An open source plug-in that makes it possible to use the image generation AI 'Stable Diffusion' with Krita
Krita ', several plug-ins have appeared that will allow you to use the topical AI ' Stable Diffusion ' that can automatically generate images from text and images.
With the open source paint software '
Stable Diffusion is an AI that can generate images with ' text2img ' that generates images along the input text and ' img2img ' that can generate detailed images even from rough images such as rough line drawings. is. Stable Diffusion not only has a demo version that can be used on the web, but it is also possible to run what is distributed under the Creative ML OpenRAIL-M license in the local environment. There are various ways to actually use Stable Diffusion, but the article below shows how to use it for free and without waiting time even on a low-spec PC.
Summary of how to use image generation AI ``Stable Diffusion'' even on low-spec PCs for free and without waiting time - GIGAZINE
One of the plugins that can use Stable Diffusion on Krita is 'sddebz'. sddebz is a Web UI-based plug-in, and the images that can be output are RGBA 8-bit color images only. You also need to add layers to output the image.
GitHub - sddebz/stable-diffusion-krita-plugin
sddebz is a Web UI-based plug-in, and how it actually works is as shown in the video played when you click the image below.
rutkowski , cumulonimbus clouds, sunset, nostalgic, very very very beautiful art, cinematic lighting, bright pastel color, blue sky' Sci-fi concept art of a futuristic city, cumulonimbus, sunset, nostalgic, very very beautiful art, featuring cinematic lighting, bright pastel colors, blue sky).
'scifi concept art of a futuristic city near to a big alke by greg
The output result is within the red frame below.
After selecting the range ......
Gaston - Art by Bussiere and Alphonse Mucha : Enter 2).
Next is 'portrait of beautiful female, blonde hair, feminine, mature, beaatiful, upper body: 2 intricate, elegant, highly detailed, digital painting, artstation, concept art, smooth, sharp focus, illustration: 2 art by gaston bussiere and alphonse mucha:2' (portrait of a beautiful woman, blonde, female, mature, beautiful, upper body: 2, intricate, elegant, highly detailed, digital painting, artstation, concept art, smooth, sharp focus, illustration: 2,
Then, the touch of the illustration changed only in the selected range.
There is also a Stable Diffusion-based Krita addon called qweryty . The feature of qweryty is that not only can you use both 'text2img' and 'img2img' of Stable Diffusion, but it also supports Real-ESRGAN- based upscaling.
GitHub - qweryty/image-ai-utils-krita
The video below shows how images can be generated using qweryty.
There is also a Stable Diffusion-based AI image generation plugin that can be used with Krita called 'koi (Krita Opensource Img2Img: StableDiffusion)'. koi is an open source project that uses the backend server of Google Colaboratory, and it is explained that they decided to release early in order to lower the barrier to entry into art and make it available to everyone.
GitHub - nousr/koi: A plug-in for Krita that enables the use of AI models for img2img generation.
In addition, a user who is developing a Krita plug-in that can operate on a PC equipped with NVIDIA's GeForce RTX 3060 Ti has also appeared on the overseas bulletin board Reddit. You can see how it actually works in the embedded video below. You can check how it evolves.
In the video below, a part of the photo is erased to output UFOs and aliens.
Furthermore, in the video below, an image is output from text. The first is 'dog wearing a suit, highly detailed' and the second is 'cat wearing a suit, highly detailed'. It's a cat.