Choose your video quality workflow

Bring

Representative source clips, candidate encodes, quality floors, and compute or delivery limits.

Measure

Perceptual score, VMAF, PSNR, SSIM, frame coverage, and separately recorded bitrate and compute cost.

1. 01

   Define the policy

   Set quality floors, scale limits, content fixtures, and the encoder search space.

2. 02

   Generate candidates

   Vary codec, resolution, rate control, preset, GOP, frame rate, and bitrate ladder.

3. 03

   Measure evidence

   Run Perceptual Analysis and Full-Reference Analysis with coverage and provenance.

4. 04

   Rank and prune

   Reject failed constraints and remove candidates dominated on quality, cost, or scale.

5. 05 / Decision

   Converged?

   Confirm that the winner is stable, sufficiently explored, and inside every limit.

↻If the result is not stable or misses a limit, refine the candidates and repeat.

A Services or customer workflow controls encoding and candidate selection.

Bring

Supported stream URLs, sampling intervals, expected properties, and operational thresholds.

Measure

TTFF, available startup components, decoded-frame jitter and gaps, source properties, plus a separate perceptual analysis.

1. 01

   Schedule endpoints

   Select live or test URLs, regions, cadence, and expected stream properties.

2. 02

   Probe delivery

   Measure startup time, decoded-frame cadence, failures, and decoded media properties.

3. 03

   Sample content

   Decode representative frames while preserving timing, coverage, and warnings.

4. 04

   Measure quality

   Run perceptual analysis separately from network-dependent evidence.

5. 05 / Continuous

   Store, alert, repeat

   Retain the result, alert when declared limits are missed, and schedule the next check.

↻At the next scheduled interval, sample again and retain both successful and failed checks.

Scheduling, storage, dashboards, and alerts require a custom workflow.

Bring

A high-quality camera reference, representative scenes and motion, and candidate compression settings.

Measure

VMAF, PSNR, SSIM, perceptual quality, coverage, bitrate, and separately measured compute demand.

1. 01

   Establish the source ceiling

   Capture representative camera output at its highest practical quality and retain it as the reference.

2. 02

   Generate encodes

   Vary codec, bitrate, resolution, rate control, preset, GOP, and other compression settings.

3. 03

   Compare with the source

   Run Full-Reference Analysis and Perceptual Analysis to quantify preserved quality and visible loss.

4. 04

   Find the efficiency knee

   Rank quality gain against bitrate and compute, then remove settings with diminishing returns.

5. 05 / Decision

   Optimal operating point?

   Select the lowest stable bitrate that meets the target and stays close to the camera-quality ceiling.

↻If quality or stability falls short, refine the encoded candidates and compare again.

Encoder control and the search policy belong to the customer workflow or Services engagement.

Bring

A controlled test plan, representative captures, and recorded device, lens, firmware, and profile settings.

Measure

Perceptual quality, metric evidence, confidence, and coverage; full-reference metrics where an aligned reference exists.

1. 01

   Define the test plan

   Choose devices, scenes, lighting, motion, resolutions, and acceptance criteria.

2. 02

   Capture consistently

   Use controlled fixtures and retain every device, lens, firmware, and profile setting.

3. 03

   Analyze evidence

   Measure perceptual quality and Full-Reference evidence when an aligned reference exists.

4. 04

   Compare results

   Assess quality, consistency, coverage, warnings, and regressions across conditions.

5. 05 / Output

   Qualify and report

   Publish ranked findings with configuration, provenance, and operating limits.

↻Repeat the same test plan after firmware, configuration, or hardware changes.

Acceptance criteria and the final qualification decision remain with the customer.