{
  "schemaVersion": "streamclarify-benchmark-examples-v2",
  "examples": [
    {
      "id": "standard",
      "type": "standard",
      "label": "Standard QA",
      "dataset": "StreamBench",
      "sampleId": "StreamBench:0e4804e0-85fa-48bc-ada3-a94167b06e53:15",
      "question": "I need to use a drill to make holes in the wooden strips. Where can I find such a tool?",
      "questionWithOptions": null,
      "options": [],
      "referenceAnswer": "There seems to be a drill on the windowsill to your right that you can use.",
      "originalQuestion": null,
      "originalAnswer": null,
      "queryTimeSeconds": 15,
      "frameTimeSeconds": 15.0,
      "image": "assets/benchmark-standard-original.jpg",
      "width": 480,
      "height": 360,
      "alt": "A power drill is visibly resting on the windowsill.",
      "visualInspection": "A power drill is visibly resting on the windowsill.",
      "provenance": {
        "annotationSource": "StreamClarify-QA, accepted paper version",
        "annotationJsonPointer": "/0/breakpoint/0",
        "annotationSha256": "609e140ab6f94b950a7d76fcc42e340c9806e546227830a9a8af7b7cb39d1573",
        "sourceKind": "original dataset video",
        "sourceVideoName": "0e4804e0-85fa-48bc-ada3-a94167b06e53.mp4",
        "sourceVideoSha256": "4fc50d8439355f9d115884a47b84d021abefd4862f3195a69213c6274b7de00c",
        "sourceFps": 30.0,
        "sourceFpsRational": "30",
        "sourceWidth": 480,
        "sourceHeight": 360,
        "width": 480,
        "height": 360,
        "clipStartSeconds": 7,
        "clipEndSeconds": 15,
        "posterTimeSeconds": 15.0,
        "posterSourceTimeSeconds": 15.0,
        "posterFrameIndex": 450,
        "transform": "scale=480:360:flags=lanczos,setsar=1",
        "transformNote": "Original frame rate and full field of view. Downscaled only when larger than 1280 × 720; no upscaling or frame interpolation. Audio omitted.",
        "videoSha256": "08aa24e60af4777438e0f7204df08e5821a38a652747260b7c6b48feace1f8df",
        "imageSha256": "986f26e612bb51d2afef1fab57a7c794295173dea0fd91888a9c6f3e80414572",
        "frameCount": 240,
        "durationSeconds": 8.0,
        "frameTimeSeconds": 15.0,
        "frameIndex": 450
      },
      "title": "A drill on the windowsill",
      "meaning": "The question names a specific tool. Its reference annotation locates that tool on the windowsill.",
      "construction": "Retain a well-specified source question.",
      "video": "assets/benchmark-standard-original.mp4",
      "clipStartSeconds": 7,
      "clipEndSeconds": 15
    },
    {
      "id": "standard-teapot",
      "type": "standard",
      "label": "Standard QA",
      "title": "Finding the red teapot",
      "meaning": "The question names a red teapot. The source annotation locates it opposite the refrigerator.",
      "construction": "Retain a well-specified source question.",
      "dataset": "StreamBench",
      "sampleId": "StreamBench:zlwC6pmyHcs:80",
      "question": "Where is a red teapot?",
      "questionWithOptions": null,
      "options": [],
      "referenceAnswer": "There is a red teapot on the cabinet opposite the refrigerator.",
      "originalQuestion": null,
      "originalAnswer": null,
      "queryTimeSeconds": 80,
      "frameTimeSeconds": 70,
      "image": "assets/benchmark-standard-teapot-original.jpg",
      "width": 640,
      "height": 480,
      "alt": "A red teapot on the kitchen island, with the refrigerator behind it.",
      "visualInspection": "A red teapot on the kitchen island, with the refrigerator behind it.",
      "provenance": {
        "annotationSource": "StreamClarify-QA, accepted paper version",
        "annotationJsonPointer": "/333/breakpoint/1",
        "annotationSha256": "609e140ab6f94b950a7d76fcc42e340c9806e546227830a9a8af7b7cb39d1573",
        "sourceKind": "original dataset video",
        "sourceVideoName": "zlwC6pmyHcs.mp4",
        "sourceVideoSha256": "cb5c33de09111a17dff3fef3e4f96d81f1c4a0952a5c171fd5dffbbdf0ba6440",
        "sourceFps": 29.97002997002997,
        "sourceFpsRational": "30000/1001",
        "sourceWidth": 640,
        "sourceHeight": 480,
        "width": 640,
        "height": 480,
        "clipStartSeconds": 64,
        "clipEndSeconds": 72,
        "posterTimeSeconds": 70,
        "posterSourceTimeSeconds": 69.9699,
        "posterFrameIndex": 2097,
        "transform": "scale=640:480:flags=lanczos,setsar=1",
        "transformNote": "Original frame rate and full field of view. Downscaled only when larger than 1280 × 720; no upscaling or frame interpolation. Audio omitted.",
        "videoSha256": "266186b5babbbabce04d8961bab1c22c8cb74d93b0fcd5b041143244860d4395",
        "imageSha256": "31b54c8ffb689d2b92b572e6dd5c8e93debbee1b40e36f726b0531fde63e0538",
        "frameCount": 239,
        "durationSeconds": 7.974633,
        "frameTimeSeconds": 70,
        "frameIndex": 2097
      },
      "video": "assets/benchmark-standard-teapot-original.mp4",
      "clipStartSeconds": 64,
      "clipEndSeconds": 72
    },
    {
      "id": "standard-eggplant",
      "type": "standard",
      "label": "Standard QA",
      "title": "An action at the sink",
      "meaning": "This question asks which food was dropped into the bowl. The retained source answer is “eggplant”; the still illustrates the cooking context.",
      "construction": "Retain an OVO-Bench question and its answer options.",
      "dataset": "OVO-Bench",
      "sampleId": "OVO-Bench:ec4a3ba3-eb00-4aa8-9b41-36043ece98f7:244",
      "question": "What did I drop in the bowl in the sink?",
      "questionWithOptions": "What did I drop in the bowl in the sink? Options: 1. meat | 2. onion | 3. eggplant | 4. carrot",
      "options": [
        "meat",
        "onion",
        "eggplant",
        "carrot"
      ],
      "referenceAnswer": "eggplant",
      "originalQuestion": null,
      "originalAnswer": null,
      "queryTimeSeconds": 244,
      "frameTimeSeconds": 240,
      "image": "assets/benchmark-standard-eggplant-original.jpg",
      "width": 960,
      "height": 720,
      "alt": "A first-person kitchen view shows cut vegetables on a cutting board and in a bowl in the sink.",
      "visualInspection": "A first-person kitchen view shows cut vegetables on a cutting board and in a bowl in the sink.",
      "provenance": {
        "annotationSource": "StreamClarify-QA, accepted paper version",
        "annotationJsonPointer": "/1/breakpoint/1",
        "annotationSha256": "609e140ab6f94b950a7d76fcc42e340c9806e546227830a9a8af7b7cb39d1573",
        "sourceKind": "original dataset video",
        "sourceVideoName": "ec4a3ba3-eb00-4aa8-9b41-36043ece98f7.mp4",
        "sourceVideoSha256": "51bc2fdfad3c9b2c84812c4830bddb8887ea6be54ca8f7e82bf574e116923a2f",
        "sourceFps": 30.0,
        "sourceFpsRational": "30",
        "sourceWidth": 1920,
        "sourceHeight": 1440,
        "width": 960,
        "height": 720,
        "clipStartSeconds": 234,
        "clipEndSeconds": 242,
        "posterTimeSeconds": 240,
        "posterSourceTimeSeconds": 240.0,
        "posterFrameIndex": 7200,
        "transform": "scale=960:720:flags=lanczos,setsar=1",
        "transformNote": "Original frame rate and full field of view. Downscaled only when larger than 1280 × 720; no upscaling or frame interpolation. Audio omitted.",
        "videoSha256": "769d435c20e0c3aa607febd3bc02b93f0b1e9b82cd9e92825ffd5972e353b3d5",
        "imageSha256": "4795edf9ee0118c462dc6b2660cc64bdd0bec588d9b5cd164f8d325a980fd3a3",
        "frameCount": 240,
        "durationSeconds": 8.0,
        "frameTimeSeconds": 240,
        "frameIndex": 7200
      },
      "video": "assets/benchmark-standard-eggplant-original.mp4",
      "clipStartSeconds": 234,
      "clipEndSeconds": 242
    },
    {
      "id": "ambiguous",
      "type": "ambiguous",
      "label": "Referential Ambiguity",
      "dataset": "StreamBench",
      "sampleId": "StreamBench:1c53277a-eff1-4d73-bab6-62644c2029a9:20",
      "question": "What color is this?",
      "questionWithOptions": null,
      "options": [],
      "referenceAnswer": "The electric drill you're holding is likely black and red.",
      "originalQuestion": "What color is the electric drill I'm holding?",
      "originalAnswer": null,
      "queryTimeSeconds": 20,
      "frameTimeSeconds": 20.0,
      "image": "assets/benchmark-ambiguous-original.jpg",
      "width": 480,
      "height": 360,
      "alt": "A red drill, a hammer, wooden framing and the wearer's sweatshirt are visible.",
      "visualInspection": "A red drill, a hammer, wooden framing and the wearer's sweatshirt are visible.",
      "provenance": {
        "annotationSource": "StreamClarify-QA, accepted paper version",
        "annotationJsonPointer": "/319/breakpoint/0",
        "annotationSha256": "609e140ab6f94b950a7d76fcc42e340c9806e546227830a9a8af7b7cb39d1573",
        "sourceKind": "original dataset video",
        "sourceVideoName": "1c53277a-eff1-4d73-bab6-62644c2029a9.mp4",
        "sourceVideoSha256": "2e7e823c51553a26194dbf7671fb78a1e22851fb56fc1b2ecf72b8181daa5cdc",
        "sourceFps": 30.0,
        "sourceFpsRational": "30",
        "sourceWidth": 480,
        "sourceHeight": 360,
        "width": 480,
        "height": 360,
        "clipStartSeconds": 12,
        "clipEndSeconds": 20,
        "posterTimeSeconds": 20.0,
        "posterSourceTimeSeconds": 20.0,
        "posterFrameIndex": 600,
        "transform": "scale=480:360:flags=lanczos,setsar=1",
        "transformNote": "Original frame rate and full field of view. Downscaled only when larger than 1280 × 720; no upscaling or frame interpolation. Audio omitted.",
        "videoSha256": "a62a457a69f4fe921a7e37fbdec349d9bddb2cccda36badce6508d6a4ad126f1",
        "imageSha256": "abe6be3a1b0eeea07820356c637433cc2ad39a63d5927c5e0bdea828a1cd7e33",
        "frameCount": 240,
        "durationSeconds": 8.0,
        "frameTimeSeconds": 20.0,
        "frameIndex": 600
      },
      "title": "Which object is “this”?",
      "meaning": "The original question names the electric drill; the rewritten question replaces it with “this”.",
      "construction": "Replace the specific object description with a pronoun.",
      "ambiguityExplanation": "“This” could refer to the drill or the sweatshirt.",
      "video": "assets/benchmark-ambiguous-original.mp4",
      "clipStartSeconds": 12,
      "clipEndSeconds": 20
    },
    {
      "id": "ambiguous-apron",
      "type": "ambiguous",
      "label": "Referential Ambiguity",
      "title": "Remembering an item of clothing",
      "meaning": "“An apron” becomes “that”. This earlier frame illustrates the intended clothing item; it does not establish that the model later retained or recovered it.",
      "construction": "Replace the named clothing item with an underspecified reference.",
      "dataset": "StreamBench",
      "sampleId": "StreamBench:P-mG7YxfFPA:447",
      "question": "Do you remember if the short-haired woman in the video was wearing that?",
      "questionWithOptions": null,
      "options": [],
      "referenceAnswer": "Yes, The short-haired woman in the video is wearing a green apron.",
      "originalQuestion": "Do you remember if the short-haired woman in the video was wearing an apron?",
      "originalAnswer": null,
      "queryTimeSeconds": 447,
      "frameTimeSeconds": 93,
      "image": "assets/benchmark-ambiguous-apron-original.jpg",
      "width": 320,
      "height": 240,
      "alt": "Two women at a cooking counter; the short-haired woman is wearing a green apron.",
      "visualInspection": "Two women at a cooking counter; the short-haired woman is wearing a green apron.",
      "provenance": {
        "annotationSource": "StreamClarify-QA, accepted paper version",
        "annotationJsonPointer": "/223/breakpoint/5",
        "annotationSha256": "609e140ab6f94b950a7d76fcc42e340c9806e546227830a9a8af7b7cb39d1573",
        "sourceKind": "original dataset video",
        "sourceVideoName": "P-mG7YxfFPA.mp4",
        "sourceVideoSha256": "6c6bb32a075c272762d5adbf0c781a0ca340b36725503aee43016c6e616ccfd7",
        "sourceFps": 25.0,
        "sourceFpsRational": "25",
        "sourceWidth": 320,
        "sourceHeight": 240,
        "width": 320,
        "height": 240,
        "clipStartSeconds": 87,
        "clipEndSeconds": 95,
        "posterTimeSeconds": 93,
        "posterSourceTimeSeconds": 93.0,
        "posterFrameIndex": 2325,
        "transform": "scale=320:240:flags=lanczos,setsar=1",
        "transformNote": "Original frame rate and full field of view. Downscaled only when larger than 1280 × 720; no upscaling or frame interpolation. Audio omitted.",
        "videoSha256": "9c9d6099b47786194f80e7240b473bc42c42420e073a75f8f8eb9c26567b89aa",
        "imageSha256": "cb1c81286bf2a473784d55b108a9c9dfeaca24d9bccf6cd2d148c0265062889d",
        "frameCount": 200,
        "durationSeconds": 8.0,
        "frameTimeSeconds": 93,
        "frameIndex": 2325
      },
      "video": "assets/benchmark-ambiguous-apron-original.mp4",
      "clipStartSeconds": 87,
      "clipEndSeconds": 95
    },
    {
      "id": "ambiguous-toy",
      "type": "ambiguous",
      "label": "Referential Ambiguity",
      "title": "Which item in the store?",
      "meaning": "The source question names a toy box; the rewritten question uses “it”.",
      "construction": "Replace the named toy box with an underspecified reference.",
      "dataset": "OVO-Bench",
      "sampleId": "OVO-Bench:33c3f556-cf99-45ae-9d01-9051b9b1f19e:51",
      "question": "Where did I put it?",
      "questionWithOptions": "Where did I put it? Options: 1. On the floor | 2. At the toy rack | 3. In the shopping basket | 4. On the shelf",
      "options": [
        "On the floor",
        "At the toy rack",
        "In the shopping basket",
        "On the shelf"
      ],
      "referenceAnswer": "At the toy rack",
      "originalQuestion": "Where did I put the alien toy box? Options: 1. On the floor | 2. At the toy rack | 3. In the shopping basket | 4. On the shelf",
      "originalAnswer": null,
      "queryTimeSeconds": 51,
      "frameTimeSeconds": 45,
      "image": "assets/benchmark-ambiguous-toy-original.jpg",
      "video": "assets/benchmark-ambiguous-toy-original.mp4",
      "clipStartSeconds": 43,
      "clipEndSeconds": 51,
      "width": 1280,
      "height": 720,
      "alt": "A first-person store view shows boxed products on display and a person holding a box.",
      "visualInspection": "A person holds a box beside shelves of boxed products.",
      "ambiguityExplanation": "“It” does not specify which item in the store is intended.",
      "provenance": {
        "annotationSource": "StreamClarify-QA, accepted paper version",
        "annotationJsonPointer": "/171/breakpoint/0",
        "annotationSha256": "609e140ab6f94b950a7d76fcc42e340c9806e546227830a9a8af7b7cb39d1573",
        "sourceKind": "original dataset video",
        "sourceVideoName": "33c3f556-cf99-45ae-9d01-9051b9b1f19e.mp4",
        "sourceVideoSha256": "6bc6fb4238943b7009af850e66da15fccea9bff1694751a6e7fec9d60a8c2b47",
        "sourceFps": 30.0,
        "sourceFpsRational": "30",
        "sourceWidth": 1920,
        "sourceHeight": 1080,
        "width": 1280,
        "height": 720,
        "clipStartSeconds": 43,
        "clipEndSeconds": 51,
        "posterTimeSeconds": 45,
        "posterSourceTimeSeconds": 45.0,
        "posterFrameIndex": 1350,
        "transform": "scale=1280:720:flags=lanczos,setsar=1",
        "transformNote": "Original frame rate and full field of view. Downscaled only when larger than 1280 × 720; no upscaling or frame interpolation. Audio omitted.",
        "videoSha256": "2eb01b0e83bc6c9f9c965b0e9f2f78343ae08b4ba5130deeaec41e543a5cf758",
        "imageSha256": "ef6d7b6726d9326cb9c8a5e05ee428749f3c35797d93d3337bcc92ba32a45ba0",
        "frameCount": 240,
        "durationSeconds": 8.0,
        "frameTimeSeconds": 45,
        "frameIndex": 1350
      }
    },
    {
      "id": "unanswerable",
      "type": "evidence_absent",
      "label": "Unanswerable QA",
      "dataset": "StreamBench",
      "sampleId": "StreamBench:fMQYY3e6kcM:16",
      "question": "What color is this man's hat?",
      "questionWithOptions": "What color is this man's hat? Options: 1. The hat is brown. | 2. None of the above | 3. The hat is blue. | 4. The hat is red.",
      "options": [
        "The hat is brown.",
        "None of the above",
        "The hat is blue.",
        "The hat is red."
      ],
      "referenceAnswer": "None of the above",
      "originalQuestion": null,
      "originalAnswer": "The man's hat is black or gray.",
      "queryTimeSeconds": 16,
      "frameTimeSeconds": 3.0,
      "image": "assets/benchmark-unanswerable-original.jpg",
      "width": 640,
      "height": 360,
      "alt": "The man is visibly wearing a black cap in the kitchen. Black is absent from the substantive answer options.",
      "visualInspection": "The man is visibly wearing a black cap in the kitchen. Black is absent from the substantive answer options.",
      "provenance": {
        "annotationSource": "StreamClarify-QA, accepted paper version",
        "annotationJsonPointer": "/217/breakpoint/0",
        "annotationSha256": "609e140ab6f94b950a7d76fcc42e340c9806e546227830a9a8af7b7cb39d1573",
        "sourceKind": "original dataset video",
        "sourceVideoName": "fMQYY3e6kcM.mp4",
        "sourceVideoSha256": "2453a5d3a492799f75b30699451e1f2b8b6e12fae5592b1a4f07b08b73d74f90",
        "sourceFps": 23.976023976023978,
        "sourceFpsRational": "24000/1001",
        "sourceWidth": 640,
        "sourceHeight": 360,
        "width": 640,
        "height": 360,
        "clipStartSeconds": 0,
        "clipEndSeconds": 5.0,
        "posterTimeSeconds": 3.0,
        "posterSourceTimeSeconds": 2.961291667,
        "posterFrameIndex": 71,
        "transform": "scale=640:360:flags=lanczos,setsar=1",
        "transformNote": "Original frame rate and full field of view. Downscaled only when larger than 1280 × 720; no upscaling or frame interpolation. Audio omitted.",
        "videoSha256": "3657e4f478d929e6be9176cc6a4f42cef6875fc83d350caec85cb95b43a46c3d",
        "imageSha256": "ba08fadb09c3fd5455a4b38be13d8368cd8ec401873ba79b662171e2fa3444a5",
        "frameCount": 120,
        "durationSeconds": 5.005,
        "frameTimeSeconds": 3.0,
        "frameIndex": 71
      },
      "title": "A missing color option",
      "meaning": "The hat is black. Black is missing from the offered choices.",
      "construction": "Remove the original answer and add “None of the above”.",
      "video": "assets/benchmark-unanswerable-original.mp4",
      "clipStartSeconds": 0,
      "clipEndSeconds": 5.0
    },
    {
      "id": "unanswerable-device",
      "type": "evidence_absent",
      "label": "Unanswerable QA",
      "title": "A missing device option",
      "meaning": "The source answer is a massage gun, which is missing from the choices.",
      "construction": "Remove the supported device identity and add “None of the above”.",
      "dataset": "StreamBench",
      "sampleId": "StreamBench:080ae20b-7914-49ff-8c1c-e384599f2b7a:25",
      "question": "What did I just plug into the outlet to charge?",
      "questionWithOptions": "What did I just plug into the outlet to charge? Options: 1. A wireless phone charger | 2. A portable power bank | 3. A handheld vacuum cleaner | 4. None of the above",
      "options": [
        "A wireless phone charger",
        "A portable power bank",
        "A handheld vacuum cleaner",
        "None of the above"
      ],
      "referenceAnswer": "None of the above",
      "originalQuestion": null,
      "originalAnswer": "It looks like a black massage gun.",
      "queryTimeSeconds": 25,
      "frameTimeSeconds": 6,
      "image": "assets/benchmark-unanswerable-device-original.jpg",
      "width": 480,
      "height": 360,
      "alt": "A person holds a black massage gun with a round head while connecting an adapter at a wall outlet.",
      "visualInspection": "The close-up shows the massage gun’s round attachment and handle, together with the wall outlet and charging adapters.",
      "provenance": {
        "annotationSource": "StreamClarify-QA, accepted paper version",
        "annotationJsonPointer": "/285/breakpoint/0",
        "annotationSha256": "609e140ab6f94b950a7d76fcc42e340c9806e546227830a9a8af7b7cb39d1573",
        "sourceKind": "original dataset video",
        "sourceVideoName": "080ae20b-7914-49ff-8c1c-e384599f2b7a.mp4",
        "sourceVideoSha256": "91723fb004411d9991a60670c85c959009f3407cea3cc5bde9a5c6b008c7b2d2",
        "sourceFps": 30.0,
        "sourceFpsRational": "30",
        "sourceWidth": 480,
        "sourceHeight": 360,
        "width": 480,
        "height": 360,
        "clipStartSeconds": 0,
        "clipEndSeconds": 8,
        "posterTimeSeconds": 6,
        "posterSourceTimeSeconds": 6.0,
        "posterFrameIndex": 180,
        "transform": "scale=480:360:flags=lanczos,setsar=1",
        "transformNote": "Original frame rate and full field of view. Downscaled only when larger than 1280 × 720; no upscaling or frame interpolation. Audio omitted.",
        "videoSha256": "5f980069dd79bb98ab3df387d4f521b03a1e4fb4aa23374f551a514cfb28fd87",
        "imageSha256": "6552102156495f54317fc3f80b3cad0f09266d01440ef6f7243e5694e00025c1",
        "frameCount": 240,
        "durationSeconds": 8.0,
        "frameTimeSeconds": 6,
        "frameIndex": 180
      },
      "video": "assets/benchmark-unanswerable-device-original.mp4",
      "clipStartSeconds": 0,
      "clipEndSeconds": 8
    },
    {
      "id": "unanswerable-oil",
      "type": "evidence_absent",
      "label": "Unanswerable QA",
      "title": "An uncertain past action",
      "meaning": "OVO-Bench labels this question “Unable to answer”. The excerpt does not establish what happened throughout the prefix.",
      "construction": "Retain the native OVO-Bench HLD question, options and reference label.",
      "dataset": "OVO-Bench",
      "sampleId": "OVO-Bench:ec4a3ba3-eb00-4aa8-9b41-36043ece98f7:359",
      "question": "Did I close the bottle of oil?",
      "questionWithOptions": "Did I close the bottle of oil? Options: 1. No, I left it open | 2. Unable to answer | 3. The video doesn't show this",
      "options": [
        "No, I left it open",
        "Unable to answer",
        "The video doesn't show this"
      ],
      "referenceAnswer": "Unable to answer",
      "originalQuestion": null,
      "originalAnswer": null,
      "queryTimeSeconds": 359,
      "frameTimeSeconds": 359,
      "image": "assets/benchmark-unanswerable-oil-original.jpg",
      "width": 960,
      "height": 720,
      "alt": "A first-person view of a kitchen counter with food preparation and several bottles beside the stove.",
      "visualInspection": "A first-person view of a kitchen counter with food preparation and several bottles beside the stove.",
      "provenance": {
        "annotationSource": "StreamClarify-QA, accepted paper version",
        "annotationJsonPointer": "/1/breakpoint/3",
        "annotationSha256": "609e140ab6f94b950a7d76fcc42e340c9806e546227830a9a8af7b7cb39d1573",
        "sourceKind": "original dataset video",
        "sourceVideoName": "ec4a3ba3-eb00-4aa8-9b41-36043ece98f7.mp4",
        "sourceVideoSha256": "51bc2fdfad3c9b2c84812c4830bddb8887ea6be54ca8f7e82bf574e116923a2f",
        "sourceFps": 30.0,
        "sourceFpsRational": "30",
        "sourceWidth": 1920,
        "sourceHeight": 1440,
        "width": 960,
        "height": 720,
        "clipStartSeconds": 351,
        "clipEndSeconds": 359,
        "posterTimeSeconds": 359,
        "posterSourceTimeSeconds": 359.0,
        "posterFrameIndex": 10770,
        "transform": "scale=960:720:flags=lanczos,setsar=1",
        "transformNote": "Original frame rate and full field of view. Downscaled only when larger than 1280 × 720; no upscaling or frame interpolation. Audio omitted.",
        "videoSha256": "3720cbd91f90853c98e774f3059e7fdaa278d9090d3b760221f177ecadb4013a",
        "imageSha256": "f80a1a15379684c07a84d3c4156d9c9eb7f02097b3c3a5b9e665f18f7fc00ff1",
        "frameCount": 240,
        "durationSeconds": 8.0,
        "frameTimeSeconds": 359,
        "frameIndex": 10770
      },
      "video": "assets/benchmark-unanswerable-oil-original.mp4",
      "clipStartSeconds": 351,
      "clipEndSeconds": 359
    }
  ],
  "note": "Short video excerpts from the observed prefix. Questions and answers are reference annotations.",
  "unanswerableNote": "Constructed StreamBench examples remove the original supported answer and include None of the above. Native OVO-Bench HLD examples preserve their original question, options and Unable to answer label; an individual displayed frame does not prove that information is absent over the entire prefix."
}