How to create better faces with stable diffusion

Stable Diffusion allows you to create virtually any text-based image. For example, your project may require you to improve an AI face, but you are having trouble improving the original. If so, you could use some of our tips to improve the Stable Diffusion face creation process. Continue reading to learn more.

How do you create better faces with stable diffusion?

To create better faces with Stable Diffusion, you need to set up a Google Colab called Deforum Stable Diffusion v0.5. Here’s how you can do it:

1. Open your browser and go to this website.
   
2. Press the “Connect” button and click “NVIDIA GPU”. Since this is Google’s server GPU, don’t worry: you don’t need an NVIDIA graphics card.
   
3. Grant system permissions for the website to use your computer.
   
4. Click on the section below the GPU card (“Model and Output Paths”).
   
5. Connect your Google Colab with your Google Drive and wait for the check mark to appear next to “Model and Output Paths.”
   
6. Select “Set up environment” and let the system connect your account. Do the same with “Python Definitions”.
   
7. Go to “Select and load model”. This section requires that you have the model you want to upload to your Google Drive. To do so, download the .ckpt file from this website pressing “sd-v1.4.ckpt” in the “Download Weights” section. Upload this file to your Google Drive AI folder containing the models you want to improve.
   
8. Press the “Play” button next to the “Select and Load Model” section. Wait for the green check mark to appear and your Google Colab will be up and running.
   

Now that your Google Colab is set up, you can enhance your AI faces with Stable Diffusion. This process is also known as improving old renders and boils down to changing the resolution of your images:

1. Crop the face you want to enhance and save it as a separate image.
   
2. Upload the face by dragging it to your temporary Google Colab drive (the left panel).
   
3. Right click on the image and click the “Copy Path” button.
   
4. Paste the image into the “init_image” field in the right menu.
   
5. Navigate to the “Directions” window and enter the directions you want for your face. Be sure to write
   in the directions inside the quotes for best results.
   
6. Go to the “Run” section and look for the “Image Settings” menu.
   
7. Change the pixel resolution to improve face clarity. For example, 800×800 works well in most cases.
   
8. Press the red “Play” button below the “Run” window to render the image.
   
9. Your image will now be rendered at a different resolution. Experiment with the procedure until you achieve the desired effect.
10. Download the image from your disk and overlay it on top of the original.
    

How do you use variations with stable diffusion to create better faces?

In addition to enhancing faces directly by changing the resolution, you can also introduce variations to produce sharper images. The first step is to set your “Animation” preferences to “none.” This is what you should do:

1. Connect your Google Drive to Google Deforum Stable Diffusion 0.5 Colab with the steps above.
   
2. Upload the face image to the left panel of Colab.
   
3. Go to the “Animation” menu and change the “animation_mode” setting to “none”.
   
4. Click the “Play” button next to the “Animation” section to save your changes.
   

Once you’ve changed the “Animation” settings, you can ignore most of the other settings since you don’t want to create an animation. However, you will need to make some adjustments in the “Messages” section. Follow these steps:

1. Use the original message as a starting point.
   
2. Make any necessary variations by changing the line, depending on whether you want to add or remove features from your AI face. Work in the “Messages” section instead of the “Animation Messages” window.
   
3. Place quotes at the beginning and end of your prompts.
   
4. Place a comma at the end of the message (after the quotes).
   
5. Click the “Play” button in the upper left corner of the “Messages” window to render your image and save your progress.
   

The next menu you want to modify is the “Run” menu. More specifically, you need to go to the “Image Settings” window and make the following changes:

1. Navigate to “Sample Settings” and set the “Seed” setting to “-1”.
   
2. Experiment with the “Sampler” settings until you get the desired effect.
   
3. Set as many “Steps” as you like. The “Scale” value is also up to you.
   
4. If you used “ddim” rendering in the “Sampler” window, set “ddim_eta” to “1”.
   
5. Scroll down until you find “Batch Settings” and go to the “n_batch” line. Type the number of variations you want to have in the text box.
   
6. Go to “Startup Settings” and press the box next to the “use_init” option. There should now be a check mark next to the settings.
   
7. Type “0.5” in the “strength” field if you want to reuse 50% of the original image. You can enter higher or lower values, depending on how much you want to reuse the source material.
   
8. Click the “File” icon in the left panel. Go to the folder containing the face image you want to optimize and drag it to the section on the left. Wait for the system to load your image, right click on it and press the “copy path button”.
   
9. Paste the path into the “init_image” field and click the red “Play” button in the top left corner of the “Image Settings” menu.
   
10. Let the program render your image and add variations. You should now see the new face at the bottom of the screen with your variations.

How to improve stable diffusion imaging

When enhancing faces with Stable Diffusion, most of the work comes down to typing the right prompts. The system relies on this function to generate an image using “txt2img” technology. It uses the text message and other parameters to produce an image based on your input.

As mentioned above, you must enter the “seed” value at a time to create the desired results. You can also change the number of “steps” by incorporating variations.

However, these features may not do much to improve the resulting faces. Instead, you’ll need to step up your “cueing” game to ensure high-quality image generation. Here are some tips to help you achieve this.

• Be specific: Your text prompts should be very detailed, especially the beginning, since the beginning of your sentence is more important than the end. If you leave out key information, you may not get the look you want.
• Stick to English: Stable Diffusion developers used more than two billion English text-image pairs. The other 100+ languages ​​represented fewer pairs, meaning that the proportion of inpidual languages ​​is much smaller than that of English. As a result, typing prompts in languages ​​other than English may not give you the correct result.
• Include genre and type of art: Stable Diffusion can generate all types of images, including acrylic and oil paintings, fashion photography, cartoons, and prints. Therefore, specifying the shape of images you want makes the algorithm more precise.
• Be on trend: If you are looking for artistic images, you can add “trending in ArtStation” or other names to help Stable Diffusion get as close as possible to the desired image.
• Play with different words: When writing your prompts, you can never go overboard in your descriptions. So, play around with different technical properties, like “sharp focus” and “dramatic lighting” or adjectives like “stunning,” “beautiful,” and “stunning.”
• Add an artist’s name: Stable Diffusion allows you to produce images in the style of famous artists. To achieve this effect, add your artists’ names to your message, such as Andry Warhol, Caspar David Friedrich, or Auguste Renoir.
• Use negative messages: Some versions of your interface react to negative messages. These allow you to indicate what your system should avoid when producing images.

Let your imagination and creativity fly in a stable diffusion

Mastering stable diffusion can take a while, especially when you want to enhance your AI faces and other images. One or two bad lines can easily degrade your project, so you need to be very careful when typing your prompts and entering build parameters. With a few adjustments, you should be able to beautify faces and any other part of your images with ease.

How often do you use Stable Diffusion? Have you ever added variations to your images? If so, what effects did they produce? Share your experiences in the comments section below.