Skip to content

Troubleshooting Guide

Jump to bottom
Akash Bora edited this page Jul 5, 2026 · 5 revisions

RemObj-Fuse Troubleshooting Guide

Welcome to the RemObj-Fuse troubleshooting wiki. This guide helps you resolve common installation and setup errors.

➤ Installation Instructions

Before troubleshooting, ensure you've followed the standard installation process:

➤ Python Installation

Requirements

➤ Automatic Setup Issues

Setup UI Not Appearing

Try these alternatives:

  1. Click the Open RemObj Setup button in the RemObj node inspector within DaVinci Resolve.
  2. Navigate to the RemObj folder and run remobj_manager.py directly (How to open Python files).

Manual Installation

If automatic setup fails, follow these steps:

Install Required Packages

Run one of the following commands in your terminal or command prompt to install dependencies:

pip install onnxruntime opencv-python numpy pillow

Or if the above doesn't work:

python -m pip install onnxruntime opencv-python numpy pillow

Or for Python 3:

python3 -m pip install onnxruntime opencv-python numpy pillow

Optional GPU Support:

  • CUDA: pip install onnxruntime-gpu opencv-python numpy pillow
  • AMD/ROCm: pip install onnxruntime-rocm opencv-python numpy pillow If GPU packages fail to load or result in conflicts, switch back to the standard CPU installation.

Download AI Models

  1. Make sure the LaMa model file lama_sim_fp32.onnx is placed inside the RemObj folder.
  2. You can download the model directly from the RemObj-Fuse Releases or the main repository.
  3. If placing manually, ensure the file path is: Fuses/RemObj/lama_sim_fp32.onnx

➤ Processing Errors

Python Executable Not Found (Windows)

Error: "Windows can't find pythonw"

Solution (requires RemObj-Fuse v0.1+):

  1. Open the RemObj node in DaVinci Resolve
  2. Click the Python Path option in the inspector
  3. Change pythonw to python

Alternative:

  1. Click the browse box next to the Python Path input and select your python.exe (or pythonw.exe if available)
  2. To find your Python installation directory, run in cmd:
where python
image

Image Save Errors

Error: "Failed to save input image" or similar clip writing errors

Possible Cause: DaVinci Resolve has limitations writing temporary files to the Temp:/ directory, or there are issues with certain vertical/high-resolution aspect ratios (like 4K vertical formats).

Solution:

  1. Ensure the DaVinci Resolve temp directory has write permissions.
  2. Scale down the image resolution in Fusion before passing it to the RemObj node.
  3. Try different aspect ratios or avoid non-ASCII characters in the file paths.

➤ GPU Support Not Working

ONNX Runtime GPU conflicts

If both onnxruntime and onnxruntime-gpu are installed at the same time, it can cause execution failures. The RemObj Setup Manager uninstalls the CPU package before installing CUDA support to prevent this.

To fix manually:

  1. Uninstall both packages:
pip uninstall onnxruntime onnxruntime-gpu -y
  1. Reinstall only the GPU package:
pip install onnxruntime-gpu

Check the official GPU documentation for ONNX Runtime:

CUDA Execution Provider: https://onnxruntime.ai/docs/execution-providers/CUDA-Execution-Provider.html

ROCm Execution Provider: https://onnxruntime.ai/docs/execution-providers/ROCm-Execution-Provider.html

Ensure that your installed CUDA Toolkit and cuDNN versions match the requirements of the onnxruntime-gpu version you installed.

➤ Uninstalling RemObj

Remove the plugin files

  1. Navigate to the RemObj folder inside your Fuses directory.
  2. Delete all the files in that folder.

Uninstall RemObj Packages

Use the uninstall button (trash icon 🗑️) in remobj_manager.py, or run this command in your terminal:

pip uninstall onnxruntime onnxruntime-gpu opencv-python numpy pillow -y

➤ Additional Support

If issues persist after following this guide:

  • Check the GitHub Issues page
  • View plugin logs in DaVinci Resolve: Fusion page → Workspace → Console
  • Ensure Disable Logging is unchecked in the RemObj node settings

Clone this wiki locally