Hyperion 2 always fails with error on Mac

Chamberhouse · May 15, 2026, 10:49am

Topaz Video 1.6, Mac Studio M3 Ultra, macOS 26.5

Even tried a fresh install of TV 1.6 on a 2nd startup volume I use for such things.

No informative error is given, just the usual “Please contact support. - Collect the logs: Help > Logging > Get Logs For Support - Contact: Help > Get Support”

Here’s the last part of the log file, to exclude personal information such as email address:

Is Owned: True
Is LoggedIn: True
Is Pro: False
Is SubLegacy: True
Getting model for request: hyp-2, num_gpu: 0, gpu_id: 0
Detected platform: Darwin
Available GPUs: ['Apple MPS']
Detected platform: Darwin
GPU manufacture: apple
torch version: 2.10.0
[11:23:48.576] 🔧 Processing context initialized (device=mps)

2026-05-15 11-23-48.579 Thread: 0x1f8465e80 Info FF Process Output: 4 INFO:models.slp25mac.src.cr8lpb3112cm:Preparing runner...
[11:23:48.577] 🏃 Creating new runner

2026-05-15 11-23-48.589 Thread: 0x1f8465e80 Info FF Process Output: 4 [11:23:48.588] 🚀 Creating enhancement model structure

2026-05-15 11-23-48.652 Thread: 0x1f8465e80 Info FF Process Output: 4 [11:23:48.651] 🎨 Skipping codec weight loading (accelerated engines will be used)

2026-05-15 11-23-48.652 Thread: 0x1f8465e80 Info FF Process Output: 4 INFO:models.slp25mac.src.cr8lpb3112cm:Initializing codec...

2026-05-15 11-23-48.687 Thread: 0x1f8465e80 Info FF Process Output: 4 INFO:models.slp25mac.src.optimization.yv4lli5izmzg:Loading codec encoder...

2026-05-15 11-23-48.713 Thread: 0x1f8465e80 Info FF Process Output: 4 INFO:models.slp25mac.src.tplf777t5333:Accelerated post processing correction enabled

2026-05-15 11-23-49.033 Thread: 0x1f8465e80 Info FF Process Output: 4 INFO:models.slp25mac.src.optimization.yv4lli5izmzg:Loading codec decoder...

2026-05-15 11-23-49.337 Thread: 0x1f8465e80 Info FF Process Output: 4 INFO:models.slp25mac.src.optimization.yv4lli5izmzg:Codec ready

2026-05-15 11-23-49.338 Thread: 0x1f8465e80 Info FF Process Output: 4 INFO:models.slp25mac.src.cr8lpb3112cm:Codec encode/decode ready

2026-05-15 11-23-49.354 Thread: 0x1f8465e80 Info FF Process Output: 4 INFO:models.slp25mac.src.cr8lpb3112cm:Loading enhancement model...

2026-05-15 11-23-49.375 Thread: 0x1f8465e80 Info FF Process Output: 4 INFO:models.slp25mac.src.cr8lpb3112cm:Loading model components

2026-05-15 11-23-49.430 Thread: 0x1f8465e80 Info FF Process Output: 4 INFO:models.slp25mac.src.cr8lpb3112cm:Model components loaded

2026-05-15 11-23-49.431 Thread: 0x1f8465e80 Info FF Process Output: 4 INFO:models.slp25mac.src.cr8lpb3112cm:Model ready

2026-05-15 11-23-49.477 Thread: 0x1f8465e80 Info FF Process Output: 4 INFO:models.slp25mac.src.cr8lpb3112cm:Video Upscaler initialized successfully!

2026-05-15 11-23-49.477 Thread: 0x1f8465e80 Info FF Process Output: 4 INFO:models.slp25mac.src.cr8lpb3112cm:Input video: /Users/andy/Desktop/Topaz/Source.mp4
INFO:models.slp25mac.src.cr8lpb3112cm:Output path: /Users/andy/Desktop/Topaz/Source_hdr.mov

2026-05-15 11-23-49.477 Thread: 0x1f8465e80 Info FF Process Output: 4 INFO:models.slp25mac.src.cr8lpb3112cm:Resolution: 640x360
INFO:models.slp25mac.src.cr8lpb3112cm:color_correction=none output_codec=prores

2026-05-15 11-23-49.477 Thread: 0x1f8465e80 Info FF Process Output: 4 INFO:models.slp25mac.src.cr8lpb3112cm:Frame range: start=0, load_cap=63634

2026-05-15 11-23-49.478 Thread: 0x1f8465e80 Info FF Process Output: 4 {"timestamp": "2026-05-15 11:23:49,477", "level": "INFO", "message": {"status": "RUNNING", "frame": 0, "progress": 0, "message": "Processing started..."}}
INFO:interface:{"status": "RUNNING", "frame": 0, "progress": 0, "message": "Processing started..."}

2026-05-15 11-23-49.576 Thread: 0x1f8465e80 Info FF Process Output: 4 [11:23:49.575]  
[11:23:49.575]  ━━━━━━━━ Phase 1: Encoding video ━━━━━━━━

2026-05-15 11-23-49.576 Thread: 0x1f8465e80 Info FF Process Output: 4 [11:23:49.575] 🎨 Input: frames streamed)

2026-05-15 11-23-58.507 Thread: 0x1f8465e80 Info FF Process Output: 4 {"timestamp": "2026-05-15 11:23:58,506", "level": "INFO", "message": {"status": "RUNNING", "frame": 0, "progress": 0, "message": "Phase 1: Encoding"}}
INFO:interface:{"status": "RUNNING", "frame": 0, "progress": 0, "message": "Phase 1: Encoding"}

2026-05-15 11-24-07.021 Thread: 0x1f8465e80 Info FF Process Output: 4 ERROR:models.slp25mac.j6g6yz18z68x:Hyp2MLX processing failed
Traceback (most recent call last):
  File "models/slp25mac/j6g6yz18z68x.py", line 126, in models.slp25mac.j6g6yz18z68x.Hyp2MLX.__call__
  File "models/slp25mac/src/cr8lpb3112cm.py", line 1374, in models.slp25mac.src.cr8lpb3112cm.VideoUpscaler.__call__
  File "models/slp25mac/src/core/gcvtfpjcr1ru.py", line 778, in models.slp25mac.src.core.gcvtfpjcr1ru.process_video_pix_chunked
  File "models/slp25mac/src/core/gcvtfpjcr1ru.py", line 200, in models.slp25mac.src.core.gcvtfpjcr1ru.process_pix_chunk
  File "models/slp25mac/src/core/afkh106v53pv.py", line 320, in models.slp25mac.src.core.afkh106v53pv.encode_whole_video
  File "/Applications/Topaz Video.app/Contents/MacOS/neuroserver/lib/python3.12/site-packages/torch/utils/_contextlib.py", line 124, in decorate_context
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "models/slp25mac/src/cr8lpb3112cm.py", line 1244, in models.slp25mac.src.cr8lpb3112cm.VideoUpscaler.__call__._hybrid_vae_encode
  File "models/slp25mac/src/optimization/yv4lli5izmzg.py", line 1313, in models.slp25mac.src.optimization.yv4lli5izmzg.MLXVideoVAE.tiled_encode
  File "models/slp25mac/src/optimization/yv4lli5izmzg.py", line 1083, in models.slp25mac.src.optimization.yv4lli5izmzg._gpu_blend_paste
ValueError: [broadcast_shapes] Shapes (1,31,20,80,32) and (1,1,80,80,1) cannot be broadcast.

2026-05-15 11-24-07.102 Thread: 0x1f8465e80 Info FF Process Output: 4 Traceback (most recent call last):

2026-05-15 11-24-07.102 Thread: 0x1f8465e80 Info FF Process Output: 4   File "c2qwgy08lhyj.py", line 283, in c2qwgy08lhyj.Hercules.run
  File "models/slp25mac/j6g6yz18z68x.py", line 133, in models.slp25mac.j6g6yz18z68x.Hyp2MLX.__call__
  File "models/slp25mac/j6g6yz18z68x.py", line 126, in models.slp25mac.j6g6yz18z68x.Hyp2MLX.__call__
  File "models/slp25mac/src/cr8lpb3112cm.py", line 1374, in models.slp25mac.src.cr8lpb3112cm.VideoUpscaler.__call__
  File "models/slp25mac/src/core/gcvtfpjcr1ru.py", line 778, in models.slp25mac.src.core.gcvtfpjcr1ru.process_video_pix_chunked
  File "models/slp25mac/src/core/gcvtfpjcr1ru.py", line 200, in models.slp25mac.src.core.gcvtfpjcr1ru.process_pix_chunk
  File "models/slp25mac/src/core/afkh106v53pv.py", line 320, in models.slp25mac.src.core.afkh106v53pv.encode_whole_video
  File "/Applications/Topaz Video.app/Contents/MacOS/neuroserver/lib/python3.12/site-packages/torch/utils/_contextlib.py", line 124, in decorate_context
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "models/slp25mac/src/cr8lpb3112cm.py", line 1244, in models.slp25mac.src.cr8lpb3112cm.VideoUpscaler.__call__._hybrid_vae_encode
  File "models/slp25mac/src/optimization/yv4lli5izmzg.py", line 1313, in models.slp25mac.src.optimization.yv4lli5izmzg.MLXVideoVAE.tiled_encode
  File "models/slp25mac/src/optimization/yv4lli5izmzg.py", line 1083, in models.slp25mac.src.optimization.yv4lli5izmzg._gpu_blend_paste
ValueError: [broadcast_shapes] Shapes (1,31,20,80,32) and (1,1,80,80,1) cannot be broadcast.
ERROR:local_model_service:Error running video restoration: [broadcast_shapes] Shapes (1,31,20,80,32) and (1,1,80,80,1) cannot be broadcast.

2026-05-15 11-24-07.103 Thread: 0x1f8465e80 Info FF Process Output: 4 ERROR:local_model_service:single_shot_process failed with status: MODEL_INFERENCE_ERROR

2026-05-15 11-24-07.115 Thread: 0x1f8465e80 Info FF Process Output: 4 single_shot_process status: TopServingErrorCode.MODEL_INFERENCE_ERROR, {}

2026-05-15 11-24-07.484 Thread: 0x1f8465e80 Info process exited: 4 7 0
2026-05-15 11-24-07.484 Thread: 0x1f8465e80 Info Going to delete path: "/Users/andy/Desktop/Topaz/Source_hdr.mov"
2026-05-15 11-24-07.484 Thread: 0x1f8465e80 Info Close and delete path "/Users/andy/Desktop/Topaz/Source_hdr.mov"
2026-05-15 11-24-07.501 Thread: 0x1f8465e80 Info ~TProcess(): destroyed
2026-05-15 11-24-07.507 Thread: 0x1f8465e80 Warning qrc:/videomanager/TVideoManagerPane.qml:1232:29: QML WIconImage: Detected anchors on an item that is managed by a layout. This is undefined behavior; use Layout.alignment instead.
2026-05-15 11-26-08.847 Thread: 0x1f8465e80 Info Opening current log /Users/andy/Library/Application Support/Topaz Labs LLC/Topaz Video/logs/2026-05-15-11-07-38-Main.tzlog

Andy

Chamberhouse · May 15, 2026, 11:55am

Update - I’ve found just ONE source file where Hyperion 2 works. I can’t see any reason why, whereas every other fails. So at least I know that Hyperion 2 CAN work on my Mac.

jo.vo · May 15, 2026, 1:07pm

Tried it here on the M2 Ultra and got an error as well.

Chamberhouse · May 15, 2026, 5:33pm

Further update: I think I have found the problem - Hyperion 2 appears to require two conditions to be met for the source:

At least 640 x 480

AND

Square pixels.

This might be over simplified, but so far that has “solved” it for me. Nevertheless, if this is the case, then Topaz should have made that very clear in the first place.

kyle.topazlabs · May 15, 2026, 5:48pm

Thanks for the partial log and the updated information @Chamberhouse going to get this in front of the devs in a meeting later to have them look into this further.