Base
- Prompt words are divided into forward prompt words (
positive prompt) and reverse prompt words (negative prompt), which are used to tell the AI what is needed and what is not. - The default weight of affixes is 1, which decreases from left to right. The weight will affect the screen generation results.
- AI will selectively execute based on probability. If there is a conflict between prompt words, AI will randomly select which prompt word to execute based on the probability determined by the weight.
-
The size of the generated image will affect the effect of Prompt. The larger the image, the more Prompts are needed, otherwise the Prompts will contaminate each other.
example
Use webui
Negative prompt
easynegative, canvas frame, cartoon, 3d, ((disfigured)), ((bad art)), ((deformed)),((extra limbs)),((close up)),((b&w)), wierd colors, blurry, (((duplicate))), ((morbid)), ((mutilated)), [out of frame], extra fingers, mutated hands, ((poorly drawn hands)), ((poorly drawn face)), (((mutation))), (((deformed))), ((ugly)), blurry, ((bad anatomy)), (((bad proportions))), ((extra limbs)), cloned face, (((disfigured))), out of frame, ugly, extra limbs, (bad anatomy), gross proportions, (malformed limbs), ((missing arms)), ((missing legs)), (((extra arms))), (((extra legs))), mutated hands, (fused fingers), (too many fingers), (((long neck))), Photoshop, video game, ugly, tiling, poorly drawn hands, poorly drawn feet, poorly drawn face, out of frame, mutation, mutated, extra limbs, extra legs, extra arms, disfigured, deformed, cross-eye, body out of frame, blurry, bad art, bad anatomy, 3d renderweight syntax
weight logic
If you want to identify a certain subject, you should move its generation steps forward, increase the number of generation steps, sort the affixes forward, and increase the weight.
Picture quality → Main elements → Details
// 超高分辨率， 8k， 漫画风， 相机取景框，一个男子， 奔跑（动词原型）， 黑色头发， 橙色太阳， 蓝色天空
absurdres, 8k, comic, viewfinder, 1 male, run, black hair, orange sun, blue sky,
If you want to clarify the style, the style affix should be better than the content affix
Picture Quality → Style → Element → Details
Unconfirmed: The style weight may need to be better than the picture quality, so that the special style texture is not contaminated by the picture quality.
// 超高分辨率， 8k， 漫画风， 相机取景框， 一个男子， 奔跑（现在分词）， 黑色头发， 橙色太阳， 蓝色天空
absurdres, 8k, comic, viewfinder, 1 male, running, black hair, orange sun, blue sky,
In these two examples, run is changed to running
separator
, Comma : separates affixes and has a certain weight sorting function. The weight before the comma is high and the weight after the comma is low. Therefore, it is recommended to sort:
- Overview (image quality + painting style + lens effect + lighting effect + theme + composition)
- Subject (Character & Object + Posture + Clothing + Props)
- Details (scene + environment + accessories + features)
Combinator
: Colon : custom weight value
format: left parenthesis + affix + colon + number + right parenthesis
// 单人女孩词缀，权重为0.75
(1girl:0.75) () Parentheses only : Increase weight by 0.1 In most cases, parentheses will suffice.
// 权重乘以1.1
(1girl) {} Curly braces : increase weight by 0.05
// 权重乘以1.05
{1girl}[] Square brackets : Reduce the weight by 0.1, some say it is reduced by 0.05. The
weight value cannot be customized in the square brackets, and the custom weight can only use (x:0.5)the form.
// 权重除以1.1
[1girl] (()), {
{}}, [[]] reuse brackets : superimpose weights
// 权重乘以1.1*1.1，即权重为1.21
((1girl))General process: First write down the scene to be described and generate it once. Based on the generated results, try and modify the unsatisfactory or missing descriptions. Use the (xxx: 1.x) grammatical form to increase the weight of the concepts to be emphasized, where xxx is your For words to be emphasized, 1.x represents the proportion to be improved, for example, 1.5 is to increase the weight by 150%.
The weight value range is 0.4-1.6. If the weight is too small, it will be easily ignored, and if it is too large, it will make errors in fitting the image. Example: (beautiful:1.3).
Stacking weight: Increase the weight by stacking parentheses. Each additional layer is equivalent to increasing the weight by 1.1 times, for example: ((((beautiful eyes)))) .
// 跳跃的权重最高
(absurdres), 8k, comic, ((viewfinder)), 1male, running, (jumping:1.5), black hair, orange sun, blue sky,Joiner
ANDSum : aggregate multiple affixes into one prompt word sequence position, and their initial weights are consistent
// 三个词缀权重一致
bird and dog and pig
// 可使用冒号标记其权重
bird:1.5 and dog:1.2 and pig:0.8
In actual use, it was found that when and is used to connect affixes, objects may merge, such as cabbage and meat, which may produce a weird picture of cabbage with meat wrapped in it.
Distributed and alternate rendering
[]:Using symbols such as and in square brackets | can achieve more complex distribution and alternate rendering requirements.
// 渲染蓝色到40%进度后，开始渲染红色
[blue:red:0.4]
// 前12步渲染蓝色后，开始渲染红色
[blue:red:12]Note:
When step > 1, it means how many steps in the first step the combination will be rendered as A and then as B.
When step < 1, it represents the percentage of iteration steps (total number of steps involved).
// 在执行40%的步骤后，开始渲染蓝色
[blue:0.4]
// 在执行15步之后，开始渲染蓝色
[blue:15]// 在执行了40%的步骤后，停止渲染蓝色
[blue::0.4]
// 在执行了15步后停止渲染蓝色
[blue::15]// 融合蓝色与红色，紫色？
[blue|red]
// 某物与某物的交替演算，用于生成融合生物，句中词缀含义为融合牛与马。
[cow|horse] // 一个蓝红混色染发的短发女孩
1 girl, [blue|red]hair, short hair
best quality, very detailed, absurdres, 8k, comic, [viewfinder], one lady with long [blue|red]hair is sitting on ground, birds around her, blue sky+[orange sun], 
// [viewfinder] ： 权重减少0.1 权重0.9
// 1male+running ：+ 连接符
// ([red:blue:0.3]hair) : 渲染红色到30%进度后，开始渲染蓝色
// [[orange sun]] : 权重 0.9*0.9
absurdres, 8k, comic, [viewfinder], (1 male)+running, 1 female ([red:blue:0.3]hair)(long hair) jumping, [[orange sun]], [blue sky], (bird and horse) in sky,
resource
Affix acquisition channels
https://www.fiverr.com/resources/guides/graphic-design/5-ai-art-ideas
-
[A:B:step][Element 1:Element 2:Step ratio] : After the progress of rendering element 1, start rendering element 2 to achieve the mutual superposition of element 1 and element 2. It has a feeling similar to [from:to:when] and can be used for both fusion of terms
[A:step][Element 1: Step Ratio] : The progress from which to start rendering element 1
[A::step][Element 1::Step Ratio] : Stop the rendering of element 1 when the rendering progress is reached (the rendering of other elements continues)
[A|B][Element 1|Element 2] : Alternate mixed rendering of element 1 and element 2, which means rendering A once, rendering B again, rendering A again~~ and repeat until the end
[A|B]C[Element 1|Element 2]Element 3 :
Parentheses plus affixes mean shared elements in the fusion process.