Skip to content

Command Line Options

Visionatrix supports various command-line options, including most of the options provided by ComfyUI. You can specify these options when starting Visionatrix manually.

For example:

python3 -m visionatrix --verbose=WARNING run --ui --disable-smart-memory

Below are the command-line options related to Visionatrix and ComfyUI.

Common Options for Multiple Commands

The following options can be specified for the install, update, install-flow, orphan-models, and openapi commands:

  • --backend_dir=BACKEND_DIR: Directory for the folder with ComfyUI. Default: vix_backend
  • --flows_dir=FLOWS_DIR: Directory for the flows. Default: vix_flows
  • --models_dir=MODELS_DIR: Directory for the models. Default: vix_models

The run Command

Starts the ComfyUI and Visionatrix backends.

Syntax

python3 -m visionatrix [--verbose=LEVEL] run [options]

Options

Visionatrix-Specific Options

  • --host=HOST: Host to listen on (DEFAULT or SERVER mode).

    Info

    This corresponds to ComfyUI's --listen argument.

  • --port=PORT: Port to listen on (DEFAULT or SERVER mode).

  • --server=SERVER_ADDRESS: Address of Vix Server (WORKER mode).

  • --mode {DEFAULT,WORKER,SERVER}: Visionatrix operating mode.

    Choices: DEFAULT, WORKER, SERVER

  • --ui [UI_DIR]: Enable WebUI (DEFAULT or SERVER mode).

    • If --ui is provided without a value, the default UI is enabled.
    • If --ui is provided with a directory, it specifies the UI directory.
  • --tasks_files_dir=FILES_DIR: Directory for input/output files. Default: vix_task_files
  • --disable-device-detection: Disable automatic device detection.

  • --verbose [LEVEL]: Set the logging level.

    Choices: DEBUG, INFO, WARNING, ERROR, CRITICAL.

    Default: INFO or the value of the LOG_LEVEL environment variable.

    Note

    Note: The --verbose option should be specified before any command (e.g., run, install, update).

  • --backend_dir=BACKEND_DIR: Directory for the folder with ComfyUI. Default: vix_backend

  • --flows_dir=FLOWS_DIR: Directory for the flows. Default: vix_flows
  • --models_dir=MODELS_DIR: Directory for the models. Default: vix_models

ComfyUI Options

Visionatrix supports most of ComfyUI's command-line options for the run command. You can pass ComfyUI options when starting Visionatrix, and they will be forwarded to ComfyUI.

Note: The following ComfyUI options are not supported in Visionatrix:

  • --tls-keyfile
  • --tls-certfile
  • --enable-cors-header
  • --verbose (Visionatrix uses its own --verbose option)
  • --dont-print-server
  • --quick-test-for-ci
  • --windows-standalone-build
  • --auto-launch
  • --disable-auto-launch
  • --multi-user (Visionatrix always supports multiple users)

All other ComfyUI options can be used. Below is a list of supported ComfyUI options:

Device and Performance Options
  • --cuda-device DEVICE_ID: Set the ID of the CUDA device this instance will use.
  • --cuda-malloc: Enable cudaMallocAsync (enabled by default for PyTorch 2.0 and up).
  • --disable-cuda-malloc: Disable cudaMallocAsync.
  • --force-fp32: Force fp32 precision (If this makes your GPU work better, please report it).
  • --force-fp16: Force fp16 precision.
UNET Precision Options
  • --bf16-unet: Run the UNET in bf16 precision. This should only be used for testing purposes.
  • --fp16-unet: Store UNET weights in fp16 precision.
  • --fp8_e4m3fn-unet: Store UNET weights in fp8_e4m3fn format.
  • --fp8_e5m2-unet: Store UNET weights in fp8_e5m2 format.
VAE Precision Options
  • --fp16-vae: Run the VAE in fp16 precision (might cause black images).
  • --fp32-vae: Run the VAE in full precision fp32.
  • --bf16-vae: Run the VAE in bf16 precision.
  • --cpu-vae: Run the VAE on the CPU.
Text Encoder Precision Options
  • --fp8_e4m3fn-text-enc: Store text encoder weights in fp8 (e4m3fn variant).
  • --fp8_e5m2-text-enc: Store text encoder weights in fp8 (e5m2 variant).
  • --fp16-text-enc: Store text encoder weights in fp16.
  • --fp32-text-enc: Store text encoder weights in fp32.
Other Performance Options
  • --force-channels-last: Force channels-last format when inferring the models.
  • --disable-ipex-optimize: Disable ipex.optimize when loading models with Intel GPUs.
Caching Options
  • --cache-classic: Use the old style (aggressive) caching.
  • --cache-lru N: Use LRU caching with a maximum of N node results cached. May use more RAM/VRAM.
Attention Optimization Options
  • --use-split-cross-attention: Use the split cross-attention optimization. Ignored when xformers is used.
  • --use-quad-cross-attention: Use the sub-quadratic cross-attention optimization. Ignored when xformers is used.
  • --use-pytorch-cross-attention: Use the new PyTorch 2.0 cross-attention function.
  • --disable-xformers: Disable xformers.
  • --force-upcast-attention: Force enable attention upcasting (please report if it fixes black images).
  • --dont-upcast-attention: Disable all upcasting of attention. Should be unnecessary except for debugging.
Memory and Device Options
  • --gpu-only: Store and run everything (text encoders/CLIP models, etc.) on the GPU.
  • --highvram: Keep models in GPU memory instead of unloading to CPU memory after use.
  • --normalvram: Force normal VRAM usage if low VRAM gets automatically enabled.
  • --lowvram: Split the UNET into parts to use less VRAM.
  • --novram: Use when --lowvram isn't enough.
  • --cpu: Use the CPU for everything (slow).
  • --reserve-vram VRAM_GB: Set the amount of VRAM in GB you want to reserve for use by your OS/other software.
  • --disable-smart-memory: Force ComfyUI to aggressively offload to regular RAM instead of keeping models in VRAM when it can.
Other Options
  • --fast: Enable some untested and potentially quality-deteriorating optimizations.

Examples

Running Visionatrix with Specific GPU Settings

To run Visionatrix using the first CUDA device and enable split cross-attention optimization:

python3 -m visionatrix run --cuda-device 0 --use-split-cross-attention

Running in WORKER Mode

To run Visionatrix in WORKER mode, connecting to a Vix Server:

python3 -m visionatrix run --mode WORKER --server http://your_vix_server_address

Enabling the Web UI

To start Visionatrix with the default Web UI:

python3 -m visionatrix run --ui

To specify a custom UI directory:

python3 -m visionatrix run --ui /path/to/your/ui_directory

The install Command

Performs cleanup and initialization.

Syntax

python3 -m visionatrix [--verbose=LEVEL] install [options]

Options

  • --backend_dir=BACKEND_DIR: Directory for the folder with ComfyUI. Default: vix_backend
  • --flows_dir=FLOWS_DIR: Directory for the flows. Default: vix_flows
  • --models_dir=MODELS_DIR: Directory for the models. Default: vix_models

Example

python3 -m visionatrix install

During installation, you will be prompted to confirm whether to clear models, flows, and backend folders.

The update Command

Performs an update to the latest version.

Syntax

python3 -m visionatrix [--verbose=LEVEL] update [options]

Options

  • --backend_dir=BACKEND_DIR: Directory for the folder with ComfyUI. Default: vix_backend
  • --flows_dir=FLOWS_DIR: Directory for the flows. Default: vix_flows
  • --models_dir=MODELS_DIR: Directory for the models. Default: vix_models

Example

python3 -m visionatrix update

The install-flow Command

Installs a flow by name, tag, or from a file. Useful for workers that do not have a user interface.

Syntax

python3 -m visionatrix [--verbose=LEVEL] install-flow [options]

Options

You must specify one of the following options:

  • --file=FILE_PATH: Path to comfyui_flow.json file or a directory containing flow files.
  • The file should contain a ComfyUI workflow with the metadata needed for Visionatrix.
  • --name=FLOW_NAME: Flow name mask of the flow(s).
  • This will install the flow by its ID, which is equal to its folder name here.
  • --tag=TAG: Flow tags mask of the flow(s).

Additional options:

  • --backend_dir=BACKEND_DIR: Directory for the folder with ComfyUI. Default: vix_backend
  • --flows_dir=FLOWS_DIR: Directory for the flows. Default: vix_flows
  • --models_dir=MODELS_DIR: Directory for the models. Default: vix_models

Examples

Installing from a File

python3 -m visionatrix install-flow --file=path_to_comfyui_flow.json

The file should contain a ComfyUI workflow with the metadata needed for Visionatrix.

Installing by Name

python3 -m visionatrix install-flow --name=photo_stickers

This will install the flow by its ID, which is equal to its folder name here.

Installing by Tag

python3 -m visionatrix install-flow --tag=your_tag

The create-user Command

Creates a new user.

Syntax

python3 -m visionatrix [--verbose=LEVEL] create-user [options]

Options

  • --name=USERNAME (required): User name (ID).
  • --password=PASSWORD (required): User password.
  • --full_name=FULL_NAME: Full user name. Default: John Doe.
  • --email=EMAIL: User's email address. Default: user@example.com.
  • --admin=BOOLEAN: Should the user be an admin. Default: True.
  • --disabled=BOOLEAN: Should the account be disabled. Default: False.

Example

python3 -m visionatrix create-user --name=username --password=userpassword --email=user@example.com

The orphan-models Command

Removes orphan models.

Syntax

python3 -m visionatrix [--verbose=LEVEL] orphan-models [options]

Options

  • --no-confirm: Do not ask for confirmation for each model.
  • --dry-run: Perform cleaning without actually removing models.
  • --include-useful-models: Include orphaned models that can be used in future flows for removal.
  • --backend_dir=BACKEND_DIR: Directory for the folder with ComfyUI. Default: vix_backend
  • --flows_dir=FLOWS_DIR: Directory for the flows. Default: vix_flows
  • --models_dir=MODELS_DIR: Directory for the models. Default: vix_models

Example

python3 -m visionatrix orphan-models --no-confirm --dry-run

The openapi Command

Generates OpenAPI specifications.

Syntax

python3 -m visionatrix [--verbose=LEVEL] openapi [options]

Options

  • --file=FILENAME: Filename to save. Default: openapi.json.
  • --indentation=SIZE: Indentation size. Default: 2.
  • --flows=FLOWS: Flows to include in OpenAPI specs (comma-separated list or *).
    • If --flows is *, include all endpoints and all installed flows.
    • If --flows is specified but empty (e.g., --flows=""), do not include any flows.
    • If --flows is a comma-separated list of flow names (e.g., --flows=flow1,flow2), include those flows.
  • --skip-not-installed: Skip flows that are not installed. Default: True.
  • --exclude-base: Exclude base application endpoints from OpenAPI specs.
  • --backend_dir=BACKEND_DIR: Directory for the folder with ComfyUI. Default: vix_backend
  • --flows_dir=FLOWS_DIR: Directory for the flows. Default: vix_flows
  • --models_dir=MODELS_DIR: Directory for the models. Default: vix_models

Examples

Generate OpenAPI Specs for All Endpoints and All Installed Flows

python3 -m visionatrix openapi --flows="*" --file=my_openapi.json

Generate OpenAPI Specs for Specific Flows Only

python3 -m visionatrix openapi --flows=flow1,flow2 --exclude-base --file=my_openapi.json

Generate OpenAPI Specs Without Any Flows

python3 -m visionatrix openapi --file=my_openapi.json