Interface: EncodeClipResponse
@kortexya/reasoninglayer / EncodeClipResponse
Interface: EncodeClipResponse
Section titled “Interface: EncodeClipResponse”Defined in: src/types/vision.ts:87
Response of POST /api/v1/vision/encode-clip.
Properties
Section titled “Properties”clipId
Section titled “clipId”clipId:
string
Defined in: src/types/vision.ts:89
Caller-chosen clip id, echoed.
disposition
Section titled “disposition”disposition:
string
Defined in: src/types/vision.ts:95
How to read the predictions. The engine always sends
"advisory_prediction": predictions stay advisory, also when every
declared OSF check passes.
embeddingDim
Section titled “embeddingDim”embeddingDim:
number
Defined in: src/types/vision.ts:97
Dimensionality of the produced embedding.
encodeElapsedMs
Section titled “encodeElapsedMs”encodeElapsedMs:
number
Defined in: src/types/vision.ts:99
Wall-clock time spent encoding, in milliseconds.
frames
Section titled “frames”frames:
FrameSummary[]
Defined in: src/types/vision.ts:101
Per-frame prediction summaries, in frame order.
headManifestSha256?
Section titled “headManifestSha256?”
optionalheadManifestSha256?:string|null
Defined in: src/types/vision.ts:113
SHA-256 of the head manifest that produced the predictions.
headWeightsSha256?
Section titled “headWeightsSha256?”
optionalheadWeightsSha256?:string|null
Defined in: src/types/vision.ts:115
SHA-256 of the head weights that produced the predictions.
modelId
Section titled “modelId”modelId:
string
Defined in: src/types/vision.ts:103
Model id that produced the embedding.
modelVersion?
Section titled “modelVersion?”
optionalmodelVersion?:string|null
Defined in: src/types/vision.ts:105
Model version, when reported. Optional.
numInputFrames
Section titled “numInputFrames”numInputFrames:
number
Defined in: src/types/vision.ts:107
Number of input frames actually consumed.
originalFrameCount?
Section titled “originalFrameCount?”
optionaloriginalFrameCount?:number|null
Defined in: src/types/vision.ts:109
Frame count of the decoded source video, when known.
paddedFrameIndices
Section titled “paddedFrameIndices”paddedFrameIndices:
number[]
Defined in: src/types/vision.ts:111
Frame indices that were padded to reach the required count.
receiptPayloadSha256?
Section titled “receiptPayloadSha256?”
optionalreceiptPayloadSha256?:string|null
Defined in: src/types/vision.ts:117
SHA-256 of the receipt payload the engine stored.
receiptTermId?
Section titled “receiptTermId?”
optionalreceiptTermId?:string|null
Defined in: src/types/vision.ts:119
Id of the receipt term the engine wrote to the tenant’s store.
sourceByteCount?
Section titled “sourceByteCount?”
optionalsourceByteCount?:number|null
Defined in: src/types/vision.ts:121
Byte count of the decoded source video.
sourceSha256?
Section titled “sourceSha256?”
optionalsourceSha256?:string|null
Defined in: src/types/vision.ts:123
SHA-256 of the decoded source video.
temporalPatchSize?
Section titled “temporalPatchSize?”
optionaltemporalPatchSize?:number|null
Defined in: src/types/vision.ts:125
Source frames per temporal token (the encoder’s temporal patch size).
timingsUs
Section titled “timingsUs”timingsUs:
Record<string,number>
Defined in: src/types/vision.ts:132
Stage wall-clock times in microseconds, keyed by stage. The engine sends
encoder_round_trip, head, validation_and_osf, processing and
receipt_write_readback; receipt I/O is timed apart from processing.
The keys are data and reach the caller unchanged.
totalObjects
Section titled “totalObjects”totalObjects:
number
Defined in: src/types/vision.ts:134
Total objects predicted across all frames.
totalResiduatedFeatures
Section titled “totalResiduatedFeatures”totalResiduatedFeatures:
number
Defined in: src/types/vision.ts:136
Total residuated features across all frames (the headline “honest uncertainty” number).