clyroq scores every frame for camera stability and person-free zones, places your ad with lighting-matched compositing, tracks the surface for the full clip, and grades its own work with a per-frame Trust Score.
Four pipeline modules score every frame and fuse into a single Reliability Index, from raw geometry to final photorealism checks.
Builds a 3D point cloud from each frame and scores surface planarity, identifying anchor regions stable enough to composite onto.
Scores camera jitter per time segment with sparse optical flow, then maps person-free confidence across a 3x3 zone grid on every stable segment.
Applies semantic segmentation to resolve depth order, so people who pass in front of the ad are always rendered on top of it, not behind.
Combines structural similarity, photometric variance, and a vision-language judge to catch bad blends before a video ships.
Each job follows the same deterministic flow, with measurable confidence at every step.
Decode, segment, and score camera stability before any compositing starts.
Compute person-free confidence across a 3x3 map and choose the safest zone.
Warp the ad to the surface, match color statistics, and lock motion with optical flow.
Produce a Trust Score, frame-level verdicts, and hotspot metadata for playback.
Deploy clyroq as a service or run it in your own environment. Every frame-level decision is inspectable for legal, editorial, and brand-safety workflows.
Short clips can finish in under a minute on CPU. Typical 10-60 second social clips complete in roughly 20-90 seconds depending on resolution and hardware.
Yes. Enterprise plans support private deployment and integration through the API and job-orchestration hooks.
The system returns a clear reject state with segment jitter and confidence diagnostics, so your team can retry with different footage or a smaller ad tier.
Upload a clip, pick your CTA, and inspect every decision from placement coordinates to frame-level Trust Score.