explaingit

acly/krita-ai-diffusion

10,074PythonAudience · designerComplexity · 3/5Setup · hard

TLDR

A Krita plugin that adds AI image generation, inpainting, live painting, and upscaling, directly inside the painting app, powered by Stable Diffusion and Flux models running locally or in the cloud.

Mindmap

mindmap
  root((krita-ai-diffusion))
    What it does
      Canvas inpainting
      Live painting hints
      Image upscaling
    Tech Stack
      Python
      ComfyUI backend
      Stable Diffusion
    Setup
      6GB GPU required
      NVIDIA recommended
      Cloud option available
    Audience
      Digital artists
      Illustrators
Click or tap to explore — scroll the page freely

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

Things people build with this

USE CASE 1

Select a region in a Krita canvas and fill or replace it with AI-generated content without leaving the app.

USE CASE 2

Enable live painting mode to see AI suggestions update in real time as you sketch rough lines on the canvas.

USE CASE 3

Upscale a low-resolution painting to a very high resolution using the AI upscaling feature inside Krita.

USE CASE 4

Assign different text prompts to separate canvas layers to generate a scene where each region has its own style.

Tech stack

PythonStable DiffusionFluxComfyUIKrita

Getting it running

Difficulty · hard Time to first run · 1h+

Requires a GPU with at least 6GB VRAM, NVIDIA is the most directly supported path, AMD, Apple Silicon, and Intel XPU need extra configuration.

In plain English

This is a plugin for Krita, a free open-source painting and image editing application, that adds AI image generation directly into your editing workflow. Instead of jumping between separate tools to generate and edit images, you can select a region of your canvas inside Krita and have the AI fill, replace, or extend it without leaving the app. The main things it can do: fill selected areas with generated content (inpainting), watch your canvas in real time and continuously suggest what the sketch or rough lines might look like as a finished image (live painting), upscale images to very high resolutions, and accept reference images or sketches as guides to steer what gets generated. It supports a range of AI model types including Stable Diffusion 1.5 and XL, Flux 2, and Illustrious. You can apply text-based edits to an existing image and assign separate text descriptions to different regions of a canvas using layers. Running it locally requires a graphics card with at least 6 GB of video memory. NVIDIA GPUs are the most directly supported path, AMD, Apple Silicon, and Intel XPU setups are possible but need additional configuration. The plugin uses ComfyUI as its backend, which it can install automatically or connect to an existing installation. A cloud-based option is also available if you want to avoid the hardware requirement to start. Installation follows the standard Krita plugin process: download the ZIP release from GitHub, import it through Krita's built-in script import menu, and then configure it through the plugin's own docker panel inside the app. The documentation and FAQ are at docs.interstice.cloud, and support goes through GitHub Discussions or a Discord server. The plugin is open source, actively maintained, and welcomes contributions. It is free to run locally and does not require any account or subscription for local use.

Copy-paste prompts

Prompt 1
Using the Krita AI Diffusion plugin, select a region of my painting and fill it with a realistic stone wall texture using Stable Diffusion XL.
Prompt 2
How do I install and configure the Krita AI Diffusion plugin with ComfyUI on an NVIDIA GPU with 8GB VRAM, step by step?
Prompt 3
In Krita with the AI Diffusion plugin, how do I assign a separate text prompt to a sky layer versus a foreground layer to control each region independently?
Prompt 4
How do I enable live painting mode in the Krita AI Diffusion plugin so that AI suggestions update automatically as I draw rough sketches?
Open on GitHub → Explain another repo

← acly on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.