What Is Clip Skip Stable Diffusion

Are you trying to generate quicker images without loss of quality? Learn how to do it with ease through the steps below.
What Is Clip Skip Stable Diffusion

Stable Diffusion usually takes time to generate an image, and if you want multiple images within a short period, they can be lacking. For this, Stable Diffusion offers a Clip skip, where the final layers of the text processing model are skipped, and an intermediate layer is used to complete the process.

In this guide, we will help you add Clip Skip to Stable Diffusion, along with a demonstration of it in action to get you all covered.

Clip Skip on Stable Diffusion

  1. Load Stable Diffusion on your device and open its Settings. Navigate to the options on the left and click on User interface.
  2. Doing so, you’re displayed with various settings from which click on the drop-down under Quick settings list and select sd_model_checkpoint, sd_vae, and CLIP_stop_at_last_layers.
Clip skip - Settings, User interface, Quick settings list, Apply, Reload UI
  1. Before navigating to the Generation tab, ensure to click on Apply and then Reload UI for the changes to appear.
Apply and Reload UI
  1. If you do it right, you will notice the options that are selected in the drop (refer to Step 2) at the top of the home screen. Now, set the Clip Skip last stop by adjusting the slider according to your requirements.
Clip skip - Quick settings updates
  1. After the quick settings are updated, scroll down to the bottom and select the Script as X/Y/Z plot. It is essential to choose this as you can generate images with different parameters.
Clip skip - Select Script as XYZ plot
  1. Now you are presented with more options, and to demonstrate Clip Skip, we have set the X plane as Clip Skip by selecting it from the drop-down.
Clip skip - Select X Type as Clip Skip
  1. While the default options can be left as is, set the X values, and the results are generated to the values.
Clip skip - Select Clip Skip Values
  1. With everything set, navigate to the top, input the positive and negative prompts, and click on the Generate button. Here are the positive and negative prompts we used by referring to an image from CivitAI.
Positive prompt: portrait of (snnmrn:1) as wonder woman, feminine, epic > (photo, studio lighting, hard light, sony a7, 50 mm, hyperrealistic, big depth of field, mate skin, pores, wrinkles, concept art, colors, hyperdetailed, hyperrealistic:1.1)

Negative prompt: hood, hoodie, text, b&w, illustration, painting, cartoon, 3d, bad art, poorly drawn, close up, blurry, disfigured, deformed, extra limbs

Here is the result after using the above prompts and options set in the steps provided:

Clip Skip results

Wrapping Up

Congratulations on learning how to set up clip skip on Stable Diffusion and use it to generate images!

While we have struggled with using the Clip Skip at the start, we suggest you avoid adding multiple Clip Skip values (refer to Step 7) and consider around 4-5. Also, get the prompts right and adjust the primary options in the Generation tab only when needed.

What Is ControlNet Stable Diffusion

How To Use img2img Stable Diffusion

What Is VAE Stable Diffusion

What Is a Negative Prompt in Stable Diffusion

How To Use Stable Diffusion for NSFW


We will be happy to hear your thoughts

Leave a reply

Enable registration in settings - general