ComfyUI-ImagePlus

ComfyUI ImagePlus

ComfyUI ImagePlus is a custom unified nodes pack primarily for Image to Image generation with a toggleable Text to Image option for easy switch back. It allows for CivitAI image inspection, local image upload/drop, metadata extraction, prompt syncing, and ImagePlus workflow helper nodes. ComfyUI ImagePlus is the unified successor to ComfyUI CivitAI ImagePlus, https://github.com/CovertBannana/ComfyUI_CivitAI_ImagePlus

Since CivitAI released their own node pack with different objectives, the direct naming relationship no longer made sense. ImagePlus has evolved into a broader image inspection, metadata extraction, gallery browsing, prompt conditioning, and workflow helper toolset.

The new nodes feature Model, Clip and VAE pass through for a smoother work flow. The KSampler Adv includes a VAE encode for direct image input and built-in empty latent image options that combined with a toggle will instantly switch between I2I and T2I. The effort was to eliminate the common vae encode, empty latent image and clip text encode (prompt) that are used in almost all workflows. The Nodes are not restricted to being used exclusively with the packs other nodes and should incorporate the necessary outputs to be used with standard nodes. A trigger word input connector has been added to connect with a STRING node for a constant input or from other node packs like Lora Manager that offer that feature. There is now an option to store Favourite images in the Gallery and a recent Import list. Favourites can be added and removed. Recent list can be cleared. Both lists are stored in the favourites.json and recent.json files.

Nodes

• ImagePlus Inspector
• ImagePlus Prompt
• ImagePlus KSampler Adv

Features

• CivitAI gallery browsing
• CivitAI gallery Favourite and Recent import list
• Fetch image/post URL support
• Local image upload and drag/drop
• Image preview with dimensions
• Metadata tabs
• Positive and negative prompt extraction
• Prompt conditioning outputs
• Model/VAE pass-through
• String/Trigger word input (optional)
• ImagePlus KSampler Adv node for switching between Image In and Empty Latent sampling paths

Installation

Clone this repository into your ComfyUI custom_nodes folder:

Open a command prompt in: ComfyUI/custom_nodes
git clone https://github.com/CovertBannana/ComfyUI-ImagePlus.git

Restart ComfyUI after installation.

Requirements

The package currently uses:

aiohttp>=3.9.0

ComfyUI normally provides torch, numpy, and Pillow.

Optional CivitAI API Key

To use a CivitAI API key, add your CivitAI API to this file inside the node folder api_key.txt

ImagePlus Inspector

ImagePlus Inspector provides the image selection, upload, preview, metadata, and prompt inspection workflow.

Outputs

• image: IMAGE output for downstream image or latent workflows
• positive: final positive prompt used for conditioning, including optional Trigger Words input
• negative: extracted negative prompt string
• info: filtered info/metadata string

Uses

• Browse/select CivitAI images
• Fetch an image/post URL
• Upload or drag/drop a local image
• Preview selected image and dimensions
• Extract prompts and metadata
• Send image output into img2img workflows
• Favourite images for later reuse
• Automatic Recent Imports history
• JSON-based persistence for Favourites and Recents

ImagePlus Prompt

ImagePlus Prompt receives model, clip, and vae inputs, then outputs prompt conditioning and pass-through model/VAE connections.

Inputs

• model: MODEL
• clip: CLIP
• vae: VAE
• trigger_words: STRING (optional)

Outputs

• model: MODEL pass-through
• vae: VAE pass-through
• positive_conditioning: CONDITIONING for KSampler positive input
• negative_conditioning: CONDITIONING for KSampler negative input
• positive_prompt: extracted positive prompt string
• negative_prompt: extracted negative prompt string (positive_prompt and positive_conditioning now reflects the actual prompt used for conditioning, including Trigger Words/String inputs)

Recommended Wiring

Checkpoint Loader:model
  → ImagePlus Prompt:model

Checkpoint Loader:clip
  → ImagePlus Prompt:clip

Checkpoint Loader:vae
  → ImagePlus Prompt:vae

ImagePlus Prompt:model
  → ImagePlus KSampler Adv:model

ImagePlus Prompt:vae
  → ImagePlus KSampler Adv:vae

ImagePlus Prompt:positive_conditioning
  → ImagePlus KSampler Adv:positive

ImagePlus Prompt:negative_conditioning
  → ImagePlus KSampler Adv:negative

ImagePlus KSampler Adv

ImagePlus KSampler Adv is an ImagePlus workflow helper node that combines the common img2img/txt2img latent setup with KSampler execution.

It can operate in two modes using the image_in toggle:

• image_in = true
  Encodes the connected IMAGE input through the connected VAE, then runs sampling.

• image_in = false
  Creates an empty latent internally using the selected width, height, and batch size, then runs sampling.

This replaces the need for a separate VAE Encode, Empty Latent Image, and latent switch node.

Inputs

• model: model input from ImagePlus Prompt or Checkpoint Loader
• vae: VAE input from ImagePlus Prompt or Checkpoint Loader
• positive: positive conditioning
• negative: negative conditioning
• image_in: toggle between image input mode and empty latent mode
• width: empty latent width
• height: empty latent height
• batch_size: empty latent batch count
• seed: sampler seed
• steps: sampler steps
• cfg: CFG scale
• sampler_name: sampler selection
• scheduler: scheduler selection
• denoise: denoise strength
• image: optional image input used when image_in is enabled

Outputs

• samples: latent output for VAE Decode
• vae: VAE pass-through output for VAE Decode

Denoise Notes

When using image_in = true, this behaves like an img2img workflow. Typical denoise values are usually below 1.0, such as:

0.35 to 0.75

When using image_in = false, this behaves like a txt2img workflow from an empty latent. Typical denoise is:

1.0

The node does not automatically change denoise. The user controls denoise manually.

License

MIT License. See LICENSE for details.