From physical action to displayed photon
Glass-to-glass measurement spans controller input, game processing, rendering, capture, encoding, network, jitter handling, decoding, presentation and the physical display.
Input
Controller polling, OS and game response.
Media chain
Render, capture, encode, transport, decode and presentation.
Display
Refresh timing, TV processing and panel response.
RTT + decode is not total latency
RTT is a packet round trip at a measured layer. Video is a one-way path with its own queues. Decode excludes capture, encode and display refresh. Therefore 11 ms RTT plus 0.9 ms decode cannot honestly be labeled 11.9 ms end-to-end.
High-speed camera or synchronized sensors, with repeated trials
- Create a clear eventMap a button to an unambiguous visual change.
- Observe input and displayUse one high-speed frame or calibrated synchronized sensors.
- RepeatPublish median, percentiles and spread—not one best result.
- Document the chainGame, GPU, codec, route, client and display mode.
- Keep incidentsHandle freezes and outliers by a predefined rule.
The number belongs to a complete chain
A TV’s image processing, 60 versus 120 Hz, relay routing or V-Sync can materially change the distribution. A number without those settings is hard to compare.
Publish only what the test supports
Until the full protocol is run, Lume publishes RTT, decode, frame rate, bitrate, freezes, duration and route type separately. A future glass-to-glass campaign should include conditions, sequences and distributions.