stable diffusion img2img walkthrough : paint an bussin watercolor in minutes (no cap) Krita + Koi

The video script outlines a step-by-step guide on transforming simple drawings into sophisticated watercolor paintings using Krita and the Koi plugin. The process begins with creating a new canvas, installing the Koi plugin, and covering the entire image with a base color. The user then draws a basic lighthouse, adds qualifiers like 'beautiful' and 'in the style of JMW Turner', and lets the AI generate four variants with specific parameters. The results showcase the potential for AI to create aesthetically pleasing images with some iterations and adjustments to settings like sketch strength and steps.

Takeaways

• 🎨 **Image to Image Transformation**: The process involves taking a simple sketch and using AI to transform it into a detailed, artistic image.
• 🖥️ **Krita Software**: Krita is a free, open-source software used for creating the initial sketch and canvas.
• 📏 **Canvas Size**: A 512x512 canvas size is recommended as it is preferred by the neural network.
• 🔌 **Koi Plugin**: The Koi plugin is essential and needs to be installed for the AI transformation, although it can be a bit tricky to set up.
• 🌊 **Background Painting**: It's important to paint the entire background to avoid undesired black areas that could confuse the AI.
• 🏰 **Lighthouse Example**: The video uses a lighthouse as an example to demonstrate the image to image process.
• 🎨 **Art Style**: To guide the AI, specifying an art style, such as that of a famous watercolor artist like J.M.W. Turner, can help achieve a desired outcome.
• 🔄 **Variants and Steps**: Requesting multiple variants and increasing the number of steps allows the neural network to refine the image.
• 🌟 **Sketch Strength**: Adjusting the sketch strength parameter gives the AI more or less freedom in the transformation process.
• 📏 **Base Seed**: Using the same base seed ensures the ability to recreate the same result.
• 🔍 **Upscaling**: The resulting images can be upscaled to reduce blurriness and improve detail.
• 🕒 **Time Efficiency**: The process is relatively quick, offering a significant output with minimal effort.

Q & A

What is the purpose of image to image transformation using Krita and the Koi plugin?

The purpose is to take a simple or poorly drawn image and transform it into a more coherent and visually appealing piece of art using the power of neural networks and AI.

Why is it important to cover the entire canvas with color before using the Koi plugin?

It's important because any uncolored pixels are interpreted as black by the neural network, which can lead to unwanted results as it may try to incorporate the black into the final painting.

What is the recommended size for a new canvas in Krita for using with the Koi plugin?

The recommended size is 512 by 512 pixels, as this is the size that the neural network prefers.

How long did it take to install the Koi plugin according to the transcript?

It took about 20 minutes to install the Koi plugin.

What is the role of the 'base seed' in the image transformation process?

The base seed is used to ensure reproducibility. If you use the same base seed and parameters, you will always get the same result.

What does 'sketch strength' control in the Koi plugin?

Sketch strength tells the AI how much attention to pay to the original drawing and how much freedom to give it in the transformation process.

How many variations did the user request to see in the example?

The user requested four different variants to see how they turn out.

What is the significance of the 'steps' parameter in the Koi plugin?

The 'steps' parameter determines how many times the neural network tries to refine and improve the image, making it better and better with each iteration.

Why might some of the generated images look blurry?

Some images might look blurry due to the limitations of the AI's interpretation and the original sketch's level of detail. Using an upscaler can help to reduce the blurriness.

What is the benefit of viewing the generated images from a distance?

Viewing the images from a distance can make them appear more impressive and better composed, as the AI tends to focus more on the overall composition rather than fine details.

What is the final step before clicking 'dream' in the Koi plugin?

The final step is to make a new layer from all the visible elements, ensuring that the AI does not interpret any part of the painting as black and causing unwanted results.

Outlines

00:00

🎨 Transforming Art Skills with Krita and KOI

This paragraph introduces the concept of transforming poor art skills into more refined images using Krita, a free digital painting software, and the KOI plugin. The speaker explains the process of creating a new canvas with specific dimensions to suit the neural network's preference and outlines the steps to install the KOI plugin, which can be a bit tricky but is guided by a tutorial video. The goal is to enhance the user's art by using the AI capabilities of the plugin, starting with a basic lighthouse drawing as an example. The importance of covering the entire canvas is emphasized to avoid misinterpretation by the AI. The paragraph concludes with the speaker's intention to create an oil painting-style image of a lighthouse on a sunny day, inspired by the watercolor artist J.M.W. Turner.

05:02

🖌️ Experimenting with AI Enhancement and Variations

In this paragraph, the speaker discusses the process of experimenting with the AI enhancement process in Krita using the KOI plugin. They explain the adjustments made to the sketch strength and the number of iterations to allow the AI more freedom in creating the image, resulting in a more expressionistic and detailed outcome. The speaker shares their satisfaction with the improved results and suggests using an upscaler to enhance the image quality further. The paragraph highlights the value of the AI's creative potential and encourages viewers to explore their artistic ideas while waiting for the AI to generate images. The speaker also invites feedback on their video presentation and appearance, showing a willingness to improve based on viewer suggestions.

Keywords

stable diffusion img2img

Stable diffusion img2img refers to a process where an initial, often low-quality image is transformed into a more refined and detailed one through the use of machine learning algorithms. In the context of the video, this process is used to take simple drawings and turn them into aesthetically pleasing watercolor paintings, reminiscent of long hours of manual artwork.

Krita

Krita is a free and open-source digital painting program that provides a variety of tools for artists to create digital art. In the video, Krita is used as the platform to draw the initial sketch and to interface with the Koi plugin for the stable diffusion process.

Koi plugin

The Koi plugin is an add-on for Krita that integrates the stable diffusion image-to-image functionality. It allows users to input a basic sketch and then apply the neural network to enhance and transform the image into a more detailed artwork, such as a watercolor painting.

neural network

A neural network is a series of algorithms modeled loosely after the human brain. It is designed to recognize underlying relationships in a set of data through a process that mimics the way the human brain operates. In the video, the neural network is the AI component that takes the sketch from Krita and the Koi plugin to generate a detailed image.

watercolor painting

Watercolor painting is a form of painting that uses water-soluble pigments. It is known for its transparency and fluidity, often creating a light and airy effect. In the video, the goal is to transform a simple sketch into a watercolor painting style, using the stable diffusion img2img process.

JMW Turner

JMW Turner was an English Romantic landscape painter, watercolorist, and printmaker. He is known for his expressive color and dynamic depictions of the natural world. In the video, the creator references Turner's style to guide the neural network in creating a watercolor painting with a similar aesthetic.

base seed

In the context of the neural network used in the stable diffusion process, the base seed is a starting point or a set of initial parameters that the AI uses to generate the image. It ensures that the same input will produce consistent results.

sketch strength

Sketch strength is a parameter that determines how much influence the original sketch has on the final image generated by the AI. A lower sketch strength gives the AI more freedom to deviate from the sketch, potentially leading to more creative or 'wild' results.

upscale

Upscaling an image involves increasing its resolution while attempting to maintain or improve its visual quality. This process can help reduce blurriness and make low-resolution images more suitable for larger formats.

iteration

In the context of the video, iteration refers to the process of repeating the stable diffusion process with adjusted parameters to improve the final result. Each iteration involves tweaking the settings and seeing how the AI responds to generate better images.

aesthetic value

Aesthetic value refers to the appreciation of beauty or good taste in art. In the video, the creator discusses the aesthetic value of the AI-generated images, noting that some of the results have a certain charm or appeal despite not being perfect.

Highlights

Stable diffusion img2img technique allows for transforming simple sketches into sophisticated art pieces.

The method has gained significant popularity online due to its ability to produce high-quality images quickly.

Krita, a free software, is used for creating the initial canvas and sketching the desired image.

A canvas size of 512 by 512 is recommended for optimal neural network performance.

The Koi plugin is essential for the img2img process and can be installed with some effort.

Starting with a basic sketch, such as a lighthouse, provides a foundation for the transformation process.

Covering the entire canvas with a base color, like blue, is crucial to avoid misinterpretation by the AI.

Descriptive qualifiers, such as 'watercolor painting' and 'beautiful', guide the AI in creating the desired style.

Selecting a famous watercolor artist like J.M.W. Turner can influence the AI to mimic their style.

Requesting multiple variants allows for a range of outcomes to choose from.

Adjusting the 'steps' parameter lets the neural network refine the image multiple times.

The 'base seed' parameter allows for the replication of identical results.

Modifying the 'sketch strength' gives the AI more or less freedom in interpreting the sketch.

Combining all visible elements into one layer before processing prevents unintended AI output.

Experimenting with different settings, such as increasing steps and decreasing sketch strength, can lead to improved results.

Viewing the images from a distance can enhance their perceived quality and aesthetic appeal.

The process from sketch to finished artwork can be achieved in minutes with the right tools and techniques.