ComfyUI Extension: ComfyUI-Jax-Nodes

Authored by Kebolder

Created

Updated

0 stars

Some custom nodes for weird niche needs when working with Krita

Custom Nodes (0)

    README

    ComfyUI-Jax-Nodes

    Custom ComfyUI nodes for some niche Krita-focused workflows and easier prompt / pipeline handling.

    Nodes Overview

    • Conditional Select (JAX_ConditionalSelect)

      • Simple if/else node: chooses between true_value and false_value based on a boolean condition and outputs the selected value.
    • Sampler Pipe In (JAX_KritaPipeIn)

      • Packs a set of sampler-related inputs (model, positive, negative, vae, image, clip, latent) into a single pipe dictionary that can be passed around a workflow.
    • Sampler Pipe Out (JAX_SamplerPipeOut)

      • Unpacks the pipe dictionary back into individual outputs (model, positive, negative, vae, image, clip, latent), useful for branching or recombining pipelines.
    • Krita Strength (JAX_KritaStrength)

      • Given sigmas, a strength, and a denoise value, computes an integer step index that approximates Krita-style strength control for diffusion, clamping values to safe ranges.
    • Krita Resize Canvas (JAX_KritaResizeCanvas)

      • Output node that, when enabled, emits a small JSON payload in the UI (application/x-krita-command) describing a resize_canvas command with the chosen width and height.
    • Image Size Multiplier (JAX_ImageSizeMultiplier)

      • Multiplies an input width and height by a multiplier and outputs the new dimensions plus a small Markdown string summarizing the final size.
    • Easy Prompt (W/ Append) (JAX_EasyPrompt)

      • CLIP text encoder that lets you write prompts with inline LoRA tags like <lora:name:weight> and append extra positive/negative text. It loads and applies LoRAs to the model/clip, then returns the updated model and positive/negative conditioning.
    • Easy Prompt (JAX_EasyPromptSimple)

      • Simpler version of the above: supports <lora:...> tags but without extra append fields, just positive and negative inputs.