What Are Negative Prompts In Stable Diffusion? [GUIDE]
Stable Diffusion is a deep learning, text-to-image model released in 2022. It is primarily used to generate detailed images conditioned on text descriptions, though it can also be applied to other tasks such as inpainting, outpainting, and generating image-to-image translations guided by a text prompt.
In order to use Stable Diffusion and generate arts or images as per your desire, you must know some parameters and one among them is a Negative Prompt.
In this guide, you will learn what is negative prompt in Stable Diffusion is and how you can use it to generate high-quality images as per your defined prompt parameters. Also, do not forget to check our post on best stable diffusion anime prompts and try some artwork with it.
Let’s get started!
What Are Negative Prompts In Stable Diffusion?
A negative prompt is a way to use Stable Diffusion in a way that allows the user to specify what he doesn’t want to see, without any extra input. It is a parameter that tells the Stable Diffusion model what not to include in the generated image.
Negative prompting influences the generation process by acting as a high-dimension anchor which the process strays away from. It can also be used to improve the image output by specifying abstract concepts such as “blurry” and “pixelated”.
Negative prompting affects the generation process by acting as a high-dimensional anchor that the process moves away from. This lets the output image be more precisely controlled. By using negative prompts, users can make unique images with more detail and accuracy.
For example: You might have generated a portrait but sometimes Stable Diffusion provides duplicates, even though you provide it with parameters like ‘Don’t add duplicates’. That’s because it understands negative prompts better than human language.
In that case what you can do instead of regenerating the same prompt, you can provide a negative prompt like ‘duplicate’ or ‘poorly Rendered face.’
Here’s what you will get without a negative prompt:
Here’s what you will get by adding a negative prompt such as “Duplicate”:
How To Use Negative Prompts In Stable Diffusion or Dream Studio?
Though you might have got an idea on using negative prompts in Stable Diffusion, let’s understand this with more examples and by the end, you will be provided an entire list of negative prompts for stable diffusion with you can use it to remove certain objects or make your images more enhanced.
As per the documentation of stability.ai, trying out negative prompts is as simple as putting “| negative prompt>: -1.0” to the prompt in DreamStudio or Stable Diffusion. An example would be to add “| deformed, ugly: The error “-1.0, too many fingers:-1.0” is sometimes a workaround for the problem of producing an excessive number of fingers.
Now, let’s understand Negative Prompts with another example.
Example #1: Lets say we generate a hybrid between a German Shephard and a Tiger with this given prompt – a hybrid between a German Shephard:0.7 and a tiger, photography, award winning, documentary, wildlife, 8k
We provided stable diffusion as a negative prompt. In response to the following prompt, Stable Diffusion prioritizes German Shephard. You should thus anticipate a visual that demonstrates the priority of a German Shepherd over a Tiger.
In this scenario, Stable Diffusion prioritizes the German Shephard and then applies the difference (0.3) to the Tiger without any user input.
Example #2: Here’s another example which you can see down below.
The first image has bushes and trees surrounding the german architecture generated through a basic prompt.
But the other image shows what will be generated if you’ll add a negative prompt in order to lower the dominance of Trees near the area.
Note: The negative prompts seed and CFG scale are not available as separate options in DreamStudio. In contrast, the negative prompt syntax described above can be used in Dream Studio.
However, the following negative prompt parameters can be used in Stability Diffusion’s specific negative prompt panel. By the way it’s highly recommended to use negative prompts in Stable Diffusion v2.0.
Universal Negative Prompt List For Stable Diffusion
Here’s the complete list of negative prompts that you can use in Stable Diffusion to remove certain parts of the image that you don’t want to keep:
- bad anatomy
- bad proportions
- cloned face
- extra arms
- extra fingers
- extra legs
- extra limbs
- fused fingers
- gross proportions
- jpeg artifacts
- long neck
- low quality
- malformed limbs
- missing arms
- missing legs
- mutated hands
- out of frame
- poorly drawn face
- poorly drawn hands
- too many fingers
- worst quality
We hope that our guide was able to help you grasp the negative prompts parameter in Stable Diffusion, and that you will put what you’ve learned to use in order to produce even more impressive works of art.
Negative prompts act as a high-dimension anchor which the generation process strays away from.
Some examples of negative prompts in Stable Diffusion include ugly, boring, bad anatomy, blurry, pixelated, trees, green, obscure, unnatural colors, poor lighting, dull, and unclear.
Suggestion: The Universal Negative Prompt List for Stable Diffusion should be provided in alphabetic order, as this will facilitate new additions (include ‘NEW’ after the new negative prompt):
• bad anatomy
• bad proportions
• cloned face
• extra arms
• extra fingers
• extra legs
• extra limbs
• fused fingers
• gross proportions
• jpeg artifacts
• long neck
• low quality
• malformed limbs
• missing arms
• missing legs
• mutated hands
• out of frame
• poorly drawn face
• poorly drawn hands
• too many fingers
• worst quality
Also: Your current list contains two ‘out of frame’ prompts (which would have been caught with an alphabetic ordering)
Request: Provide a list of negative prompts for use with appropriate image genres, such as human (faces, bodies); scenery; imaginative; etc.
Nice work, guys!
Thanks Bill! We will look into that…