framefields 2.10.0-canary.0 → 2.11.0

package/dist/{src-BU6vm9sP-DFJ3phW0.mjs → src-Bnr8DUC--BRDqRF6I.mjs}

@@ -6,14 +6,13 @@ import { readFile, rename, unlink, writeFile } from "node:fs/promises";
 import { createHash } from "node:crypto";
 import { homedir } from "node:os";
 
-//#region ../vision/dist/registry-DOHiTjt1.mjs
+//#region ../vision/dist/registry-RNCQoD9e.mjs
 const hf = (repo, revision, path$1) => `https://huggingface.co/${repo}/resolve/${revision}/${path$1}`;
 const RTMDET_INS_REPO = "tori29umai/rtmdet-ins-onnx-person-masks";
 const RTMDET_INS_REV = "e5218e8d47d7c50077f27f9893a28e0ef2ad5656";
 const RTMDET_INS_SOURCE = "RTMDet-Ins (OpenMMLab mmdetection, Apache-2.0), mmdeploy ONNX export";
 const RTMO_SOURCE = "RTMO (OpenMMLab mmpose, Apache-2.0), ONNX export";
 const SELFIE_SOURCE = "MediaPipe Selfie Segmenter (Google, Apache-2.0), ONNX export";
-const DEPTH_SOURCE = "Depth Anything V2 Small (TikTok / HKU, Apache-2.0), ONNX export";
 const models = {
 	"rtmdet-ins-t": {
 		key: "rtmdet-ins-t",
@@ -94,16 +93,6 @@ const models = {
 		sha256: "7a0adcfdb1715d3b0ff0f61486d8a181c33f4a208343a5abfccbc6540e872d24",
 		input: [256, 144],
 		source: SELFIE_SOURCE
-	},
-	"depth-anything-v2-s": {
-		key: "depth-anything-v2-s",
-		family: "depth-anything",
-		filename: "depth-anything-v2-s.onnx",
-		url: hf("onnx-community/depth-anything-v2-small", "4472b7362082ad9968fee890ca0f1e5aca36b93d", "onnx/model.onnx"),
-		bytes: 99060839,
-		sha256: "afb6a5c28f3b6bf1618c6e43f02073ef9dfdc70e937502d51603e57b0a1df10c",
-		input: [518, 518],
-		source: DEPTH_SOURCE
 	}
 };
 const VISION_MODELS = models;
@@ -205,7 +194,7 @@ const COCO_CLASSES = [
 ];
 
 //#endregion
-//#region ../vision/dist/src-BU6vm9sP.mjs
+//#region ../vision/dist/src-Bnr8DUC-.mjs
 function computeIoU(boxA, boxB) {
 	const xA = Math.max(boxA.originX, boxB.originX);
 	const yA = Math.max(boxA.originY, boxB.originY);
@@ -677,35 +666,6 @@ async function analyzeSequence(frames, options) {
 		masks
 	};
 }
-/** Model input size for a frame: the short side at `short`, the other kept to the frame's aspect, both multiples of the 14 px ViT patch. */
-function depthInputSize(width, height, short) {
-	const snap = (v) => Math.max(14, Math.round(v / 14) * 14);
-	return width >= height ? [snap(short * width / height), snap(short)] : [snap(short), snap(short * height / width)];
-}
-/** Share of samples clipped at each end, so a stray hot pixel does not set the range. */
-const TAIL = .01;
-/**
-* Depth Anything decode: `predicted_depth [1, h, w]` relative inverse depth →
-* 0 (farthest) … 1 (nearest), at the model's resolution. The range is the
-* frame's 1st…99th percentile, so it is relative to the frame, not metric.
-*/
-function decodeDepth(predicted, width, height) {
-	const n = width * height;
-	const step = Math.max(1, Math.floor(n / 4096));
-	const samples = [];
-	for (let i = 0; i < n; i += step) samples.push(predicted[i]);
-	samples.sort((a, b) => a - b);
-	const lo = samples[Math.floor((samples.length - 1) * TAIL)];
-	const hi = samples[Math.ceil((samples.length - 1) * (1 - TAIL))];
-	const scale = hi > lo ? 1 / (hi - lo) : 0;
-	const data = new Float32Array(n);
-	for (let i = 0; i < n; i++) data[i] = Math.min(1, Math.max(0, (predicted[i] - lo) * scale));
-	return {
-		data,
-		width,
-		height
-	};
-}
 const IMAGENET_BGR_MEAN = [
 	103.53,
 	116.28,
@@ -754,21 +714,6 @@ const PREPROCESS_BY_FAMILY = {
 			255
 		],
 		padValue: 0
-	},
-	"depth-anything": {
-		fit: "stretch",
-		channels: "rgb",
-		mean: [
-			123.675,
-			116.28,
-			103.53
-		],
-		std: [
-			58.395,
-			57.12,
-			57.375
-		],
-		padValue: 0
 	}
 };
 /** True for the families `VisionRunner` can preprocess and decode itself. */
@@ -2906,15 +2851,6 @@ var VisionRunner = class VisionRunner$1 {
 			...this.featherRadius !== void 0 ? { featherRadius: this.featherRadius } : {}
 		});
 	}
-	/** Relative depth of the whole frame (Depth Anything V2 Small), near = 1, at the model's resolution. */
-	async depth(image) {
-		const key = "depth-anything-v2-s";
-		const input = depthInputSize(image.width, image.height, VISION_MODELS[key].input[1]);
-		const { outputs } = await this.infer(key, image, input);
-		const predicted = output(outputs, "predicted_depth", key);
-		const dims = predicted.dims;
-		return decodeDepth(predicted.data, dims[dims.length - 1] ?? input[0], dims[dims.length - 2] ?? input[1]);
-	}
 	close() {
 		if (!this._ownsCore) return;
 		const core = this._core;
@@ -2964,9 +2900,9 @@ var VisionRunner = class VisionRunner$1 {
 	* Runs one model on one image. Runs of the same model are queued: they share
 	* the input scratch buffer, and one run already uses the session's threads.
 	*/
-	infer(key, image, input) {
+	infer(key, image) {
 		const queue = this._core.runQueue;
-		const run = (queue.get(key) ?? Promise.resolve()).then(() => this.inferNow(key, image, input), () => this.inferNow(key, image, input));
+		const run = (queue.get(key) ?? Promise.resolve()).then(() => this.inferNow(key, image), () => this.inferNow(key, image));
 		const tail = run.catch(() => void 0);
 		queue.set(key, tail);
 		tail.then(() => {
@@ -2974,11 +2910,11 @@ var VisionRunner = class VisionRunner$1 {
 		});
 		return run;
 	}
-	async inferNow(key, frame, input) {
+	async inferNow(key, frame) {
 		const session = await this.session(key);
 		const desc = VISION_MODELS[key];
 		if (!isPreprocessFamily(desc.family)) throw new Error(`VisionRunner cannot run '${desc.family}' models; use the family's own tracker (e.g. Sam2Tracker or Sam3Tracker from @framefields/vision-sam).`);
-		const spec = preprocessSpec(desc.family, input ?? desc.input);
+		const spec = preprocessSpec(desc.family, desc.input);
 		const dims = [
 			1,
 			3,
@@ -3141,9 +3077,8 @@ function output(outputs, name, key) {
 /**
 * Where a render-time Vision effect finds subject masks someone already
 * computed for its clip (`comp.vision(clip)` in framefields), so a matte layer
-* reuses them instead of segmenting the frame again, a skeleton layer draws
-* the pose matched to that subject instead of the first person it detects, and
-* a depth layer shows the depth analysed for each frame (not one behind). Set
+* reuses them instead of segmenting the frame again, and a skeleton layer draws
+* the pose matched to that subject instead of the first person it detects. Set
 * on the Vision effect instance under this key; the renderer reads it from the effect.
 */
 const CLIP_SUBJECT = Symbol.for("framefields.vision.clipSubject");
@@ -4381,4 +4316,4 @@ var VisionNode = class VisionNode$1 {
 
 //#endregion
 export { setVisionDevice as A, matchPoseToTracks as C, readTexturePixels as D, pinNodeToObject as E, registerVisionModel as F, varint as M, COCO_CLASSES as N, readVarint as O, VISION_MODELS as P, maskObject as S, pinNodeToLandmark as T, fields as _, PoseSkeletonRenderer as a, lengthDelimited as b, TemporalObjectTracker as c, VisionNode as d, VisionRunner as f, encoder as g, createDefaultSessionProvider as h, COCO17_KEYPOINT_NAMES as i, strings as j, sessionCompute as k, VisionBundle as l, concat as m, COCO17_BONES as n, SegmentationTexturePool as o, analyzeSequence as p, COCO17_KEYPOINTS as r, SpatialLandmarkTransformer as s, CLIP_SUBJECT as t, VisionModelStore as u, firstString as v, mergeSubjectMask as w, maskBounds as x, imageToTensor as y };
-//# sourceMappingURL=src-BU6vm9sP-DFJ3phW0.mjs.map
+//# sourceMappingURL=src-Bnr8DUC--BRDqRF6I.mjs.map