orchestra

ORCHESTRA / SYNTHETIC SIGNAL LAB NO LIVE VIDEO

Follow the signal.
Keep the evidence in frame.

Watch a fictional pattern move from video fragments to a cautious review lead. Then change the coverage and independent-view evidence and see what the system can actually say.

READY · SAMPLE DATA ONLY

Every clip, site, time, count and model output here is generated locally in your browser. This demo does not query the archive or run inference.

MOTIONCONTEXTCOVERAGE
06SAMPLE
MONTHS

SAMPLE WINDOW / 06

A repeating route appears.

GENERATED SCENE
Abstract synthetic city views connected by a repeated path Three fictional observation windows show glowing markers moving between abstract blocks. This is not a map or a recording. VIEW A VIEW B VIEW C VIEW D WINDOW 06 · ROUTE SAMPLE FICTIONAL GRID · NO GEOGRAPHIC COORDINATES · NO PEOPLE DEPICTED
SAMPLE WINDOW 06VECTOR VIEW / NOT A RECORDING
REPEATED MOTION CUESIMULATED FRAME 00:14
recurring sample path independent viewAbstract motion only

THE ANALYSIS RUN

A cue moves through six checks.

READY
  1. 01▤Archive slicesOrdered sample windowsWAITING
  2. 02⌁V-JEPA motionCompare movement over timeWAITING
  3. 03✳Cosmos contextDescribe the visible sceneWAITING
  4. 04⊹Independent viewsCheck whether views agreeWAITING
  5. 05▥Coverage gateCount complete evidence daysWAITING
  6. 06↗Human reviewDecide what evidence supportsWAITING

Run the sample to animate each stage. All output is generated in this page.

01 / MOTION

V-JEPA compares movement.

It can surface similar motion windows; it does not name people or infer motive.

02 / CONTEXT

Cosmos adds scene descriptions.

Model-written context is a hypothesis to check against the source evidence.

03 / EVIDENCE

Coverage sets the limit.

Gaps stay unknown. A repeated pattern needs enough complete, independent evidence before review.