This tool is a custom node for ComfyUI designed to facilitate the automatic testing and comparison of various samplers and schedulers. It generates both latent and RGB images, providing detailed annotations and performance metrics for each combination tested.
- Automates the testing of all combinations of samplers and schedulers, eliminating manual input.
- Outputs images with overlays that include critical performance metrics and generation details.
- Allows customization of overlay text appearance, enhancing the clarity of the results displayed.
Context
This custom node enhances ComfyUI by allowing users to systematically evaluate different combinations of samplers and schedulers. Its primary purpose is to streamline the process of testing, making it easier to identify the most effective settings for image generation without the need for manual configurations.
Key Features & Benefits
The tool's key features include the ability to iterate through all available samplers and schedulers, producing comprehensive outputs that include both generated images and detailed performance metrics. The inclusion of metrics such as Laplacian Variance, Gradient Mean, and FFT-based sharpness provides users with quantifiable data to assess image quality, thereby enabling informed decisions on optimal settings.
Advanced Functionalities
This node offers advanced functionalities such as customizable overlays for generated images, allowing users to adjust font size, text color, and background color. Additionally, it supports the specification of particular samplers and schedulers for testing, giving users the flexibility to focus on specific configurations that interest them.
Practical Benefits
By automating the testing process, this tool significantly enhances workflow efficiency in ComfyUI. Users gain greater control over their image generation processes, resulting in improved quality and a more streamlined approach to exploring various configurations. This not only saves time but also allows for a more thorough investigation of potential settings.
Credits/Acknowledgments
This project was developed by the Icelandic Center for Artificial Intelligence (ICAI) and is licensed under the European Union Public Licence v.1.2 (EUPL-1.2). The source code benefitted from AI-assisted development, contributing to its functionality and robustness.