# ComfyUI 运行日志 — 2D # 原文:/home/kf/comfy/ComfyUI/user/comfyui.log 第 480-594 行 # 输出:output/video/MiniMax_H3_00415_.mp4 # 提示词风格:Style: 2D traditional anime # 采样器 seed:612688989492486 ---------------------------------------------------------------------- [2026-11-16 16:35:21.632] got prompt [2026-11-16 16:35:21.680] Model storage policy: fast_disk=False paths=['/media/comfyui_storage/vae/MiniMax/minimax_h3_audio_vae_fp32.safetensors'] [2026-11-16 16:35:21.709] VAE load device: cuda:0, offload device: cpu, dtype: torch.float32 [2026-11-16 16:35:21.711] #2 [VAELoader]: 0.1秒 (累计0.1秒) [2026-11-16 16:35:21.735] Model storage policy: fast_disk=False paths=['/media/comfyui_storage/vae/MiniMax/minimax_h3_video_vae_fp16.safetensors'] [2026-11-16 16:35:21.805] VAE load device: cuda:0, offload device: cpu, dtype: torch.float16 [2026-11-16 16:35:21.806] #1 [VAELoader]: 0.1秒 (累计0.2秒) [2026-11-16 16:35:21.840] Found quantization metadata version 1 [2026-11-16 16:35:21.840] Detected mixed precision quantization [2026-11-16 16:35:21.840] Using mixed precision operations [2026-11-16 16:35:21.840] Native ops: float8_e4m3fn, int8_tensorwise, asym_w4a8_int8, mxfp8, convrot_w4a4, w6a8_int8, float8_e5m2, nvfp4 [2026-11-16 16:35:21.847] model weight dtype torch.bfloat16, manual cast: torch.bfloat16 [2026-11-16 16:35:21.848] model_type FLOW_AV [2026-11-16 16:35:21.848] Model storage policy: fast_disk=False paths=['/media/comfyui_storage/diffusion_models/MiniMax/minimax_h3_hybrid_fl2va_ref2va_b25-49-int8.safetensors'] [2026-11-16 16:35:22.364] #4 [UNETLoader]: 0.6秒 (累计0.7秒) [2026-11-16 16:35:22.455] #5 [LoraLoaderBypassModelOnly]: 0.1秒 (累计0.8秒) [2026-11-16 16:35:22.456] #84 [SelfLiftAvatarH3TST]: 0.0秒 (累计0.8秒) [2026-11-16 16:35:22.457] [sol_attn] chaining onto an existing attention override [2026-11-16 16:35:22.457] #87 [SolAttnMiniMax]: 0.0秒 (累计0.8秒) [2026-11-16 16:35:22.510] #47 [LoraLoaderModelOnly]: 0.1秒 (累计0.9秒) [2026-11-16 16:35:22.564] #83 [LoraLoaderModelOnly]: 0.1秒 (累计0.9秒) [2026-11-16 16:35:22.613] #45 [LoraLoaderModelOnly]: 0.0秒 (累计1.0秒) [2026-11-16 16:35:22.619] #76 [BasicScheduler]: 0.0秒 (累计1.0秒) [2026-11-16 16:35:22.625] #77 [H3SigmaRefiner]: 0.0秒 (累计1.0秒) [2026-11-16 16:35:22.626] #75 [KSamplerSelect]: 0.0秒 (累计1.0秒) [2026-11-16 16:35:22.648] #35 [LoadImage]: 0.0秒 (累计1.0秒) [2026-11-16 16:35:22.658] #6 [LoadImage]: 0.0秒 (累计1.0秒) [2026-11-16 16:35:22.705] Found quantization metadata version 1 [2026-11-16 16:35:22.705] [MultiGPU Core Patching] text_encoder_device_patched returning device: cuda:0 (current_text_encoder_device=cuda:0) [2026-11-16 16:35:22.705] Using MixedPrecisionOps for text encoder [2026-11-16 16:35:22.955] Model storage policy: fast_disk=False paths=['/media/comfyui_storage/text_encoders/MiniMax/qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors'] [2026-11-16 16:35:41.312] CLIP/text encoder model load device: cuda:0, offload device: cpu, current: cpu, dtype: torch.float16 [2026-11-16 16:35:41.313] #3 [CLIPLoader]: 18秒 (累计19秒) [2026-11-16 16:35:41.315] #27 [PrimitiveFloat]: 0.0秒 (累计19秒) [2026-11-16 16:35:41.316] #30 [ComfyMathExpression]: 0.0秒 (累计19秒) [2026-11-16 16:35:41.317] #29 [ResolutionSelector]: 0.0秒 (累计19秒) [2026-11-16 16:35:41.317] #49 [PrimitiveStringMultiline]: 0.0秒 (累计19秒) [2026-11-16 16:36:55.608] #14 [MiniMaxH3AudioConditioningT8]: 1分14秒 (累计1分33秒) - 显存 816.8MB [2026-11-16 16:36:55.615] #78 [ConditioningZeroOut]: 0.0秒 (累计1分33秒) [2026-11-16 16:36:55.621] [selflift-Avatar global] mode=global_stream token_chunk=4096 weight_cache=4.00 GiB; full-sequence video/audio masks enabled [2026-11-16 16:36:55.622] [selflift-Avatar global] attention override detected (make_override..override); disabling head/GPU streaming; Falling back to the full-sequence global path with chunked MLP. Expect higher VRAM use; disable TST/other attention overrides for global_stream/global_ram. [2026-11-16 16:36:55.622] [selflift-Avatar global chunking v2.1] chunk=4096; native sparse blocks=0/50; native backend fallback retained; native memory planning [2026-11-16 16:36:55.622] [selflift-Avatar plan] low_latent=(1, 24, 87, 24, 44) target_latent=(1, 24, 87, 48, 86) spatial_lift=(2.0000, 1.9545) low_nfe=6 high_nfe=3 sigma_prediction=0.87804878 sigma_resume=0.80000001 rho=0.0000 weights=(0.5000, 1.0000) direct_lift=external pixel_anchor=False cfg=1.0000 mask=none hires_model=same [2026-11-16 16:36:55.639] [selflift-Avatar timing] low_resolution start (wall time; no forced CUDA sync) latent=(24, 44) pixels=(384, 704) [2026-11-16 16:38:34.536] [selflift-Avatar timing] low_resolution step 1/6 (includes setup): 98.897s [2026-11-16 16:38:41.335] [selflift-Avatar timing] low_resolution step 2/6: 6.799s [2026-11-16 16:38:44.302] [selflift-Avatar timing] low_resolution step 3/6: 2.967s [2026-11-16 16:38:47.316] [selflift-Avatar timing] low_resolution step 4/6: 3.014s [2026-11-16 16:38:50.335] [selflift-Avatar timing] low_resolution step 5/6: 3.019s [2026-11-16 16:38:53.355] [selflift-Avatar timing] low_resolution step 6/6: 3.020s [2026-11-16 16:38:53.355] 100%|█████████████████████████████████████████████| 6/6 [00:25<00:00, 3.02s/it] 100%|█████████████████████████████████████████████| 6/6 [00:25<00:00, 4.27s/it] [2026-11-16 16:38:53.357] [selflift-Avatar timing] low_resolution total: 117.719s; after last mark: 0.002s [2026-11-16 16:38:53.358] [selflift-Avatar timing] transition start (wall time; no forced CUDA sync) latent=(48, 86) pixels=(768, 1376) [2026-11-16 16:38:53.401] [selflift-Avatar timing] transition prepare_endpoint: 0.043s [2026-11-16 16:38:53.451] [SelfLift upscaler] model=minimax_h3_latent_upscaler_3d_fp16.safetensors dtype=torch.float16 input=(1, 24, 87, 24, 44) target_hw=(48, 86) spatial_lift=(2.0000, 1.9545) scale_embedding=0.9773 mode=chunked chunk=32 overlap=5 windows=3 max_input_window=52 budget_window=52 estimated_workspace=1677.00 MiB [2026-11-16 16:38:58.280] [selflift-Avatar timing] transition paired_lifts: 4.880s [2026-11-16 16:38:58.284] [selflift-Avatar timing] transition correction_and_debug: 0.003s [2026-11-16 16:38:58.334] [selflift-Avatar timing] transition renoise: 0.050s [2026-11-16 16:38:58.334] [selflift-Avatar timing] transition total: 4.976s; after last mark: 0.000s [2026-11-16 16:38:58.334] [selflift-Avatar timing] high_resolution start (wall time; no forced CUDA sync) latent=(48, 86) pixels=(768, 1376) [2026-11-16 16:38:58.334] [selflift-Avatar plan] high-resolution global context enabled; mode=global_stream token_chunk=4096 weight_cache=4.00 GiB; masks remain full-sequence [2026-11-16 16:39:13.739] [selflift-Avatar timing] high_resolution step 1/3 (includes setup): 15.404s [2026-11-16 16:39:30.285] [selflift-Avatar timing] high_resolution step 2/3: 16.547s [2026-11-16 16:39:59.025] FETCH ComfyRegistry Data [DONE] [2026-11-16 16:39:59.140] [ComfyUI-Manager] default cache updated: https://api.comfy.org/nodes [2026-11-16 16:39:59.187] FETCH DATA from: /home/kf/comfy/ComfyUI/user/__manager/cache/1514988643_custom-node-list.json [DONE] [2026-11-16 16:39:59.209] [ComfyUI-Manager] All startup tasks have been completed. [2026-11-16 16:39:59.209] Exception in thread Thread-1 (): [2026-11-16 16:39:59.218] Traceback (most recent call last): [2026-11-16 16:39:59.218] File "/usr/lib/python3.14/threading.py", line 1082, in _bootstrap_inner self._context.run(self.run) ~~~~~~~~~~~~~~~~~^^^^^^^^^^ [2026-11-16 16:39:59.219] File "/usr/lib/python3.14/threading.py", line 1024, in run self._target(*self._args, **self._kwargs) ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ [2026-11-16 16:39:59.219] File "/home/kf/下载/comfy-env/lib/python3.14/site-packages/comfyui_manager/glob/manager_server.py", line 2172, in  threading.Thread(target=lambda: asyncio.run(default_cache_update())).start() ~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^ [2026-11-16 16:39:59.219] File "/usr/lib/python3.14/asyncio/runners.py", line 203, in run with Runner(debug=debug, loop_factory=loop_factory) as runner: ~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ [2026-11-16 16:39:59.219] File "/usr/lib/python3.14/asyncio/runners.py", line 63, in __exit__ self.close() ~~~~~~~~~~^^ [2026-11-16 16:39:59.219] File "/usr/lib/python3.14/asyncio/runners.py", line 73, in close loop.run_until_complete( ~~~~~~~~~~~~~~~~~~~~~~~^ loop.shutdown_default_executor(constants.THREAD_JOIN_TIMEOUT)) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ [2026-11-16 16:39:59.219] File "/home/kf/下载/comfy-env/lib/python3.14/site-packages/nest_asyncio.py", line 98, in run_until_complete return f.result() ~~~~~~~~^^ [2026-11-16 16:39:59.219] File "/usr/lib/python3.14/asyncio/futures.py", line 208, in result raise self._exception.with_traceback(self._exception_tb) [2026-11-16 16:39:59.219] File "/usr/lib/python3.14/asyncio/tasks.py", line 289, in __step_run_and_handle_result result = coro.send(None) [2026-11-16 16:39:59.219] File "/usr/lib/python3.14/asyncio/base_events.py", line 609, in shutdown_default_executor async with timeouts.timeout(timeout): ~~~~~~~~~~~~~~~~^^^^^^^^^ [2026-11-16 16:39:59.219] File "/usr/lib/python3.14/asyncio/timeouts.py", line 88, in __aenter__ raise RuntimeError("Timeout should be used inside a task") [2026-11-16 16:39:59.219] RuntimeError: Timeout should be used inside a task [2026-11-16 16:39:59.219] [selflift-Avatar timing] high_resolution step 3/3: 71.475s [2026-11-16 16:40:41.761] 100%|█████████████████████████████████████████████| 3/3 [01:44<00:00, 71.47s/it] 100%|█████████████████████████████████████████████| 3/3 [01:44<00:00, 34.86s/it] [2026-11-16 16:40:43.268] [selflift-Avatar timing] high_resolution total: 104.934s; after last mark: 1.508s [2026-11-16 16:40:43.331] Comfy model compiler graph breaks: 9, rogues: 0 [2026-11-16 16:40:43.351] #89 [SelfLiftAvatarH3Sampler]: 3分47秒 (累计5分21秒) - 显存 2.5GB [2026-11-16 16:41:38.437] #20 [MiniMaxH3AVDecodeT8]: 55秒 (累计6分16秒) - 显存 676.8MB [2026-11-16 16:41:38.442] #51 [CreateVideo]: 0.0秒 (累计6分16秒) [2026-11-16 16:41:43.749] #53 [SaveVideo]: 5.3秒 (累计6分22秒) [2026-11-16 16:41:43.787] Prompt executed in 382.15 seconds