Make Images of Yourself in Playground AI/Stable Diffusion (without training or downloads)

9 Jan 202306:38

TLDRThe video tutorial demonstrates how to use the Playground AI/Stable Diffusion tool to modify personal photos without the need for training or downloads. The process begins with uploading a photo and adjusting its strength to 100% for a direct likeness. The user can then replace the background by painting over the desired areas, ensuring to cover contrasting colors to avoid artifacts. After selecting a new background, the user can generate multiple images with the same prompt and removal settings. To further personalize the image, the user can switch to an image-to-image filter and adjust settings for a cinematic look. Additional modifications include changing clothing with the painting mask feature and generating a set of images to choose from. The video also covers facial restoration and upscaling the final image by four times, though it clarifies that the facial restoration does not upscale. The tutorial concludes with tips on refining the image to match personal preferences and keeping the final selection.


  • 🖼️ To convert a photo, drag and drop it into the provided interface.
  • 🔧 Increase the image strength to 100 for the highest conversion quality.
  • 🎨 Use the largest brush to select and modify the background, ensuring to cover contrasting colors.
  • 🖌️ Erase unwanted parts of the background, like red, to avoid artifacts.
  • 📌 If you want to restore a part of the image, use the paint tool to cover it.
  • 🌟 Select a desired background from the Playground AI's main page to match the style you want.
  • 🧩 Modify the prompt and removal settings to create a different look while keeping the desired background.
  • 👕 Change your clothing in the image by using the image-to-image feature and adding a painting mask.
  • 🛡️ Personalize your look, for example, by giving yourself a 'silver Knight armor'.
  • 🔄 Generate multiple images and select the one that best fits your preference.
  • 🔍 Use facial restoration and upscaling features to enhance the final image quality.

Q & A

  • What is the process of converting a photo using Stable Diffusion 1.5?

    -The process involves dragging and dropping an image into the provided box, setting the image strength to 100, and turning on a private session. The system then generates the same image based on the prompt provided.

  • How can you change the background of an image using the tool?

    -You can change the background by selecting the 'paint' tool and using a brush to cover the unwanted background. You can also erase parts as needed, ensuring to cover colors that differ greatly from the skin tone to avoid artifacts.

  • What is the purpose of using a red background in the script?

    -The red background is used to contrast the subject and create a clean separation for easier manipulation. However, it is suggested to get rid of the red background to avoid creating artifacts in the final image.

  • How can you restore parts of the image that you want to keep but accidentally erased?

    -If you accidentally erase a part of the image you want to keep, such as an ear, you can use the 'paint' tool to redraw that part of the image.

  • What does the term 'prompt' refer to in the context of image generation?

    -In the context of image generation, a 'prompt' is a set of instructions or a description that guides the AI in creating the desired output. It helps the system understand what kind of image to generate.

  • How can you generate multiple images based on a single prompt?

    -You can generate multiple images by setting the system to generate a specific number of images, such as four, using the same prompt and removal settings.

  • What is the significance of selecting a background that resonates with you?

    -Selecting a background that resonates with you helps in creating an image that matches your personal preferences and the desired outcome. It serves as a foundation for further modifications and enhancements.

  • How can you modify the image to match a specific style or theme?

    -You can modify the image to match a specific style or theme by using filters and adjusting settings such as the 'image to image' filter and the '85' parameter, which is mentioned as a 'happy number' for achieving a good balance in the image.

  • What is the purpose of the 'painting mask' tool?

    -The 'painting mask' tool is used to make specific changes to the image, such as altering the subject's clothing or adding elements like a 'silver Knight armor' in the example provided.

  • How can you refine the generated images to better suit your taste?

    -You can refine the generated images by continuously modifying the prompt and settings until you achieve the desired look. If you don't like any of the generated images, you can keep generating new ones until you find one that meets your satisfaction.

  • What are the options for enhancing the final image quality?

    -To enhance the final image quality, you can use features like 'facial restoration' and 'upscale by four'. However, note that 'upscale by four' does not apply to the facial restored image, and it's important to download and keep the image you want to enhance.



🖼️ Background and Clothing Transformation

The speaker demonstrates how to use an image editing tool, presumably stable diffusion 1.5, to alter a personal photo. They start by uploading an image and adjusting its strength to 100% for a clear conversion. The focus then shifts to changing the background using a paint tool, carefully covering contrasting colors and avoiding artifacts. After modifying the background, the speaker selects a new background from a pre-existing image, making minor adjustments to achieve the desired look. The process continues with changing the subject's clothing to match the new background, using image-to-image transformation and a painting mask for customization. The speaker emphasizes the iterative nature of the process, encouraging viewers to keep generating images until they find one they like.


🔍 Image Refinement and Enhancement

Once satisfied with the background and clothing changes, the speaker discusses the process of refining the image further. They mention the option to declutter by removing unwanted elements and suggest generating more images if the initial results are not satisfactory. The speaker also talks about using facial restoration to improve the quality of the subject's face in the image. However, they clarify that facial restoration needs to be downloaded separately and does not come with the upscale feature. They demonstrate upscaling the image by four times, noting that this process does not apply to the facial restored version, and advise users to be aware of this common confusion.



💡Stable Diffusion

Stable Diffusion is an AI model that is used for generating images from textual descriptions. In the context of the video, it is used to convert a person's photo with modifications such as changing the background or clothing without the need for training the model or downloading additional software.

💡Image Strength

Image strength refers to the intensity or the degree to which the input image's features are retained in the generated output. In the video, the user sets the image strength to 100 to ensure that the generated image closely resembles the original.

💡Private Session

A private session in the context of the video script implies a mode where the AI operates to ensure user privacy, possibly not storing the images or data used in the session. It's important for users who are concerned about their data's confidentiality.

💡Paint and Erase Tools

These are digital tools used to manually edit the input image before processing it with the AI. In the video, the user selects and paints over the background to prepare it for the AI to generate a new background, and the erase tool is mentioned for correcting any over-painting.


In digital imaging, an artifact is an unwanted additional pattern or effect that appears in an image due to the image processing method. The video mentions avoiding artifacts by covering colors that contrast greatly with the skin tone during the painting process.


A prompt is a text input that guides the AI in generating an image. It can describe the desired outcome or the style of the image. In the video, the user uses a prompt to guide the AI to generate a background similar to an example image.


Image-to-Image is a process where an AI takes an existing image and transforms it into a new image based on a given prompt or filter. In the video, the user selects an image and uses the image-to-image feature to modify the subject's appearance.


Cinematic refers to a style or quality that is reminiscent of or suitable for a movie. In the context of the video, the user wants to give a more cinematic look to the generated image, possibly to enhance its visual appeal.

💡Facial Restoration

Facial restoration is a process where the AI attempts to correct or enhance the facial features in an image. In the video, the user uses facial restoration to improve the quality of the generated image's face.


Upscaling is the process of increasing the resolution of an image. In the video, the user upscales the image by four times to enhance its detail and clarity, although it's noted that this does not apply to the facially restored version.

💡Ghost Body

Ghost body refers to an undesirable effect in image processing where parts of the body appear transparent or distorted. The video mentions avoiding ghost body effects during the image generation process.


Drag and drop a photo into the provided box to convert it using Stable Diffusion 1.5.

Increase the image strength to 100 for the highest likeness to the original image.

Turn on private session for privacy during the image conversion process.

Use the largest brush in paint to modify the background, ensuring to cover contrasting colors.

Erase unwanted red background colors to avoid creating artifacts in the generated image.

Recover parts of the image, like an ear, if needed using the editing tools.

Select a desired background from the Playground AI main page to use as a reference.

Copy the reference image prompt and use it with removal settings for a similar background.

Remove unwanted elements from the background to customize the final look.

Generate multiple images to find the one that best fits your preference.

Use the 'Select this one' feature to switch to a more cinematic filter.

Adjust the filter settings to find a happy medium between the original and desired look.

Experiment with different clothing styles, such as a silver knight armor, using the image-to-image feature.

Generate a set of images with the new prompt and modifications to see various outcomes.

Remove elements that don't align with your taste or the desired image outcome.

If satisfied with an image, use the space restoration or upscale by four feature for better quality.

Note that facial restoration requires downloading the image and keeping it for further use.

The upscale by four feature only upscales the current image, not the facial restored one.

Drag the upscaled image to your desired location for storage.