nx_realism_score.nx

nx_realism_score.nx -- end-to-end pipeline realism gate ("matches reality"). module: nishi-core.perception.realism_score depends: nishi-core.perception.profile + nishi-core.perception.judge + nishi-core.perception.transducer_characterizer + nishi-core.perception.signal_compensator + nishi-core.perception.codec_perceptual_grader + nishi-core.io.syscalls disk_kb: 5 capability: PERCEPTION wired_status: PARTIAL_WIRED MISSING_CAPABILITIES: - SOURCE_REAL_CAPTURE_LOAD (load the calibrated reference recording of the actual reality the pipeline is trying to reproduce) - PIPELINE_ROUND_TRIP (source -> capture -> encode -> network -> decode -> render -> physical transducer -> verification mic -> compare) - DELTA_COMPUTE (frequency-domain + temporal-domain + spatial-domain comparison; needs nx_fft + image-domain comparison primitives) - PER_AXIS_REALISM_SCORE (decompose pipeline delta into perceptual axes matching nx_perceptual_judge's 5-axis breakdown) - LATENCY_DIMENSION (separate axis: was the pipeline FAST enough that real-time interactivity preserved? composes with first-byte-of-call latency from voice codec) license_tier: PUBLIC_NISHI_SUBSTRATE genealogy_id: feedback-multi-species-perceptual-substrate-not-anthropocentric_2026 + feedback-substrate-primitives-meta-not-one-off_2026 + feedback-bits-up-exceed-never-match + feedback-honest-perf-verdict-no-aspirational-claims + feedback-end-to-end-bit-traceability-architecture + nx_codec_perceptual_grader User directive 2026-05-20: "make sure our video and audio and all that codecs are truly s class best in the world so they match reality both on a video with a person in our nishifamily.com/video for speed and quality and same for audio." THIS primitive measures whether the FULL pipeline (source capture -> codec encode -> network transit -> codec decode -> compensator pre-correction -> physical transducer -> air -> verification mic) reproduces reality close enough that a declared target profile (e.g.,