Capturing execution frames: scene_seg_interactive.json [ComfyUI] [comfy-env] [VRAM] Before offload 'Sam3VideoInferenceWithInstanceInteractivity_1' (3334 MB) -> cpu: 4191 / 24124 MB [ComfyUI] [ComfyUI-SAM3] Health check: ping (timeout=5.0s)... [ComfyUI] [ComfyUI-SAM3] Health check: ok [ComfyUI] [SubprocessWorker] call_id=13 sending model_to_device [ComfyUI] [worker:ComfyUI-SAM3] [VRAM] -303MB (now 3148MB) reserved=3552MB peak=3833MB [ComfyUI] [worker:ComfyUI-SAM3] [VRAM] -1447MB (now 1702MB) reserved=3552MB peak=3833MB [ComfyUI] [worker:ComfyUI-SAM3] [VRAM] -1458MB (now 244MB) reserved=3552MB peak=3833MB [ComfyUI] [comfy-env] [VRAM] After offload 'Sam3VideoInferenceWithInstanceInteractivity_1' (3334 MB): 863 / 24124 MB Frame 1 saved (frame_000.jpg, t=0.0s) Validating workflow... [ComfyUI] === /validate endpoint called === Queuing workflow for execution... [ComfyUI] Received JSON with keys: ['prompt'] [ComfyUI] Prompt has 4 nodes [ComfyUI] Node 1: LoadImage [ComfyUI] Inputs: {"image": "image.png"} [ComfyUI] Node 4: MaskPreview [ComfyUI] Inputs: {"mask": ["26", 0]} [ComfyUI] Node 12: LoadSAM3Model [ComfyUI] Inputs: {"precision": "auto", "compile": false} [ComfyUI] Node 26: SAM3InteractiveCollector [ComfyUI] Inputs: {"multi_prompts_store": "[{\"positive_points\":[{\"x\":3348.288109906738,\"y\":2452.2208696844737}],\"negative_points\":[],\"positive_boxes\":[],\"negative_boxes\":[],\"name\":\"Prompt 1\"},{\"positiv [ComfyUI] Generated prompt_id: 6f897af2-3eda-4fb8-883a-1f7fd9738280 [ComfyUI] Calling execution.validate_prompt()... [ComfyUI] Validation result: valid=True [ComfyUI] === Validation PASSED === [ComfyUI] got prompt [ComfyUI] [comfy-env] Running LoadSAM3Model: precision: 'auto', compile: False [ComfyUI] [VRAM] Before LoadSAM3Model: 863 / 24124 MB [ComfyUI] [SubprocessWorker] call_method: nodes.load_model.LoadSAM3Model.EXECUTE_NORMALIZED [ComfyUI] [SubprocessWorker] acquired lock, ensuring started... [ComfyUI] [ComfyUI-SAM3] Health check: ping (timeout=5.0s)... [ComfyUI] [ComfyUI-SAM3] Health check: ok [ComfyUI] [SubprocessWorker] worker started/confirmed [ComfyUI] [SubprocessWorker] serializing kwargs to shm... [ComfyUI] [SubprocessWorker] created 0 shm blocks [ComfyUI] [SubprocessWorker] sending request via socket... [ComfyUI] [SubprocessWorker] call_id=14 sending call_method [ComfyUI] [SubprocessWorker] got response: ok [ComfyUI] [comfy-env] LoadSAM3Model done (0.07s): result: NodeOutput [ComfyUI] [VRAM] After LoadSAM3Model: 863 / 24124 MB [ComfyUI] [comfy-env] Running SAM3InteractiveCollector: multi_prompts_store: '[{"positive_points":[{"x":3348.288109906738,"y":2452.220..., sam3_model_config: dict, image: torch.float32 [1,4480,6720,3] cpu, unique_id: '26' [ComfyUI] [VRAM] Before SAM3InteractiveCollector: 863 / 24124 MB [ComfyUI] [SubprocessWorker] call_method: nodes.sam3_interactive.SAM3InteractiveCollector.EXECUTE_NORMALIZED [ComfyUI] [SubprocessWorker] acquired lock, ensuring started... [ComfyUI] [ComfyUI-SAM3] Health check: ping (timeout=5.0s)... [ComfyUI] [ComfyUI-SAM3] Health check: ok [ComfyUI] [SubprocessWorker] worker started/confirmed [ComfyUI] [comfy-env] PRE-SERIALIZE 'image' shape: torch.Size([1, 4480, 6720, 3]) [ComfyUI] [SubprocessWorker] serializing kwargs to shm... [ComfyUI] [SubprocessWorker] created 0 shm blocks [ComfyUI] [SubprocessWorker] sending request via socket... [ComfyUI] [SubprocessWorker] call_id=15 sending call_method [ComfyUI] [worker:ComfyUI-SAM3] [comfy-env] _from_shm got dict with keys: ['__type__', 'strategy', 'parent_pid', 'fd', 'storage_size'] [ComfyUI] [worker:ComfyUI-SAM3] [comfy-env] DESERIALIZE TensorRef: tensor_size=[1, 4480, 6720, 3] [ComfyUI] [worker:ComfyUI-SAM3] [comfy-env] DESERIALIZED tensor shape: torch.Size([1, 4480, 6720, 3]) Frame 2 saved (frame_001.jpg, t=2.4s) [capture-periodic] freeze=3ms shot=215ms unfreeze=4ms saved=True [ComfyUI] [SubprocessWorker] callback 'request_vram_budget' from call_id=15 [ComfyUI] [comfy-env] VRAM request: 1.84GB, free before eviction: 24.39GB [ComfyUI] [comfy-env] VRAM after eviction: 24.39GB [ComfyUI] [worker:ComfyUI-SAM3] [VRAM] +701MB (now 945MB) reserved=992MB peak=3833MB [ComfyUI] [SubprocessWorker] callback 'report_progress' from call_id=15 [ComfyUI] [worker:ComfyUI-SAM3] [VRAM] +1204MB (now 2150MB) reserved=2720MB peak=3833MB Frame 3 saved (frame_002.jpg, t=4.8s) [capture-periodic] freeze=4ms shot=224ms unfreeze=6ms saved=True [capture-loop] iter=50 t=6.4s state={'complete': False, 'error': None, 'executedCount': 0, 'wsState': 1} eval_ms=4 ram=12.2/31.2GB (39%) swap=0.0/32.0GB (0%) cpu=6% browser_rss=1.35GB [ComfyUI] [SubprocessWorker] got response: ok [ComfyUI] [comfy-env] Created 1 model patchers: ['Sam3VideoInferenceWithInstanceInteractivity_2'] [ComfyUI] [comfy-env] SAM3InteractiveCollector done (6.57s): result: NodeOutput [ComfyUI] [VRAM] After SAM3InteractiveCollector: 3369 / 24124 MB [ComfyUI] [VRAM] Loaded: SubprocessModel (1756 MB) [ComfyUI] Prompt executed in 8.40 seconds [ComfyUI] [comfy-env] [VRAM] Before offload 'Sam3VideoInferenceWithInstanceInteractivity_2' (1756 MB) -> cpu: 3369 / 24124 MB [ComfyUI] [ComfyUI-SAM3] Health check: ping (timeout=5.0s)... [ComfyUI] [ComfyUI-SAM3] Health check: ok [ComfyUI] [SubprocessWorker] call_id=16 sending model_to_device [ComfyUI] [worker:ComfyUI-SAM3] [VRAM] -501MB (now 1649MB) reserved=2720MB peak=3833MB Node executed (1 total), capturing... [ComfyUI] [comfy-env] [VRAM] After offload 'Sam3VideoInferenceWithInstanceInteractivity_2' (1756 MB): 1257 / 24124 MB Frame 4 saved (frame_003.jpg, t=9.5s) [capture-node] freeze=380ms shot=1072ms unfreeze=10ms saved=True Node executed (2 total), capturing... [capture-node] freeze=3ms shot=296ms unfreeze=0ms saved=False Execution complete (t=11.3s) Captured 4 unique frames over 11.63s [debug] iframes: 0 [timing] trigger_3d_previews: 0.0s Saved high-quality screenshot: scene_seg_interactive_executed.png [timing] final_screenshot: 0.6s Captured 5 video frames