List of Negative Prompts for Stable Diffusion

Generating AI images has never been easier. With Stable Diffusion, users have freedom to express their creativity. This can be achieved by creating intricate prompts and utilizing negative prompts. Negative prompts can be useful to get rid of unwanted objects or be more specific in the generating process. In this blog post, we will list negative prompts that can be useful in generating AI images.

Negative prompts to generate Photo Realistic Image

It needs to be noted that generating Photo-realistic images is mostly supported by using the correct and good quality Stable Diffusion Models. Models like Realistic Vision, Juggernaut, and HelloWorld are good starting points. Experienced users may also add LoRa, ControlNet and adjusting settings like CFG scale, denoising strength etc. Here are some examples of prompts to enhance your generation:

Anime, 2D, Sketch, Drawing, Bad photography, Bad photo, Deviant art, Cartoon, CGI, Render, 3D, Artwork, Illustration, 3D render

Negative prompts to generate correct body anatomy

AI Image can sometimes look a bit wonky with googly eyes, or misshaped body part. This is where negative prompts can help to enhance the generation. Here are some examples of prompts to get you correct body anatomy.

Missing limbs, Missing arms, Extra fingers, Extra hands, Extra limbs, Mutated hands, Mutated, Mutation, Multiple heads, Malformed limbs, Bad anatomy, Bad hands, Amputee, Missing fingers, Missing hands, Disfigured, Poorly drawn hands, Poorly drawn face, Long neck, Fused fingers, Fused hands, Dismembered, Duplicate, Improper scale, Ugly body, Cloned face, Cloned body, Gross proportions, Body horror, Too many fingers

Negative Prompts for Scenery, Nature, Landscape Image

Here are some ideas for negative prompts when you are trying to generate scenery or nature images:

Overexposed, Simple background, Plain background, Grainy, Portrait, Grayscale, Monochrome, Underexposed, Low contrast, Low quality, Dark, Distorted, White spots, Deformed structures, Macro, Multiple angles

FAQs:

Do I need to write long negative prompts?

No, however relevant negative prompt is quite important.

What does it mean when there’s a prompt structure like this (prompt:1), ((prompt)), [prompt]?

Using () in the prompt increases the model’s attention to enclosed words, and [] decreases it. You can combine multiple modifiers:

a (word) – increase attention to word by a factor of 1.1

a ((word)) – increase attention to word by a factor of 1.21 (= 1.1 * 1.1)
a [word] – decrease attention to word by a factor of 1.1
a (word:1.5) – increase attention to word by a factor of 1.5
a (word:0.25) – decrease attention to word by a factor of 4 (= 1 / 0.25)
a \(word\) – use literal () characters in prompt

With (), a weight can be specified like this: (text:1.4). If the weight is not specified, it is assumed to be 1.1. Specifying weight only works with () not with []

Conclusion

Negative Prompts in Stable Diffusion are useful for getting the desired images. Aside from that, it is critical to provide other parameters such as proper positive image and the checkpoint/Stable Diffusion model. DiffusionHub comes with some preloaded models and check out our blog to help you create the desired image.