[SD advanced skills] How to better generate prompts

[SD advanced skills] How to better generate prompts

March 10th, 2023
View OriginalTranslated by Google

Apply rules and techniques

website :

1. https://publicprompts.art/ (best for OpenArt online models https://openart.ai/) 2. https://docs.qq.com/doc/DWFdSTHJtQWRzYk9k (transferred from Element Codex, including various style terms) 3. https://www.ptsearch.info/home/4. http://prompttool.com/NovelAI (the experience is not bad)

use: 1. General Tag starting gesture (commonly required content!!) Prompt: masterpiece, best quality, more quality words, picture description Negative prompt: nsfw, lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, according to the picture Output plus unwanted images. nsfw (not safe for work=沉图) 2. The closer the tag is to the front, the greater the weight; for example, if the scenery tag is in the front, the character will be small, and on the contrary, the character will become larger or half-length. 3. The size of the generated image will affect the effect of the prompt. The larger the image, the more prompts are needed, otherwise the prompts will pollute each other. 4. Use () English brackets in Stable Diffusion to increase the weight of the Tag in the brackets by x1.1, and [] to decrease the weight of the Tag by x0.91 . 5. Prompt supports the use of emoji, and the expressiveness is better, and the expressive effect can be achieved by adding emoji. If 😍 describes the expression, 🖐 can repair hands. 6. "+" "AND" "|" usage: Both "+" and "AND" are used to connect short tags, but spaces should be added at both ends of AND; "+" is approximately equal to "and"; "|" draws symbols for loops (fused symbols) (Prompt A: w1)|(Prompt B: w2) The above expressions are applicable to WebUI, and w1 and w2 are weights. AI will draw the Prompts A and B in a loop. You can join Prompt unlimitedly in the future.

format optimization

1. Simple line break three-segment expression The first paragraph: image quality tag, painting style tag The second paragraph: the main body of the picture, the main body is emphasized, and the main body details are summarized. (The main body can be people, things, things, scenery) The third paragraph of the core content of the picture: the details of the scene of the picture, or the details of the characters, embedding tag, supplementary content of the picture details: the second paragraph generally provides the number of people, the main characteristics of the characters, and the main actions (generally placed in front of the person), the main features of the object, the main scene or the frame of the scene, etc. 2. Grammatical quality words in the adjusted version of the element dictionary → front painting style → front lens effect → front lighting effect → (person or thing with description AND secondary description of person or thing AND lens effect and lighting) * factor → Global illumination effect → global lens effect → style filter (embedding) 3. Tips Different models (ckpt) have different sensitivity to prompts, and a complete set of prompts will have different expression effects in different models. Each model has its own characteristics, and the combination of prompts needs to be adjusted slowly according to the characteristics of the model.

Please click here for SD local deployment and usage tutorials .


no dataCoffee time! Feel free to comment