Upload 19 files

.gitattributes

@@ -53,3 +53,13 @@ sample/toast-shmello/first_frame.jpg filter=lfs diff=lfs merge=lfs -text
 sample/crush-can/first_frame.jpg filter=lfs diff=lfs merge=lfs -text
 sample/bowling/first_frame.jpg filter=lfs diff=lfs merge=lfs -text
 sample/BigBen/input_video.mp4 filter=lfs diff=lfs merge=lfs -text
+sample/ducky-float/first_frame.jpg filter=lfs diff=lfs merge=lfs -text
+sample/ducky-float/input_video.mp4 filter=lfs diff=lfs merge=lfs -text
+sample/ducky-float/trimask_quadmask.mp4 filter=lfs diff=lfs merge=lfs -text
+sample/spinner/input_video.mp4 filter=lfs diff=lfs merge=lfs -text
+sample/spinner/trimask_quadmask.mp4 filter=lfs diff=lfs merge=lfs -text
+sample/trampoline/input_video.mp4 filter=lfs diff=lfs merge=lfs -text
+sample/trampoline/masks_output/sam3_masks/black_mask.mp4 filter=lfs diff=lfs merge=lfs -text
+sample/trampoline/masks_output/sam3_masks/grey_mask.mp4 filter=lfs diff=lfs merge=lfs -text
+sample/trampoline/masks_output/sam3_masks/input_video.mp4 filter=lfs diff=lfs merge=lfs -text
+sample/trampoline/trimask_quadmask.mp4 filter=lfs diff=lfs merge=lfs -text

sample/ducky-float/input_video.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:07acee2c2d67a7ac912aa044aeff5af6edf68189681b5cd7c0ab9969e697d171
+size 1327905

sample/ducky-float/prompt.json

@@ -0,0 +1,3 @@
+{
+  "bg": "A video of water with a floating object."
+}

sample/ducky-float/segmentation_info.json

@@ -0,0 +1,215 @@
+{
+  "total_frames": 62,
+  "frame_width": 1296,
+  "frame_height": 720,
+  "fps": 12.0,
+  "num_points": 23,
+  "points_by_frame": {
+    "7": [
+      [
+        268,
+        16
+      ],
+      [
+        353,
+        16
+      ],
+      [
+        740,
+        14
+      ],
+      [
+        900,
+        14
+      ]
+    ],
+    "18": [
+      [
+        931,
+        121
+      ],
+      [
+        688,
+        121
+      ],
+      [
+        511,
+        121
+      ],
+      [
+        283,
+        110
+      ],
+      [
+        549,
+        6
+      ],
+      [
+        657,
+        17
+      ],
+      [
+        771,
+        27
+      ],
+      [
+        461,
+        66
+      ]
+    ],
+    "36": [
+      [
+        821,
+        46
+      ],
+      [
+        636,
+        46
+      ],
+      [
+        445,
+        50
+      ],
+      [
+        401,
+        50
+      ],
+      [
+        476,
+        186
+      ],
+      [
+        733,
+        187
+      ],
+      [
+        730,
+        333
+      ],
+      [
+        531,
+        349
+      ],
+      [
+        364,
+        332
+      ],
+      [
+        264,
+        317
+      ],
+      [
+        941,
+        322
+      ]
+    ]
+  },
+  "video_path": "teaser4/ducky-float.mp4",
+  "instruction": "",
+  "primary_points_by_frame": {
+    "7": [
+      [
+        268,
+        16
+      ],
+      [
+        353,
+        16
+      ],
+      [
+        740,
+        14
+      ],
+      [
+        900,
+        14
+      ]
+    ],
+    "18": [
+      [
+        931,
+        121
+      ],
+      [
+        688,
+        121
+      ],
+      [
+        511,
+        121
+      ],
+      [
+        283,
+        110
+      ],
+      [
+        549,
+        6
+      ],
+      [
+        657,
+        17
+      ],
+      [
+        771,
+        27
+      ],
+      [
+        461,
+        66
+      ]
+    ],
+    "36": [
+      [
+        821,
+        46
+      ],
+      [
+        636,
+        46
+      ],
+      [
+        445,
+        50
+      ],
+      [
+        401,
+        50
+      ],
+      [
+        476,
+        186
+      ],
+      [
+        733,
+        187
+      ],
+      [
+        730,
+        333
+      ],
+      [
+        531,
+        349
+      ],
+      [
+        364,
+        332
+      ],
+      [
+        264,
+        317
+      ],
+      [
+        941,
+        322
+      ]
+    ]
+  },
+  "primary_frames": [
+    7,
+    18,
+    36
+  ],
+  "first_appears_frame": 7
+}

sample/ducky-float/trimask_quadmask.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5754998a3cafd74d9136f392240b2a8932a7d2dff452a8c392800191043becf0
+size 366065

sample/spinner/input_video.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc5e7d25aaa1df31ace8c4e7aaf8bd19bbe0e7db9756e2d05360f6ef966120ec
+size 4652019

sample/spinner/prompt.json

@@ -0,0 +1,3 @@
+{
+  "bg": "a video of spinners spinning"
+}

sample/spinner/trimask_quadmask.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a432d080d4b3e306e5a1d25257074732ae2036473ac2402251b31e49c1c78e9
+size 1745346

sample/trampoline/input_video.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c82d889212e4d34d1348db7f1f85977e19c8a5bd6d56663f9eb7155aeb2bf700
+size 3690934

sample/trampoline/masks_output/sam3_masks/black_mask.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fdc08ea77db7442e41f4488dab89d713d58a2bea000f8f2abae77eca93e668a4
+size 2716337

sample/trampoline/masks_output/sam3_masks/find_objects.json

@@ -0,0 +1,61 @@
+{
+  "total_objects": 9,
+  "scaling_factor": 50,
+  "note": "Pixel values in find.png are scaled by 50x for visibility. Object 1 = value 50, Object 2 = value 100, etc.",
+  "objects": [
+    {
+      "id": 1,
+      "noun": "trampoline",
+      "description": "The main rectangular trampoline with a blue padded frame and white mesh bed in the foreground.",
+      "pixel_count": 389175
+    },
+    {
+      "id": 2,
+      "noun": "mat",
+      "description": "A blue rectangular mat lying flat on the trampoline frame on the left side.",
+      "pixel_count": 1660139
+    },
+    {
+      "id": 3,
+      "noun": "mat",
+      "description": "A large, thick blue mat standing vertically against the wall in the background.",
+      "pixel_count": 1660139
+    },
+    {
+      "id": 4,
+      "noun": "mat",
+      "description": "Another large, thick blue mat standing vertically against the wall on the left.",
+      "pixel_count": 1660139
+    },
+    {
+      "id": 5,
+      "noun": "mat",
+      "description": "A folded red and yellow mat placed on the right side of the trampoline area.",
+      "pixel_count": 1660139
+    },
+    {
+      "id": 6,
+      "noun": "mat",
+      "description": "A stack of dark green/grey mats piled up in the background.",
+      "pixel_count": 1660139
+    },
+    {
+      "id": 7,
+      "noun": "door",
+      "description": "A wooden door visible on the left side of the room.",
+      "pixel_count": 36410
+    },
+    {
+      "id": 8,
+      "noun": "mat",
+      "description": "A blue mat placed on top of the trampoline frame in the background.",
+      "pixel_count": 1660139
+    },
+    {
+      "id": 9,
+      "noun": "mat",
+      "description": "A red mat visible on the floor to the left.",
+      "pixel_count": 1660139
+    }
+  ]
+}

sample/trampoline/masks_output/sam3_masks/grey_mask.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1aaa8f346241e27c05f3bdab6a11a43dcbf6b18f5578e536035e6d83783d5a9d
+size 2392634

sample/trampoline/masks_output/sam3_masks/input_video.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c82d889212e4d34d1348db7f1f85977e19c8a5bd6d56663f9eb7155aeb2bf700
+size 3690934

sample/trampoline/masks_output/sam3_masks/vlm_analysis.json

@@ -0,0 +1,278 @@
+{
+  "primary_subject": {
+    "noun": "person",
+    "aliases": [
+      "man",
+      "gymnast",
+      "athlete"
+    ],
+    "position": [
+      [
+        1,
+        3
+      ],
+      [
+        1,
+        4
+      ],
+      [
+        2,
+        2
+      ],
+      [
+        2,
+        3
+      ],
+      [
+        2,
+        4
+      ],
+      [
+        3,
+        3
+      ],
+      [
+        3,
+        4
+      ],
+      [
+        4,
+        3
+      ],
+      [
+        4,
+        4
+      ],
+      [
+        5,
+        3
+      ],
+      [
+        5,
+        4
+      ],
+      [
+        5,
+        5
+      ]
+    ],
+    "reasoning": "The person's head starts in row 1 (cols 3-4), torso and arms occupy rows 2-3 (cols 2-4), and legs/feet extend down through rows 4-5 (cols 3-5)."
+  },
+  "integral_belongings": [],
+  "affected_subjects": [
+    {
+      "noun": "shadow",
+      "why": "cast by the person on the trampoline surface and surrounding mats",
+      "current_position": [
+        [
+          5,
+          3
+        ],
+        [
+          5,
+          4
+        ],
+        [
+          5,
+          5
+        ],
+        [
+          6,
+          4
+        ],
+        [
+          6,
+          5
+        ],
+        [
+          6,
+          6
+        ]
+      ],
+      "current_reasoning": "Shadows are visible directly under the feet on the blue padding at [5,3], [5,4], and [5,5], and a projected body shadow falls on the white trampoline mesh in row 6, columns 4-6.",
+      "contact_position": [
+        [
+          5,
+          3
+        ],
+        [
+          5,
+          4
+        ],
+        [
+          5,
+          5
+        ]
+      ],
+      "contact_reasoning": "The feet are in physical contact with the blue padding at [5,3], [5,4], and [5,5], which is where the contact shadow originates.",
+      "will_move": false,
+      "movement_path": [],
+      "final_position": [],
+      "movement_reasoning": "The shadow is a visual artifact caused by the person blocking light; removing the person removes the shadow."
+    },
+    {
+      "noun": "trampoline bed",
+      "why": "depressed and rebounding due to the person's weight and impact",
+      "current_position": [
+        [
+          6,
+          4
+        ],
+        [
+          6,
+          5
+        ],
+        [
+          6,
+          6
+        ],
+        [
+          6,
+          7
+        ],
+        [
+          6,
+          8
+        ],
+        [
+          6,
+          9
+        ],
+        [
+          7,
+          4
+        ],
+        [
+          7,
+          5
+        ],
+        [
+          7,
+          6
+        ],
+        [
+          7,
+          7
+        ],
+        [
+          7,
+          8
+        ],
+        [
+          7,
+          9
+        ]
+      ],
+      "current_reasoning": "The white mesh surface of the trampoline spans rows 6-7 and columns 4-9. It is currently reacting to the gymnast's movement.",
+      "contact_position": [],
+      "contact_reasoning": "In this specific frame, the person's feet are positioned on the blue padding frame (rows 5, cols 3-5), not directly touching the white mesh bed surface, though the bed is reacting to the motion.",
+      "will_move": true,
+      "movement_path": [
+        [
+          6,
+          4
+        ],
+        [
+          6,
+          5
+        ],
+        [
+          6,
+          6
+        ],
+        [
+          6,
+          7
+        ],
+        [
+          6,
+          8
+        ],
+        [
+          6,
+          9
+        ],
+        [
+          7,
+          4
+        ],
+        [
+          7,
+          5
+        ],
+        [
+          7,
+          6
+        ],
+        [
+          7,
+          7
+        ],
+        [
+          7,
+          8
+        ],
+        [
+          7,
+          9
+        ]
+      ],
+      "final_position": [
+        [
+          6,
+          4
+        ],
+        [
+          6,
+          5
+        ],
+        [
+          6,
+          6
+        ],
+        [
+          6,
+          7
+        ],
+        [
+          6,
+          8
+        ],
+        [
+          6,
+          9
+        ],
+        [
+          7,
+          4
+        ],
+        [
+          7,
+          5
+        ],
+        [
+          7,
+          6
+        ],
+        [
+          7,
+          7
+        ],
+        [
+          7,
+          8
+        ],
+        [
+          7,
+          9
+        ]
+      ],
+      "movement_reasoning": "The trampoline bed will cease vibrating/oscillating and settle into a perfectly flat, static position once the influence of the person is removed."
+    }
+  ],
+  "scene_description": "A gymnasium scene featuring a trampoline with a white mesh bed and blue padding, surrounded by other gymnastic equipment. Without the person, the trampoline bed would appear static and flat.",
+  "camera_motion_note": "The camera is static throughout the shot.",
+  "confidence": 0.95,
+  "grid_rows": 8,
+  "grid_cols": 14
+}

sample/trampoline/prompt.json

@@ -0,0 +1,3 @@
+{
+  "bg": "A gymnasium scene featuring a trampoline with a white mesh bed and blue padding, surrounded by other gymnastic equipment. Without the person, the trampoline bed would appear static and flat."
+}

sample/trampoline/trimask_quadmask.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e6d5293398af384dd9e90d4cb34b5e6d054ed98fc573ae0ac1260cf0ba83157
+size 2714264