Multi-channel preview acceptance is insufficient evidence of recording acceptance at the same channel count. Preview and recording may use different output configurations, encoding, retention policies and resources. Check actual recording files, indexes, time coverage and playback against recording objectives, rather than visible images alone.
What preview and recording checks cover
| Scope | Checks |
|---|---|
| Preview | Valid images, continuing updates, configuration and client states |
| Recording | File existence, valid content, time coverage and write errors |
| Indexes and playback | Queries, seeking, readability and corresponding segments |
| Storage | Disk resources, retention, insufficient space and alerts |
| Recovery | Input loss, restarts, abnormal termination and subsequent output |
| Combined workload | Preview, inference, recording and playback running together |
Do segment files establish recording completeness
File existence alone is insufficient evidence that content, time range and indexes meet requirements. Check playable content, unexpected gaps, retention coverage and playback against agreed conditions. Sampling a few segments does not automatically verify the full target scope.
Does processor encoding capability remove the need for testing
Specifications alone are insufficient evidence that the application uses the hardware path. Check the actual program, drivers, configuration and encoded output, then examine resources and results under target load. Record software transcoding, hardware encoding and model-inference backends separately.
Wording when evidence is incomplete
Specify verified functions, channel counts, configurations and duration. List outstanding multi-channel recording and concurrent-combination tests. Without those tests, a preview demonstration is insufficient evidence of recording acceptance.
Related services: edge AI deployment and optimization and project enquiries.