ComfyUI-Minitools Introduction
ComfyUI-Minitools is a collection of small ComfyUI nodes for tasks that can otherwise take several steps in a workflow. It includes color conversion, prompt translation and optimization, transparent-image utilities, numeric sliders, a video-prompt workspace, and a layer editor. You can use the tools independently: choose the ones that help with your project rather than adding the whole collection to every workflow.
For example, you might convert a selected color into separate red, green, and blue values, crop empty transparent space from an image, or develop a video prompt from reference images. The author describes the collection as a way to avoid unnecessary node clutter.
How ComfyUI-Minitools Works
A ComfyUI workflow passes information between connected nodes. Each Minitools node handles a particular part of that process:
- Conversion nodes change the format of information, such as turning a color into values another node can use.
- Image nodes prepare an image or its transparent areas for the next step.
- Prompt nodes help translate, expand, or edit the words that describe what you want to create.
- Editing controls let you adjust numbers or work with image layers inside a workflow.
Think of the nodes as small tools on an artist’s desk. You do not need the color converter to use the prompt translator, and you do not need the layer editor to use a numeric slider.
Some features work with external services. promptsTranslateEN uses the Baidu Translate API, while PromptOptimizer uses the Alibaba Cloud Bailian model API. Their API credentials and service availability matter when you use those nodes. The other features described here should not be assumed to require those services.
ComfyUI-Minitools Features
color2rgb: Split a selected color into RGB values
color2rgb is designed to work with a color-picker node from ComfyUI LayerStyle. It converts the selected color into RGB and provides separate R, G, and B outputs.
Use it when a later step needs individual color-channel values instead of a color-picker output. For example, you could select a color visually, then connect its red, green, and blue components to inputs that accept those values separately.
promptsTranslateEN: Translate prompts into English
promptsTranslateEN takes input text and translates it into English using the Baidu Translate API. It can help when you prefer to draft an image prompt in another language but need an English version for the next step in your workflow.
This node requires a Baidu Translate app ID prepared in advance. The original documentation says you can apply for one through Baidu Translate’s official service.
Example of the promptsTranslateEN node
RGB or HEX Convert to DEC: Change color notation
This node converts hexadecimal color codes or RGB color values into decimal values. It is useful when you know the color you want but the next input expects a different numeric representation. Check which form the receiving node expects before connecting the result.
Example of the color conversion node
CropTransparentEdges: Remove empty transparent borders
CropTransparentEdges crops an image according to its transparent area. If an object sits inside a large transparent canvas, the node can remove the unused transparent space around it. This can make the visible subject easier to position or pass to a later image-processing step.
Example of CropTransparentEdges
ImageToMaskWithAlpha: Prepare an image as a mask
ImageToMaskWithAlpha converts an image into a mask with an alpha channel. An alpha channel describes transparency: it helps distinguish visible and transparent areas when preparing artwork for masking or compositing.
Example of ImageToMaskWithAlpha
NumericSlider: Adjust numbers with sliders
NumericSlider provides sliders for both integers (whole numbers) and floating-point numbers (numbers with decimals). Use the integer slider for settings that call for whole-number steps and the floating-point slider when you need finer adjustments. Moving a slider lets you try values without repeatedly typing them.
Example of the NumericSlider controls
PromptOptimizer: Develop prompts from text and images
PromptOptimizer is based on prompt-enhancement ideas from the official Bernini demonstration. It can help with prompt inference from reference images, optimization, and expansion. The node uses Bernini’s prompt presets to complete and refine the inferred content, but the author notes that you can also use it independently as a prompt generator.
The original documentation describes two reference-image approaches:
- One image: Develop a prompt from a single image for image generation, video generation, first-and-last-frame video work, or video editing.
- Multiple images: Infer a prompt from several images or video-related references for video generation, first-and-last-frame video work, or editing.
The node depends on the Alibaba Cloud Bailian model API. Its default uses a multimodal model, which can handle more than text. The author also describes using a text-only large language model separately from a vision or multimodal model, often marked VL, when you want to manage token use. A text-only model is appropriate for text-based refinement; interpreting images requires a model that supports visual input.
Minimax H3 Studio: Work on video prompts in one workflow
Minimax H3 Studio, also described as a Hailuo H3 workspace, is intended to support text-to-video, image-to-video including first and last frames, and reference-based video work within one workflow.
You can upload images and refer to them in the prompt box with @. Clicking an uploaded image can also insert its name into the prompt, making it easier to keep track of the reference you mean.
The node supports interfaces compatible with the OpenAI API protocol; the original documentation names local Ollama and Alibaba Cloud Bailian APIs as examples. It can also be used offline as a prompt editor without an LLM interface. In that mode, it helps you compose prompts but does not call an API to run video generation.
image layer editor: Edit and combine transparent layers
The image layer editor began as a way to work with the Qwen-Image-Layered model, which the original documentation describes as generating images with transparent layers. The author also notes that the editor has uses beyond that initial purpose.
The node supports editing transparent layers from Qwen Image 2.1. Together with extension nodes, it can combine layers with other images, produce a composited image, and output separate layered images with RGBA information. RGBA means red, green, blue, and alpha; the alpha component preserves transparency. This is useful when you want both a finished composition and layers you can continue to work with.
Example of the image layer editor
Troubleshooting ComfyUI-Minitools
The translation node is not producing an English prompt. What should you check?
- Confirm that the text you want to translate is connected to
promptsTranslateEN. - Confirm that you have prepared the Baidu Translate app ID required by the node.
- Check any error shown by the node or API service. The supplied documentation does not specify additional error messages or credential settings.
PromptOptimizer cannot process a reference image. What should you check?
- Make sure the image is connected to the node.
- Check that the selected model supports visual input. The documentation distinguishes text-only models from vision or multimodal models.
- Confirm that the Alibaba Cloud Bailian API credential required by the node is available.
Can you use Minimax H3 Studio without an LLM API?
Yes, for offline prompt editing. Without an LLM interface, you can work on the prompt, but the offline editor mode does not make an API call to generate a video.
Why are transparent areas important for cropping, masks, and layers?
Transparency tells a workflow which parts of an image are visible. CropTransparentEdges uses it to identify empty borders; ImageToMaskWithAlpha uses alpha information when preparing a mask; and the layer editor preserves transparency in RGBA layer output. If your image has no transparent areas, check whether it is the right input for the task you are trying to perform.
Learn More about ComfyUI-Minitools
- ComfyUI LayerStyle provides the color-picker node mentioned alongside
color2rgb. - The official Bernini demonstration offers context for the prompt-enhancement approach behind
PromptOptimizer. - The feature screenshots above show the nodes in use and can help you recognize their controls and outputs in a workflow.
