: preset.hdr,
},
abortSignal,
(progress) => {
job.framesRendered = progress.capturedFrames;
const frameProgress = progress.capturedFrames / progress.totalFrames;
const progressPct = 25 + frameProgress * 45;
);
assertNotAborted();
}
if (enableStreamingEncode && streamingEncoder) {
// ── Streaming capture + encode (Stage 4 absorbs Stage 5) ──────────
const reorderBuffer = createFrameReorderBuffer(0, job.totalFrames!);
const currentEncoder = streamingEncoder;
if (workerCount > 1) {
// Parallel capture → streaming encode
const tasks = distributeFrames(job.totalFrames, workerCount, workDir);
const onFrameBuffer = async (frameIndex: number, buffer: Buffer): Promise<void> => {
await reorderBuffer.waitForFrame(frameIndex);
currentEncoder.writeFrame(buffer);
reorderBuffer.advanceTo(frameIndex + 1);
};
await executeParallelCapture(
fileServer.url,
workDir,
tasks,
captureOptions,
() => createVideoFrameInjector(frameLookup),
abortSignal,
(progress) => {
job.framesRendered = progress.capturedFrames;
const frameProgress = progress.capturedFrames / progress.totalFrames;
const progressPct = 25 + frameProgress * 55;
if (
progress.capturedFrames % 30 === 0 ||
progress.capturedFrames === progress.totalFrames
) {
updateJobStatus(
job,
"rendering",
`Streaming frame ${progress.capturedFrames}/${progress.totalFrames} (${workerCount} workers)`,
Math.round(progressPct),
onProgress,
);
}
},
onFrameBuffer,
cfg,
);
if (probeSession) {
lastBrowserConsole = probeSession.browserConsoleBuffer;
await closeCaptureSession(probeSession);
probeSession = null;
}
} else {
// Sequential capture → streaming encode
const videoInjector = createVideoFrameInjector(frameLookup);
const session =
probeSession ??
(await createCaptureSession(
fileServer.url,
framesDir,
captureOptions,
videoInjector,
cfg,
));
if (probeSession) {
prepareCaptureSessionForReuse(session, framesDir, videoInjector);
probeSession = null;
}
try {
if (!session.isInitialized) {
await initializeSession(session);
}
assertNotAborted();
lastBrowserConsole = session.browserConsoleBuffer;
for (let i = 0; i < job.totalFrames!; i++) {
assertNotAborted();
const time = i / job.config.fps;
const { buffer } = await captureFrameToBuffer(session, i, time);
await reorderBuffer.waitForFrame(i);
currentEncoder.writeFrame(buffer);
reorderBuffer.advanceTo(i + 1);
job.framesRendered = i + 1;
const frameProgress = (i + 1) / job.totalFrames!;
const progress = 25 + frameProgress * 55;
if (
progress.capturedFrames % 30 === 0 ||
progress.capturedFrames === progress.totalFrames
) {
updateJobStatus(
job,
"rendering",
`Capturing frame ${progress.capturedFrames}/${progress.totalFrames} (${workerCount} workers)`,
Math.round(progressPct),
`Streaming frame ${i + 1}/${job.totalFrames}`,
Math.round(progress),
onProgress,
);
}
},
undefined,
cfg,
);
await mergeWorkerFrames(workDir, tasks, framesDir);
if (probeSession) {
lastBrowserConsole = probeSession.browserConsoleBuffer;
await closeCaptureSession(probeSession);
probeSession = null;
} finally {
lastBrowserConsole = session.browserConsoleBuffer;
await closeCaptureSession(session);
}
}
// Close encoder and get result
const encodeResult = await currentEncoder.close();
assertNotAborted();
if (!encodeResult.success) {
throw new Error(`Streaming encode failed: ${encodeResult.error}`);
}
perfStages.captureMs = Date.now() - stage4Start;
perfStages.encodeMs = encodeResult.durationMs; // Overlapped with capture
} else {
// Sequential capture
// ── Disk-based capture (original flow) ────────────────────────────
if (workerCount > 1) {
// Parallel capture
const tasks = distributeFrames(job.totalFrames, workerCount, workDir);
const videoInjector = createVideoFrameInjector(frameLookup);
const session =
probeSession ??
(await createCaptureSession(
await executeParallelCapture(
fileServer.url,
framesDir,
workDir,
tasks,
captureOptions,
videoInjector,
() => createVideoFrameInjector(frameLookup),
abortSignal,
(progress) => {
job.framesRendered = progress.capturedFrames;
const frameProgress = progress.capturedFrames / progress.totalFrames;
const progressPct = 25 + frameProgress * 45;
if (
progress.capturedFrames % 30 === 0 ||
progress.capturedFrames === progress.totalFrames
) {
updateJobStatus(
job,
"rendering",
`Capturing frame ${progress.capturedFrames}/${progress.totalFrames} (${workerCount} workers)`,
Math.round(progressPct),
onProgress,
);
}
},
undefined,
cfg,
));
if (probeSession) {
prepareCaptureSessionForReuse(session, framesDir, videoInjector);
probeSession = null;
}
try {
if (!session.isInitialized) {
await initializeSession(session);
}
assertNotAborted();
lastBrowserConsole = session.browserConsoleBuffer;
for (let i = 0; i < job.totalFrames; i++) {
assertNotAborted();
const time = i / job.config.fps;
await captureFrame(session, i, time);
job.framesRendered = i + 1;
const frameProgress = (i + 1) / job.totalFrames;
const progress = 25 + frameProgress * 45;
updateJobStatus(
job,
"rendering",
`Capturing frame ${i + 1}/${job.totalFrames}`,
Math.round(progress),
onProgress,
);
}
} finally {
lastBrowserConsole = session.browserConsoleBuffer;
await closeCaptureSession(session);
}
}
perfStages.captureMs = Date.now() - stage4Start;
// ── Stage 5: Encode ─────────────────────────────────────────────────
const stage5Start = Date.now();
updateJobStatus(job, "encoding", "Encoding video", 75, onProgress);
const frameExt = needsAlpha ? "png" : "jpg";
const framePattern = `frame_%06d.${frameExt}`;
const encoderOpts = baseEncoderOpts;
const encodeResult = enableChunkedEncode
? await encodeFramesChunkedConcat(
framesDir,
framePattern,
videoOnlyPath,
encoderOpts,
chunkedEncodeSize,
abortSignal,
)
: await encodeFramesFromDir(
framesDir,
framePattern,
videoOnlyPath,
encoderOpts,
abortSignal,
);
assertNotAborted();
if (!encodeResult.success) {
throw new Error(`Encoding failed: ${encodeResult.error}`);
await mergeWorkerFrames(workDir, tasks, framesDir);
if (probeSession) {
lastBrowserConsole = probeSession.browserConsoleBuffer;
await closeCaptureSession(probeSession);
probeSession = null;
}
} else {
// Sequential capture
const videoInjector = createVideoFrameInjector(frameLookup);
const session =
probeSession ??
(await createCaptureSession(
fileServer.url,
framesDir,
captureOptions,
videoInjector,
cfg,
));
if (probeSession) {
prepareCaptureSessionForReuse(session, framesDir, videoInjector);
probeSession = null;
}
try {
if (!session.isInitialized) {
await initializeSession(session);
}
assertNotAborted();
lastBrowserConsole = session.browserConsoleBuffer;
for (let i = 0; i < job.totalFrames; i++) {
assertNotAborted();
const time = i / job.config.fps;
await captureFrame(session, i, time);
job.framesRendered = i + 1;
const frameProgress = (i + 1) / job.totalFrames;
const progress = 25 + frameProgress * 45;
updateJobStatus(
job,
"rendering",
`Capturing frame ${i + 1}/${job.totalFrames}`,
Math.round(progress),
onProgress,
);
}
} finally {
lastBrowserConsole = session.browserConsoleBuffer;
await closeCaptureSession(session);
}
}
perfStages.captureMs = Date.now() - stage4Start;
// ── Stage 5: Encode ─────────────────────────────────────────────────
const stage5Start = Date.now();
updateJobStatus(job, "encoding", "Encoding video", 75, onProgress);
const frameExt = needsAlpha ? "png" : "jpg";
const framePattern = `frame_%06d.${frameExt}`;
const encoderOpts = {
fps: job.config.fps,
width,
height,
codec: preset.codec,
preset: preset.preset,
quality: preset.quality,
pixelFormat: preset.pixelFormat,
useGpu: job.config.useGpu,
hdr: preset.hdr,
};
const encodeResult = enableChunkedEncode
? await encodeFramesChunkedConcat(
framesDir,
framePattern,
videoOnlyPath,
encoderOpts,
chunkedEncodeSize,
abortSignal,
)
: await encodeFramesFromDir(
framesDir,
framePattern,
videoOnlyPath,
encoderOpts,
abortSignal,
);
assertNotAborted();
if (!encodeResult.success) {
throw new Error(`Encoding failed: ${encodeResult.error}`);
}
perfStages.encodeMs = Date.now() - stage5Start;
}
perfStages.encodeMs = Date.now() - stage5Start;
}
} // end SDR capture paths block
if (probeSession !== null) {
const remainingProbeSession: CaptureSession = probeSession;
+2 -1
View File
@@ -11,7 +11,8 @@
"rootDir": "./src",
"declaration": true,
"declarationMap": true,
"sourceMap": true
"sourceMap": true,
"types": ["@webgpu/types"]
},
"include": ["src/**/*"],
"exclude": ["node_modules", "dist", "src/**/*.test.ts"]