Prompt Master

This ComfyUI workflow, "Prompt Master," provides a versatile set of tools for crafting and refining prompts for the Flux model. It offers multiple input methods, allowing you to tailor your prompts to achieve the desired results.

Features

This workflow includes the following prompting methods:

• Img2Prompt with JoyCaption2 LLM: Generate detailed and descriptive prompts directly from images using the powerful JoyCaption2 Large Language Model. This is ideal for quickly capturing the essence of an image and using it as a starting point for your Flux generations.

• Manual Prompt Input: For precise control, you can directly input your prompts. This allows for fine-tuning and experimentation with specific keywords and phrases.

• LLM Polished Prompt: Leverage an LLM to refine and enhance your manually created or Img2Prompt-generated prompts. This helps to improve clarity, grammar, and overall effectiveness, leading to better results from the Flux model. (Specify which LLM is used here if applicable)

• Wildcard Prompts: Introduce variability and surprise into your generations using wildcard placeholders. This feature allows you to explore a wider range of possibilities with minimal effort. (Explain how wildcards are implemented - e.g., using a specific node or text file input)

Usage

1. Installation: Ensure you have ComfyUI installed and running. (Mention any specific custom nodes required and how to install them. Provide links if available)

2. Workflow Loading: Download the `prompt_master.json` file (or whatever the file extension is) and load it into ComfyUI.

3. Prompt Input: Choose your preferred prompting method: (only one can be active at a time)

• Img2Prompt: Connect the image you want to use to the Img2Prompt input node.

• Manual Prompt: Enter your prompt text into the designated input box.

• LLM Polished Prompt: Connect your manually created or Img2Prompt prompt to the LLM Polished Prompt input node.

* Wildcard Prompts: Initial support for wildcard prompting.

4. Flux Model Integration: Connect the output of your chosen prompt method to the appropriate input of your Flux model nodes.

5. Generation: Run the workflow to generate images based on your prompts.

## Custom Nodes

• AddLabel /

• Anything Everywhere /

• Bookmark (rgthree) /

• DisplayText /

• ExtraOptionsNode /

• FaceDetailer /

• Fast Groups Bypasser (rgthree) /

• Image Comparer (rgthree) /

• Image Filter /

• ImpactWildcardProcessor /

• InjectLatentNoise+ /

• JWInteger /

• JoinStrings /

• JoyCaption2_simple /

• LF_RegexReplace /

• OverrideVAEDevice /

• Power Lora Loader (rgthree) /

• ProjectFilePathNode /

• Prompts Everywhere /

• SAMLoader /

• SaveText|pysssss /

• Searge_LLM_Node /

• Seed (rgthree) /

• Seed Everywhere /

• String /

• String Replace (mtb) /

• Switch any [Crystools] /

• Text Concatenate /

• Text Input [Dream] /

• Text to Conditioning /

• UltimateSDUpscale /

• UltralyticsDetectorProvider /

• UnetLoaderGGUF /

• easy boolean /

• easy showAnything /

• ## LLM Models

• * unsloth/Meta-Llama-3.1-8B-Instruct /

```bash

$ huggingface-cli download unsloth/Meta-Llama-3.1-8B-Instruct --local-dir ComfyUI/models/LLM/unsloth--Meta-Llama-3.1-8B-Instruct --exclude "*.git*" ""

```

* Orenguteng/Llama-3.1-8B-Lexi-Uncensored-V2 /

```bash

$ huggingface-cli download Orenguteng/Llama-3.1-8B-Lexi-Uncensored-V2 --local-dir ComfyUI/models/LLM/Orenguteng--Llama-3.1-8B-Lexi-Uncensored-V2 --exclude "*.git*" ""

```