A comprehensive set of custom nodes for ComfyUI, focusing on utilities for image processing, JSON manipulation, model operations and working with object via URLs
Loads images from URLs.
Inputs:
image
: List of URLs or base64 image data, separated by new lineskeep_alpha_channel
: Preserve alpha channeloutput_mode
: List or batch output. UseList
if you have different resolutions.
Loads JSON data from URLs.
Inputs:
url
: JSON URLprint_to_console
: Print JSON to console
Loads JSON data from text.
Inputs:
data
: JSON textprint_to_console
: Print JSON to console
Includes GetObjectFromJson
, GetTextFromJson
, GetFloatFromJson
, GetIntFromJson
, GetBoolFromJson
.
Use key format key.[index].subkey.[sub_index]
to access nested objects.
Converts strings to numbers.
Inputs:
string
: Input stringrounding
: Rounding method
Randomizes the order of lines in a multiline string.
Inputs:
text
: Input textamount
: Number of lines to randomizeseed
: Random seed
Switch between multiple cases based on a condition.
Inputs:
switch_cases
: Switch cases, separated by new linescondition
: Condition to switch ondefault_value
: Default value when no condition matchesdelimiter
: Delimiter between case and value, default is:
The switch_cases
format is case<delimiter>value
, where case
is the condition to match and value
is the value to return when the condition matches. You can have new lines in the value to return multiple lines.
Prepares images and masks for inpainting operations. It's to mimic the behavior of the inpainting in A1111.
Inputs:
image
: Input image tensormask
: Input mask tensormask_blur
: Blur amount for mask (0-64)inpaint_masked
: Whether to inpaint only the masked regions, otherwise it will inpaint the whole image.mask_padding
: Padding around mask (0-256)width
: Manually set inpaint area width. Leave 0 default to the masked area plus padding. (0-2048)height
: Manually set inpaint area height. (0-2048)
Outputs:
inpaint_image
: Processed image for inpaintinginpaint_mask
: Processed maskoverlay_image
: Preview overlaycrop_region
: Crop coordinates (input of OverlayInpaintedImage)
Overlays inpainted images with original images.
Inputs:
inpainted
: Inpainted imageoverlay_image
: Original imagecrop_region
: Crop region coordinates
Outputs:
IMAGE
: Final composited image
Remove objects from images using LaMa model.
Configures generic LLM API parameters.
Inputs:
model
: Model name (GPT-3.5, GPT-4, etc)max_token
: Maximum tokenstemperature
: Temperature parameter
Configures OpenAI API access.
Inputs:
openai_api_key
: OpenAI API keyendpoint
: API endpoint URL
Configures Anthropic Claude API access.
Inputs:
claude_api_key
: Claude API keyendpoint
: API endpointversion
: API version
Configures AWS Bedrock Claude API access.
Inputs:
aws_access_key_id
: AWS access keyaws_secret_access_key
: AWS secret keyregion
: AWS regionversion
: API version
Configures AWS Bedrock Mistral API access.
Inputs:
aws_access_key_id
: AWS access keyaws_secret_access_key
: AWS secret keyregion
: AWS region
Creates a message for LLM interaction.
Inputs:
role
: Message role (system/user/assistant)text
: Message contentimage
: Optional image inputmessages
: Previous message history
Handles chat interactions with LLMs.
Inputs:
messages
: Message historyapi
: LLM API configurationconfig
: Model configurationseed
: Random seed
Handles completion requests to LLMs.
Inputs:
prompt
: Input promptapi
: LLM API configurationconfig
: Model configurationseed
: Random seed