AS_LLM_nodes

This ComfyUI extension provides custom nodes for working with Google Gemini and OpenAI ChatGPT.

AS_GeminiCaptioning

User Guide

The AS_GeminiCaptioning node lets you generate a descriptive text prompt from a single image using the Google Gemini API. Supply your image and adjust any optional text fields to tailor the output.

Inputs

• IMAGE (Required)
  The image you want to describe (e.g., JPEG, PNG). Connect your image input.

• PROMPT TYPE (Required)
  Choose between the preset styles "SD1.5 – SDXL" or "FLUX" to select the base style of the prompt.
  If you do not provide a custom reference, the selected style determines which default text is used.

• APY KEY PATH (Required)
  The file path to your API key for the Google Gemini API.

• GEMINI MODEL (Required)
  The model that will process your request. Possible options include:

  • Gemini 2.0 Flash
  • Gemini 2.0 Flash-Lite
  • Gemini 1.5 Flash
  • Gemini 1.5 Pro
    The default is Gemini 2.0 Flash.

• PROMPT LENGTH (Optional)
  An approximate word count for the final prompt. If empty, there is no length restriction.

• PROMPT REFERENCE (Optional)
  A sample text prompt format that serves as a reference. If empty, a default reference is used.

• PROMPT STRUCTURE (Optional)
  A guideline for organizing details in the prompt (e.g., building type, materials, location).

• IGNORE (Optional)
  Specific words or concepts to exclude from the prompt.

• EMPHASIS (Optional)
  Words or concepts to emphasize.

• SAVE TO PATH (Optional)
  A directory path for saving the generated text file.

• TXT NAME (Optional)
  Name for the .txt file. If empty and a path is provided, defaults to result.txt.

Outputs

• RESULT PROMPT
  The text response from Gemini.

• REQUEST TEXT
  The exact text payload sent to the API.

• LOG
  A log of execution steps and errors.

---

AS_MultimodalGemini

User Guide

The AS_MultimodalGemini node sends text plus up to three images to the Google Gemini API.

Inputs

• TEXT_INPUT (Required)
  A text string to be sent along with the images.

• API_KEY_PATH (Required)
  The file path to your Gemini API key.

• GEMINI MODEL (Required)

  • Gemini 2.0 Flash
  • Gemini 2.0 Flash-Lite
  • Gemini 1.5 Flash
  • Gemini 1.5 Pro
    Default is Gemini 2.0 Flash.

• IMAGE_1, IMAGE_2, IMAGE_3 (Optional)
  Up to three images to attach.

Outputs

• RESULT
  The text returned by Gemini.

• LOG
  A log of steps and errors.

---

AS_ComfyGPT

User Guide

The AS_ComfyGPT node integrates with OpenAI’s ChatGPT. Provide the path to your API key, choose a model, and enter a prompt. The node returns ChatGPT’s reply.

Inputs

• api_key_file (Required)
  Path to a file containing your OpenAI API key.

• model (Required)
  Name of the OpenAI model (e.g., "gpt-4", "gpt-3.5-turbo").

• prompt (Required)
  The user prompt text to send to the GPT model.

Outputs

• STRING
  The response from ChatGPT.

---

Required Libraries

• Pillow
• requests
• google-generativeai
• openai