Error running launch.bat

[catclaw]

Hi!
I hope you're doing well.
I followed every instruction for installation, and when I type launch.bat, I get these errors:

J:\Nexfocus>launch.bat
========================================
  Nexfocus Launcher
========================================

[OK] Python 3.12.10
[OK] NVIDIA driver
[OK] Virtual environment (Python 3.12.10)
[OK] PyTorch 2.13.0+cu132 with CUDA
[OK] xformers
[OK] aria2c
[OK] uv

All checks passed. Launching Nexfocus ...

[System ARGV] ['J:\\Nexfocus\\launch.py']
Python 3.12.10 (tags/v3.12.10:0cc8128, Apr  8 2025, 12:21:36) [MSC v.1943 64 bit (AMD64)]
Nexfocus version: V1.0.0
Dependency management is handled manually in Colab cells.
W0816 19:36:48.286000 59156 venv\Lib\site-packages\torch\utils\flop_counter.py:29] triton not found; flop counting will not work for triton kernels
[Startup] Memory environment profile: Local Normal (name=local_normal, source=auto, ram=130940MB, vram=12227MB, colab=False)
[Startup] Memory policy summary: {'profile': 'local_normal', 'low_ram_headroom_mb': 4096.0, 'critical_ram_headroom_mb': 2048.0, 'checkpoint_switch_ram_headroom_mb': 6144.0, 'low_vram_threshold_mb': 8192.0, 'medium_vram_threshold_mb': 16384.0, 'linux_malloc_trim_enabled': False, 'linux_malloc_trim_trigger_mb': 2048.0, 'aggressive_checkpoint_switch_reclaim': False}
Traceback (most recent call last):
  File "J:\Nexfocus\launch.py", line 119, in <module>
    from modules.hash_cache import init_cache
  File "J:\Nexfocus\modules\hash_cache.py", line 8, in <module>
    from modules.util import sha256, HASH_SHA256_LENGTH, get_file_from_folder_list
  File "J:\Nexfocus\modules\util.py", line 8, in <module>
    import cv2
  File "J:\Nexfocus\venv\Lib\site-packages\cv2\__init__.py", line 181, in <module>
    bootstrap()
  File "J:\Nexfocus\venv\Lib\site-packages\cv2\__init__.py", line 175, in bootstrap
    if __load_extra_py_code_for_module("cv2", submodule, DEBUG):
       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "J:\Nexfocus\venv\Lib\site-packages\cv2\__init__.py", line 28, in __load_extra_py_code_for_module
    py_module = importlib.import_module(module_name)
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Python312\Lib\importlib\__init__.py", line 90, in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "J:\Nexfocus\venv\Lib\site-packages\cv2\gapi\__init__.py", line 323, in <module>
    cv.gapi.wip.GStreamerPipeline = cv.gapi_wip_gst_GStreamerPipeline
                                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AttributeError: partially initialized module 'cv2' has no attribute 'gapi_wip_gst_GStreamerPipeline' (most likely due to a circular import)

Thankful for any help.
Regards.

[MLuc24]

That error is coming from your OpenCV install rather than from anything you did during setup, and there's a line earlier in your own log that explains why it happened here — "Dependency management is handled manually in Colab cells". Fooocus normally runs prepare_environment() at startup, which installs everything from requirements_versions.txt into the venv, and that file pins one specific OpenCV build: opencv-contrib-python-headless==4.10.0.84. Note it's the contrib headless variant, not plain opencv-python. Whatever you're running has that step disabled, so nothing enforced the pin and your venv got whichever OpenCV pip happened to resolve.

The "partially initialized module" failure in cv2/gapi/__init__.py is what you get when the cv2 folder in site-packages isn't a single coherent install — most often because more than one OpenCV distribution has been installed over the top of each other. They all unpack into the same cv2/ directory, so the Python shims end up from one build while the compiled extension is from another, and the gapi bootstrap goes looking for a symbol that isn't in the binary it actually loaded.

So the first thing I'd check is pip list inside that venv, filtered for opencv. If you see more than one of opencv-python, opencv-python-headless, opencv-contrib-python, opencv-contrib-python-headless, that's your answer. Uninstall every one of them, confirm the cv2 folder is actually gone from venv\Lib\site-packages afterwards since a stale directory sometimes survives, then install just the pinned one.

One thing worth flagging separately: the launcher is printing "Nexfocus Launcher" and "Nexfocus version: V1.0.0", so what you're running is a fork rather than Fooocus itself. That's not a criticism, but it does mean the dependency handling has been changed from what's in this repo, and the people here can't really see what it does. If the pinned install doesn't clear it, the fork's own tracker is the better place, since the missing step is on their side.

[catclaw]

That error is coming from your OpenCV install rather than from anything you did during setup, and there's a line earlier in your own log that explains why it happened here — "Dependency management is handled manually in Colab cells". Fooocus normally runs prepare_environment() at startup, which installs everything from requirements_versions.txt into the venv, and that file pins one specific OpenCV build: opencv-contrib-python-headless==4.10.0.84. Note it's the contrib headless variant, not plain opencv-python. Whatever you're running has that step disabled, so nothing enforced the pin and your venv got whichever OpenCV pip happened to resolve.

The "partially initialized module" failure in cv2/gapi/__init__.py is what you get when the cv2 folder in site-packages isn't a single coherent install — most often because more than one OpenCV distribution has been installed over the top of each other. They all unpack into the same cv2/ directory, so the Python shims end up from one build while the compiled extension is from another, and the gapi bootstrap goes looking for a symbol that isn't in the binary it actually loaded.

So the first thing I'd check is pip list inside that venv, filtered for opencv. If you see more than one of opencv-python, opencv-python-headless, opencv-contrib-python, opencv-contrib-python-headless, that's your answer. Uninstall every one of them, confirm the cv2 folder is actually gone from venv\Lib\site-packages afterwards since a stale directory sometimes survives, then install just the pinned one.

One thing worth flagging separately: the launcher is printing "Nexfocus Launcher" and "Nexfocus version: V1.0.0", so what you're running is a fork rather than Fooocus itself. That's not a criticism, but it does mean the dependency handling has been changed from what's in this repo, and the people here can't really see what it does. If the pinned install doesn't clear it, the fork's own tracker is the better place, since the missing step is on their side.

Hi!

Thank you so much!
I'm so sorry for posting here!
Actually... I was going to post in the Nexfocus GitHub discussion, but for some strange reason, I ended up here!
The page where I clicked Ask the community for help, is still open in another tab on the Nexfocus GitHub page!

Anyhow, I checked with pip list, and I have these:

opencv-contrib-python 5.0.0.93
opencv-contrib-python-headless 5.0.0.93
opencv-python 4.11.0.86
opencv-python-headless 4.13.0.92

So - I'm supposed to uninstall them and check so the cv2 folder is deleted from site-packages (and if not, delete it manually)?

Hmm...
How did I end up here??! 😶
Thanks for being nice to me despite this - you're awesome! 🙏

With regards.