The problem was indeed with Node.js/Chromium implementation.
While trying to use hardware acceleration in headless mode with Node.js (through Puppeteer/Playwright), Chromium consistently fell back to software rendering despite various launch parameters and configurations.
The solution was to switch to Python implementation of Playwright, which handles GPU acceleration more effectively on Raspberry Pi OS. Python's implementation works better with the underlying graphics stack and Xvfb virtual framebuffer, allowing proper hardware acceleration even in headless mode.
This explains why you were getting those GL (OpenGL) implementation errors with Node.js version, but achieved successful hardware-accelerated screenshots using Python + Playwright + Xvfb combination.
The Python ecosystem appears to have better compatibility with Raspberry Pi's GPU architecture in this specific use case.
While trying to use hardware acceleration in headless mode with Node.js (through Puppeteer/Playwright), Chromium consistently fell back to software rendering despite various launch parameters and configurations.
The solution was to switch to Python implementation of Playwright, which handles GPU acceleration more effectively on Raspberry Pi OS. Python's implementation works better with the underlying graphics stack and Xvfb virtual framebuffer, allowing proper hardware acceleration even in headless mode.
This explains why you were getting those GL (OpenGL) implementation errors with Node.js version, but achieved successful hardware-accelerated screenshots using Python + Playwright + Xvfb combination.
The Python ecosystem appears to have better compatibility with Raspberry Pi's GPU architecture in this specific use case.
Statistics: Posted by sns8 — Mon Oct 28, 2024 9:00 am