Jasna v0.11.2 LTX模型简单测试以及报错问题

qiuhuahua

新模型出来，闲鱼入手激活码，支持一下作者
电脑配置普通，CPU AMD 5700X 内存 DDR4 3200 32G 显卡3060ti 8G
跑了一本片源 1080P 5M码率 大小4点几个G  时长120分，共耗时20小时。
1比10的比例自娱自乐还是能接受滴，关键是全默认设置效果OK

问题是现在电脑跑了Jasna完全干不了其他的，打开网页看看剧也会影响Jasna报错


08:08:42 PM ERROR   Failed to process TEST1111.mp4: LTX inference failed in the protected worker: ProtectionError: LTX private load rejected: native block commit (core.py:177 <- core.py:218 <- protected_tensors.py:242)
Traceback (most recent call last):
  File "E:\jasna-0.11.2\jasna\gui\processor.py", line 256, in _process_job
  File "E:\jasna-0.11.2\jasna\gui\processor.py", line 342, in _run_pipeline
  File "E:\jasna-0.11.2\jasna\gui\processor.py", line 435, in _run_video_job
  File "E:\jasna-0.11.2\jasna\pipeline.py", line 606, in run
  File "E:\jasna-0.11.2\jasna\pipeline.py", line 307, in _run_ltx
  File "E:\jasna-0.11.2\jasna\ltx\restore.py", line 979, in restore_video
  File "E:\jasna-0.11.2\jasna\ltx\restore.py", line 920, in restore_spans
  File "E:\jasna-0.11.2\jasna\ltx\restore.py", line 938, in _restore_spans
  File "E:\jasna-0.11.2\jasna\ltx\inference.py", line 129, in load_transformer
  File "E:\jasna-0.11.2\jasna\ltx\inference.py", line 102, in request
  File "E:\jasna-0.11.2\jasna\protection\ltx_worker.py", line 304, in request
  File "E:\jasna-0.11.2\jasna\protection\ltx_worker.py", line 254, in _exchange
jasna.protection.ltx_worker.WorkerError: LTX inference failed in the protected worker: ProtectionError: LTX private load rejected: native block commit (core.py:177 <- core.py:218 <- protected_tensors.py:242)


08:13:06 PM ERROR   Failed to process TEST1111.mp4: LTX inference failed in the protected worker: ProtectionError: LTX private load rejected: native block commit (core.py:177 <- core.py:218 <- protected_tensors.py:242)
Traceback (most recent call last):
  File "E:\jasna-0.11.2\jasna\gui\processor.py", line 256, in _process_job
  File "E:\jasna-0.11.2\jasna\gui\processor.py", line 342, in _run_pipeline
  File "E:\jasna-0.11.2\jasna\gui\processor.py", line 435, in _run_video_job
  File "E:\jasna-0.11.2\jasna\pipeline.py", line 606, in run
  File "E:\jasna-0.11.2\jasna\pipeline.py", line 307, in _run_ltx
  File "E:\jasna-0.11.2\jasna\ltx\restore.py", line 979, in restore_video
  File "E:\jasna-0.11.2\jasna\ltx\restore.py", line 920, in restore_spans
  File "E:\jasna-0.11.2\jasna\ltx\restore.py", line 938, in _restore_spans
  File "E:\jasna-0.11.2\jasna\ltx\inference.py", line 129, in load_transformer
  File "E:\jasna-0.11.2\jasna\ltx\inference.py", line 102, in request
  File "E:\jasna-0.11.2\jasna\protection\ltx_worker.py", line 304, in request
  File "E:\jasna-0.11.2\jasna\protection\ltx_worker.py", line 254, in _exchange
jasna.protection.ltx_worker.WorkerError: LTX inference failed in the protected worker: ProtectionError: LTX private load rejected: native block commit (core.py:177 <- core.py:218 <- protected_tensors.py:242)

大概率是显存内存太小了，但我觉得还有优化空间吧，使用高峰时内存占用20G，还有12G空闲。但愿作者能看到，谢谢

qiuhuahua

AI说有可能是权限问题，以管理员运行结果一样报错的

Deadman12

我本來也想用LTX的，但看看需要那麼長時間處理，我就放棄了，感覺新片子一直出，時間不夠花啊

qiuhuahua

Deadman12 发表于 2026-10-18 21:23
我本來也想用LTX的，但看看需要那麼長時間處理，我就放棄了，感覺新片子一直出，時間不夠花啊 ...

睡觉前挂机，第二天晚上应该能出来，挑几个片子自己玩玩还是可以的

15029976885

1111111111111111