Prompts

Main Prompt

Prompt is a description of the image you aim to generate and is inputted within the Prompt Field. Here you describe what you'd like to generate.

The quality of images generated by Stable Diffusion heavily relies on the effectiveness of the prompt. A well-crafted prompt can guide the AI model to generate images that closely match the desired result, while an unclear or vague prompt may result in unexpected or unsatisfactory outcomes. A suggested approach to organizing your prompts is by taking into account the subject, background, medium, style, artist style, resolution, rendering method, additional details, color, and lighting. ā€ A good Stable Diffusion prompt should have the following qualities: ā€Relevant: Employ pertinent keywords, adjectives, and phrases that are linked to the subject and scene, separated by commas. ā€Concise: Use precise language and avoid unnecessary words that could confuse the model or dilute the intended meaning. ā€Unambiguous: Avoid ambiguous words or phrases that can be interpreted in multiple ways.


Style Prompt

Style Prompt allows you to divide your prompt into two fields, enabling you to focus on modifying the main prompt while keeping the style section unchanged. You can create your own Style Prompt by selecting Custom within the Style Prompt dropdown and entering your own style prompts.

Note: The final prompt submitted to the generation model is a combination of the two: "Prompt, Style Prompt." You can also choose to leave the style empty and only use the main prompt field. Examples of style keywords include: fantasy art, artstation, cinematic, anime, analog, 3D, digital art, oil painting, ink, drawing, surreal, sci-fi, portrait, disco diffusion, inkpunk, comic, funko pop, paper cut, waifu, dark side, double exposure, openjourney, abstract, impressionistic, realistic, minimalistic, surreal, pop art, cubism, expressionistic, street art, graffiti, retro, vintage, futuristic, gothic, romantic, steampunk, noir, ethereal, whimsical, cartoonish, mosaic, geometric, collage, psychedelic. You can also reference art history periods, camera angles, camera lenses, lighting, and create any combination of adjectives and technical terms. Examples of camera angle names include: bird's eye view, high angle, low angle, eye level, Dutch angle, worm's eye view, over the shoulder, point of view (POV), close-up, medium shot, wide shot, long shot, extreme close-up, overhead shot, tracking shot, pan shot, tilt shot, crane shot, steadicam shot, zoom shot, split screen shot, aerial shot, canted angle (also known as a tilted or oblique angle), rack focus shot.


Negative Prompts (Avoid)

The Negative (Avoid) Prompt represents elements you want to exclude from your image. Use this field to specify items, image characteristics, styles, etc., that you do not want the model to incorporate during the image generation process.

You can add to and edit the Avoid field to provide your negative prompts.

Example terms for various use cases: Default: malformation, lowres, bad anatomy, bad hands, error, missing fingers, cropped, worst quality, low quality, normal quality, bad quality, jpeg artifacts, watermark. For illustrated characters: bad anatomy, bad hands, three hands, three legs, bad arms, missing legs, missing arms, poorly drawn face, bad face, fused face, cloned face, worst face, three crus, extra crus, fused crus, worst feet, three feet, fused feet, fused thigh, three thigh, fused thigh, extra thigh, worst thigh, missing fingers, extra fingers, ugly fingers, long fingers, horn, realistic photo, extra eyes, huge eyes, 2girl, amputation, disconnected limbs. For realistic images: Enter a combination of the previous terms and: cartoon, cg, 3d, unreal, animated, illustration. For children content: Add the following terms: nsfw, nude, censored Employ negative prompts such as "ugly" or "bad quality" to encourage the model to put in more effort. You can also consider terms like "3D", "realistic", if you're aiming for specific drawing styles.


Attention Modifiers

Using ( ) in your prompts increases the model's attention to enclosed words, while [ ] decreases it. Employ these modifiers in the following manner: (word) - increases attention by a factor of 1.1 ā€ ((word)) - increases attention by a factor of 1.21 (= 1.1 * 1.1) ā€ [word] - decreases attention by a factor of of 1.1 ā€ [[word]] - decreases attention by a factor of of 1.21 (word:1.5) - increases attention by a factor of of 1.5 (word:0.25) - decreases attention by a factor of 4 (= 1 / 0.25) When using ( ), a weight can be specified as follows: (word:1.4). If no weight is specified, it is assumed to be 1.1. Please note that specifying weight only works with ( ), not with [ ]. You can also apply this to multiple words, for example: (astronaut in space:1.2), bright lights, (reflective helmet), cinematic, amazing lights, [soft focus] If you intend to include the literal ( ) [ ] characters in the prompt without affecting attention, and instead want them to be treated as a literal part of the prompt, use a backslash "\" to escape them like this: \(word\)


Need help? Ask in our Discord!

For any questions regarding the use of Sogni, tips, suggestions, bugs, sharing art, discussing AI and life, or simply hanging out... Join our Discord community! Join Sogni Discord āœØ

Last updated