feat(video): allow recording video at a custom fps by pavelfeldman · Pull Request #42752 · microsoft/playwright

Pavel Feldman (pavelfeldman) · 2026-09-17T00:47:21Z

Summary

New fps option for page.screencast.start(), recordVideo (recordVideoFps in other languages), test runner video and CLI/MCP video-start.
VP8 bitrate and encoder threads scale with pixel rate relative to 800x450@25fps; default recordings keep -b:v 1M -threads 1.
Screencast frames are locked to consecutive output slots, since Chromium timestamps frames on delivery rather than render (https://issues.chromium.org/issues/562567707). Without it, 1080p60 recordings duplicated and skipped ~4% of frames.
Firefox and WebKit still capture at most 25 fps.

Yury Semikhatsky (yury-s) · 2026-09-17T00:57:52Z

+    //   1920x1080 at 60fps -> scale 13.8, bitrate 13824k, 4 threads
+    const pixelRateScale = Math.max(1, w * h * this._fps / (800 * 450 * kDefaultFps));
+    const bitrate = Math.round(pixelRateScale * 1000);
+    const threads = Math.min(8, Math.ceil(pixelRateScale / 4));


let's trust ffmpeg's auto mode in case we allow >1 thread?

"-threads 0 or -threads auto means “choose automatically”; the generic codec option defaults to auto. Current ffmpeg also explicitly sets thread_count = 0 when you haven’t supplied a threads option."

- `fps` option for `page.screencast.start()`, `recordVideo`, test runner `video` and `video-start` - scale vp8 bitrate and encoder threads with the pixel rate - lock screencast frames to consecutive output slots to absorb Chromium timestamp jitter (https://issues.chromium.org/issues/562567707)

Devin Rousso (dcrousso) · 2026-09-17T16:27:38Z

        width: z.number().describe('Video width'),
        height: z.number().describe('Video height'),
      }).optional().describe('Video size'),
+      fps: z.number().optional().describe('Video frame rate in frames per second, defaults to 25'),


should this be z.number().int().positive().optional()?

github-actions · 2026-09-17T16:32:34Z

Test results for "MCP"

3 failed
❌ [chrome] › mcp/annotate.spec.ts:173 › user-initiated annotate downloads zip with feedback.md @mcp-windows-latest-chrome
❌ [firefox] › mcp/annotate.spec.ts:417 › should cancel browser_annotate when the MCP client disconnects @mcp-windows-latest-firefox
❌ [firefox] › mcp/cli-session.spec.ts:54 › idle timeout shuts the session down @mcp-windows-latest-firefox

8587 passed, 1446 skipped

Merge workflow run.

github-actions · 2026-09-17T16:34:39Z

Test results for "tests 1"

2 failed
❌ [installation tests] › playwright-cdn.spec.ts:43 › playwright cdn failover should work (https://cdn.playwright.dev/dbazure/download/playwright) @package-installations-macos-latest
❌ [installation tests] › playwright-cdn.spec.ts:43 › playwright cdn failover should work (https://playwright.download.prss.microsoft.com/dbazure/download/playwright) @package-installations-macos-latest

5 flaky

⚠️ [chromium-library] › library/video.spec.ts:725 › screencast › should capture full viewport on hidpi `@chromium-ubuntu-22.04-arm-node20`
⚠️ [firefox-library] › library/browsercontext-cookies-third-party.spec.ts:257 › third party 'Partitioned;' cookies `@firefox-ubuntu-22.04-node20`
⚠️ [firefox-library] › library/browsercontext-cookies-third-party.spec.ts:470 › top level 'Partitioned;' cookie and same origin iframe `@firefox-ubuntu-22.04-node20`
⚠️ [firefox-page] › page/page-emulate-media.spec.ts:144 › should keep reduced motion and color emulation after reload `@firefox-ubuntu-22.04-node20`
⚠️ [firefox-page] › page/page-event-request.spec.ts:181 › should return response body when Cross-Origin-Opener-Policy is set `@firefox-ubuntu-22.04-node20`

51973 passed, 1247 skipped

Merge workflow run.

github-actions · 2026-09-17T16:37:27Z

🟢 All three failures are pre-existing flakes — the PR is clear

Hi, I'm the Playwright bot and I took a look at the failing CI on this PR.

All three failures are Windows-only MCP tests that each fail with the same error on SHAs and PRs unrelated to this one. Nothing here points at the fps change.

Details

Overall: the PR touches video/screencast encoding (videoRecorder.ts, screencast.ts, protocol fps plumbing). The three failures are MCP annotate/session tests that don't record video, and each has a failure history predating this PR. The only video-adjacent signal in this PR's CI, library/video.spec.ts:690 › screencast › should capture full viewport, was flaky (passed on retry), not failed.

Pre-existing flake / infra

[chrome] › mcp/annotate.spec.ts:173 › user-initiated annotate downloads zip with feedback.md — fails with Cannot read properties of undefined (reading 'endpoint'). Same failure, same message, on unrelated work: run 32395156145 (push, 66bfad7), run 33656387949 (chore: bump WebKit version to 26.6 #42514), and run 35230562603 (feat(mcp): offer page-registered WebMCP tools as MCP tools #42671, mcp-windows-latest-chromium, hours before this PR's run). 4 failures in 727 runs on chrome, green otherwise.
[firefox] › mcp/annotate.spec.ts:417 › should cancel browser_annotate when the MCP client disconnects — fails on expect(locator).toBeVisible() for the main role. Same failure on Windows across chrome/msedge/firefox and across unrelated PRs: run 33089818898 (feat(reporter): introduce TestStep.subtitle, unify step formatting #42422), run 34890098331 (fix(reporter): fail perfetto writes when the output stream errors #42715), run 34349086953 (feat(firefox): roll to r1544 #42631), plus push SHAs 2ea99e1 and 14f3c0c. 11 failures across ~2900 Windows runs.
[firefox] › mcp/cli-session.spec.ts:54 › idle timeout shuts the session down — the noisiest of the three: 12 of 71 runs (17%) fail on mcp-windows-latest-firefox, always expect(received).toContain("(no…"), and never on any other browser (0/74 chromium, 0/74 chrome, 0/74 webkit, 0/70 msedge). Recent unrelated failures: run 35195191777 (push f1ea64b), run 35145363072 (feat(coverage): experimental istanbul coverage collection via tracing #42684), run 35070023943 (test: evaluate() with falsy result #42728), run 35069966233 (fix(firefox): use bypassServiceWorker flag for storage state pages #42741). This one looks worth a separate look on its own merits — it's been failing roughly every sixth run since 2026-09-11.

Evidence comes from the aggregated CI results database, topped up through this PR's run.

_{Triaged by the Playwright bot - agent run}

github-actions · 2026-09-17T16:38:15Z

🟢 All five failures are pre-existing flakes — the PR looks clear

Hi, I'm the Playwright bot and I triaged the red CI here. None of the failing tests touch video, screencast, or fps; each one fails with the same error on other PRs and on main, one of them on this PR's own base commit bf9d04f2.

Details

The PR changes video/screencast recording (videoRecorder.ts, browserContext.ts, pageDispatcher.ts, tools/backend/video.ts, cli-daemon/commands.ts) plus docs and generated types. The failures are in MCP annotate, MCP CLI session listing, and the installation CDN suite — none of which record video. Counts below are from the aggregated CI results DB, grouped by (project_name, file, test_title) on the same bot.

Pre-existing flake / infra

[chrome] › mcp/annotate.spec.ts:173 › user-initiated annotate downloads zip with feedback.md — Cannot read properties of undefined (reading 'endpoint') then a test timeout. Same failure on chore(deps-dev): bump browserslist from 4.28.4 to 4.28.8 #42489 (80fb1e9b), chore: bump WebKit version to 26.6 #42514 (9f7700c7), feat(mcp): offer page-registered WebMCP tools as MCP tools #42671 (789b6fc8, as chromium) and on main (66bfad76). 4 of 712 runs on mcp-windows-latest-chrome.
[firefox] › mcp/annotate.spec.ts:417 › should cancel browser_annotate when the MCP client disconnects — expect(locator).toBeVisible() on the Dashboard: annotate region. The busiest of the three: 7 of 698 runs on mcp-windows-latest-firefox, and it fails on chrome/chromium/msedge Windows too. Recent unrelated hits include chore(trace): split actions and context options into separate files #42654, feat(mcp): add --timeout-idle to close the browser after inactivity #42663, fix(reporter): fail perfetto writes when the output stream errors #42715, fix(webkit): do not report blob: URLs to route handlers #42738, plus several main pushes (bae3292f, 118ee440, d4b2db46).
[firefox] › mcp/cli-session.spec.ts:54 › idle timeout shuts the session down — expect(received).toContain("(no browsers)"). This one is badly broken independently of any PR: 24 of 68 runs fail (35%) on mcp-windows-latest-firefox, spread over 12 different PRs (feat(mcp): offer page-registered WebMCP tools as MCP tools #42671, feat(coverage): experimental istanbul coverage collection via tracing #42684 ×3, fix(mcp): do not close browser while tool call is running #42696, test: roll stable-test-runner to 1.64.0-alpha-2026-09-14 #42706, test: fix click tests on android #42726, test: evaluate() with falsy result #42728, feat(firefox): roll to r1546 #42735, fix(firefox): use bypassServiceWorker flag for storage state pages #42741, fix(agents): add missing parentheses in test-generator example #42744) and 11 main pushes. The PR does touch cli-daemon/commands.ts, so this was the one worth a second look — but the failure rate is unchanged and predates the branch.
[installation tests] › playwright-cdn.spec.ts:43 › playwright cdn failover should work (×2 hosts) — npx playwright install exits 1 on package-installations-macos-latest. Infra: both host variants failed together on feat(webkit): roll to r2362 #42736, feat(webkit): roll to r2363 #42737, feat(webkit): roll to r2364 #42748, feat(chromium): roll to r1246 #42757, and on main at bf9d04f2 — which is exactly the base commit this PR branched from.

Nothing to fix here; a rerun should be green, modulo the cli-session flake's 1-in-3 odds.

_{Triaged by the Playwright bot - agent run}

Yury Semikhatsky (yury-s) reviewed Sep 17, 2026

View reviewed changes

Yury Semikhatsky (yury-s) approved these changes Sep 17, 2026

View reviewed changes

This comment has been minimized.

Sign in to view

Pavel Feldman (pavelfeldman) force-pushed the video-fps branch from 19a6e22 to eb5e436 Compare September 17, 2026 15:51

Devin Rousso (dcrousso) approved these changes Sep 17, 2026

View reviewed changes

github-actions Bot mentioned this pull request Sep 17, 2026

feat(chromium): roll to r1246 #42757

Merged

Pavel Feldman (pavelfeldman) merged commit 8bb2c09 into microsoft:main Sep 17, 2026
43 of 46 checks passed

Navigation Menu

feat(video): allow recording video at a custom fps - #42752

feat(video): allow recording video at a custom fps#42752
Pavel Feldman (pavelfeldman) merged 1 commit into
microsoft:mainfrom
pavelfeldman:video-fps

Pavel Feldman (pavelfeldman) commented Sep 17, 2026

Uh oh!

Yury Semikhatsky (yury-s) Sep 17, 2026

Uh oh!

This comment has been minimized.

This comment has been minimized.

Devin Rousso (dcrousso) Sep 17, 2026

Uh oh!

github-actions Bot commented Sep 17, 2026

Uh oh!

github-actions Bot commented Sep 17, 2026

Uh oh!

github-actions Bot commented Sep 17, 2026

Uh oh!

github-actions Bot commented Sep 17, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

Pavel Feldman (pavelfeldman) commented Sep 17, 2026

Summary

Uh oh!

Yury Semikhatsky (yury-s) Sep 17, 2026

Choose a reason for hiding this comment

Uh oh!

This comment has been minimized.

This comment has been minimized.

Devin Rousso (dcrousso) Sep 17, 2026

Choose a reason for hiding this comment

Uh oh!

github-actions Bot commented Sep 17, 2026

Test results for "MCP"

Uh oh!

github-actions Bot commented Sep 17, 2026

Test results for "tests 1"

Uh oh!

github-actions Bot commented Sep 17, 2026

🟢 All three failures are pre-existing flakes — the PR is clear

Uh oh!

github-actions Bot commented Sep 17, 2026

🟢 All five failures are pre-existing flakes — the PR looks clear

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants