[Feature]: Add size option to page.screencast.start()

[laizezhong]

Feature request

Please consider exposing the size option on Python's page.screencast.start() API, matching the Node.js Playwright API.

In Node.js, page.screencast.start() supports controlling the output frame/video size:

await page.screencast.start({
  onFrame: ({ data }) => sendToVisionModel(data),
  size: { width: 800, height: 600 },
});

This is useful for real-time frame capture, thumbnails, live previews, AI vision pipelines, and controlling video receipt resolution independently from the page viewport.

Current Python behavior

In Python, the async API currently exposes:

await page.screencast.start(
    on_frame=...,
    path="video.webm",
    quality=100,
)

There is no size parameter.

I verified this in both Playwright Python 1.59.0 and 1.60.0 by inspecting the generated async API signature. In 1.60.0, the generated signature is still:

async def start(
    self,
    *,
    on_frame: Optional[Callable[[ScreencastFrame], Any]] = None,
    path: Optional[Union[pathlib.Path, str]] = None,
    quality: Optional[int] = None,
) -> AsyncContextManager:

Why viewport is not enough

Changing the viewport affects page layout, but it does not give Python users a way to control screencast output resolution.

For example, with a 1920x1080 viewport, the page layout is rendered at 1920x1080, but page.screencast.start(path="video.webm") still appears to produce a video/frame stream scaled down to the default 800x800 bounding behavior, typically around 800x450 for a 16:9 viewport.

For many automation/agent workflows, this is the wrong tradeoff: we may want to keep the page viewport at 1920x1080 for desktop layout fidelity, while requesting lower-resolution frames for a vision model, or higher-resolution video receipts for human review.

Expected API

Something like this would provide parity with Node.js:

await page.screencast.start(
    on_frame=lambda frame: send_to_vision_model(frame["data"]),
    size={"width": 800, "height": 600},
    quality=80,
)

and/or:

await page.screencast.start(
    path="video.webm",
    size={"width": 1920, "height": 1080},
    quality=100,
)

Use case

I am using Playwright Python for browser automation with agentic video receipts and page-level evidence collection. page.screencast is a much better fit than context-level video recording because it is page-scoped and works well with multi-tab/shared-context workflows. However, without size, Python users cannot control the output resolution the way Node.js users can.

Thanks for considering this API parity improvement.