How to change clothes with AI.

Sebastian Kamph
21 Jan 202414:44

TLDRThe video script offers a step-by-step guide on using stable diffusion to alter an image, specifically focusing on changing a character's clothing. The tutorial covers the use of inpainting models, zooming in for detailed work, and applying various clothing styles and materials. It emphasizes the importance of mask drawing for precise changes and adjusting settings like D noising strength for optimal results. The process is showcased with examples of transforming a woman's outfit from blue jeans to leather pants and a white crop top to different styles and colors, highlighting the creative potential of AI in image editing.


  • 🎨 The tutorial demonstrates how to alter an image using stable diffusion and inpainting techniques.
  • 🖌️ The process begins by using the 'in paint' tab and can be enhanced with an inpainting model for better results.
  • 👕 Changing clothing on a character involves masking the area of interest and applying desired styles and materials.
  • 🌟 Preloaded painting styles, available through Patreon, can simplify the process and offer various fabric options.
  • 🖼️ The 's' key allows for a zoomed-in view for precise drawing within the masked area.
  • 🎨 Adjusting the 'D noising strength' and selecting the appropriate 'masked content' can improve the quality of the generated image.
  • 👖 Examples include changing a pair of blue jeans to leather with varying levels of success depending on settings.
  • 👚 The color and material of a crop top can be altered by redrawing the mask and selecting new styles.
  • 🧢 Adding new clothing items like a hat can be done using the 'in painting sketch' tab and refining the result in the 'in paint' tab.
  • 🔄 Iteration is key, as results may vary, and adjustments to settings like 'D noising strength' may be necessary.
  • 📚 The tutorial encourages exploration of the tools and styles, acknowledging the trial-and-error nature of generative AI.

Q & A

  • What is the main topic of the video?

    -The main topic of the video is teaching viewers how to change clothing and styles of a character in an image using stable diffusion and inpainting techniques.

  • What software or tool is being used in the video?

    -The software being used in the video is stable diffusion, which is an AI model for image generation and manipulation.

  • What is the significance of the 'Inpaint' tab in the software?

    -The 'Inpaint' tab in the software is used for making specific changes to an image, such as altering clothing or adding new elements, by manually painting over the area of interest.

  • How does one zoom in on an image for detailed work in the software?

    -To zoom in on an image for detailed work, the user can press the 's' key, which provides a zoomed-in view for precise painting and editing.

  • What are 'inpainting styles' and how are they used in the video?

    -Inpainting styles are preloaded sets of parameters that help guide the AI in creating specific textures or materials, such as leather, silk, or velvet. They can be found and used through the video creator's Patreon and are used to improve the quality of the generated clothing in the image.

  • What is the purpose of the 'Masked Content' setting in the inpainting process?

    -The 'Masked Content' setting determines what information the AI uses to generate the inpainted content. 'Original' reads from the existing pixels, while 'Latent Noise' introduces new data, which can be helpful for changing colors or materials.

  • How does the 'Denoising Strength' setting affect the inpainting process?

    -The 'Denoising Strength' setting controls the level of change introduced to the image during the inpainting process. Higher values may introduce more significant changes, which can be useful for drastic alterations but may also risk introducing unwanted artifacts.

  • What is the 'Inpaint Sketch' tab used for in the software?

    -The 'Inpaint Sketch' tab is used for adding new elements to an image that were not originally present. It allows the user to draw a rough sketch of the desired element, and the AI will attempt to generate it based on the input.

  • How can one add a new clothing item to an image using the software?

    -To add a new clothing item, one can use the 'Inpaint Sketch' tab to draw the item in a rough sketch, specify the material and color, and then use the AI to generate the final result. Additional fine-tuning can be done using the regular 'Inpaint' tab.

  • What is the importance of the 'Auto Detect Ruler' tool when generating images at a specific resolution?

    -The 'Auto Detect Ruler' tool is used to correct the resolution of the generated image to match the original image. This is important to avoid distortions or loss of detail when zooming in or out on specific areas of the image.

  • What is the role of trial and error in achieving satisfactory results with the software?

    -Trial and error play a crucial role as the AI's output can be random and may not always meet expectations. Adjusting settings like 'Denoising Strength' and 'Masked Content', as well as redrawing masks, can help refine the results until the desired outcome is achieved.



🎨 Introduction to Image Transformation with Stable Diffusion

The video begins with an introduction to the process of transforming an image using Stable Diffusion. The host demonstrates how to change the appearance of an image, specifically altering the clothing of a woman depicted in the image. The video promises to teach viewers how to use the inpainting model for better results and shares tips on overcoming challenges. The host also humorously mentions eating a clock the previous day, emphasizing the time-consuming nature of certain tasks. The starting image is a 1.5 image generated through Stable Diffusion, featuring a woman in a white crop top and blue jeans in a park setting. The focus is on showing various ways to change the clothing of characters in an image, with an emphasis on versatility and adaptability in style.


👖 Transforming Clothing with Inpainting Techniques

The second paragraph delves into the specifics of transforming the woman's clothing using inpainting techniques. The host explains how to use the 'in paint' tab and the inpainting model for more precise alterations. The process involves zooming in on the image, adjusting the brush size, and drawing outside the clothing item to be changed. The host discusses the importance of managing the edges and outlines when making changes. Various clothing styles are mentioned, such as leather, cotton, silk, and velvet, and the host shares a method for preloading in painting cloth styles, which can be accessed through Patreon. The video demonstrates how to change a pair of blue jeans to black leather pants, highlighting the effectiveness of the inpainting process and the potential for further refinement.


👕 Iterative Adjustments and Adding New Elements

The final paragraph focuses on the iterative process of adjusting and refining the image, as well as adding new elements that were not originally present. The host attempts to change the woman's white crop top to a green silk top, experimenting with different settings to achieve the desired result. The paragraph emphasizes the importance of adjusting the denoising strength and masked content to refine the image. Despite some trial and error, the host successfully changes the top's material and color. The video also introduces the 'in painting sketch' tab for adding new clothing items, such as a red leather hat. The host shares insights on improving the quality of the generated images through additional iterations and adjustments. The video concludes with a reminder to visit Patreon for style access and a farewell, encouraging viewers to apply the demonstrated techniques in their own projects.



💡stable diffusion

Stable diffusion is an AI model used for generating images. In the context of the video, it serves as the primary tool for altering the appearance of clothing in a photograph. The video demonstrates how to use stable diffusion to change the style and color of a woman's outfit, showcasing its capabilities in image manipulation.


Inpainting is a technique used in image editing to fill in missing or damaged parts of an image with new content that matches the surrounding area. In the video, the user employs inpainting to modify the clothing of a character in the image, using a specific model for this purpose and detailing the process of selecting and applying different styles of clothing.


Masking in image editing refers to the process of selecting a specific area of an image to apply edits to, while leaving the rest of the image untouched. In the video, masking is crucial for isolating the parts of the image that the user wants to change, such as the woman's pants or top, without affecting the rest of the image.

💡Denoising strength

Denoising strength is a parameter in AI image generation models like stable diffusion that determines how much the AI tries to retain the structure of the original image when generating new content. A higher denoising strength means the AI will attempt to keep more of the original image's details, while a lower value allows for more creative freedom and potential changes.

💡Latent noise

Latent noise refers to the random variation introduced by the AI model during the image generation process. It is used to create new data and can lead to more diverse and sometimes unexpected results. In the video, the user selects 'latent noise' as the masked content option to generate new colors and patterns for the clothing.


In the context of the video, styles refer to pre-defined sets of parameters or characteristics that can be applied to the AI model to achieve specific visual effects. The user mentions having preloaded in painting cloth styles, which can be used to quickly apply different textures and materials to the clothing in the image.

💡Image to image

Image to image is a feature in AI image generation models that allows users to start with one image and transform it into another by applying various edits and adjustments. In the video, the user navigates to the 'image to image' tab to access the inpainting tools for changing the clothing in the image.


Resolution refers to the detail and clarity of an image, typically measured in pixels. In the video, the user discusses changing the resolution of the generated image to balance detail and blending with the original image. Higher resolution provides more detail but may not blend as well, while lower resolution may blend more easily but with less detail.


Zooming in an image editing context allows users to focus on a specific area of the image for more precise editing. In the video, the user presses 's' to zoom in on the image, which enables them to draw more accurately when creating a mask or applying styles to the clothing.


Patreon is a platform where creators can offer exclusive content to their supporters for a monthly fee. In the video, the user mentions that they have preloaded in painting cloth styles available through their Patreon, suggesting that it's a source for additional resources and styles for users who wish to enhance their image editing experience.

💡Redo and iteration

Redo and iteration refer to the process of repeating or revising an action or procedure to improve upon the initial result. In the video, the user emphasizes the importance of redoing and iterating the image changes to achieve the desired outcome, as the AI model may not produce perfect results on the first attempt.


The video tutorial demonstrates how to use stable diffusion to alter an image, specifically changing the clothing of a character.

The process is user-friendly and accessible, allowing individuals to make changes to images without prior expertise.

The tutorial introduces the concept of inpainting, which can be used to modify specific parts of an image.

The use of an inpainting model can enhance the experience, though it's not mandatory and a regular model can also be used.

The 's' key provides a zoomed-in view for precise drawing within the image.

Adjusting the brush size is crucial for detailed work, and can be done by holding control or command and scrolling the mouse wheel.

The tutorial showcases how to change a character's clothing style, such as from blue jeans to leather pants.

Preloaded painting cloth styles can be utilized for efficient transformations, which can be accessed through Patreon.

The 'only masked' setting allows for detailed changes in a specific area, while 'whole picture' can help with blending.

Changing the 'Denoising strength' can introduce variations in the generated image, which can be beneficial or detrimental depending on the desired outcome.

The 'masked content' setting can be toggled between 'original' and 'latent noise' to influence the image generation process.

The tutorial provides examples of how to change the color and material of clothing, such as from a white crop top to a green silk one.

The generative AI process involves trial and error, as results can vary with each iteration.

The 'in painting sketch' tab can be used to add new elements to an image, such as drawing a hat.

The final results showcase the versatility of stable diffusion in altering and adding clothing to images.

The video concludes by encouraging viewers to explore Patreon for styles and emphasizing the power of the workflow.