This tutorial is collected at: Summary of AIGC tutorials from entry to mastery
AIGC technology is constantly updated and iterated, and more and more new ways of playing have appeared in China, such as the AI painting creative text that everyone has been discussing recently.
In the past week, I have studied these new gameplays and summarized a complete production process.
The mainstream creative text play methods are mainly divided into four categories:
- Light and shadow text
- embedded text
- hidden text
- poster text
Each type of text production process is roughly these four steps
Make Text Basemap
select large model
Write keywords (Prompt)
ControlNet parameter setting
Core: Vincent map + text base map (ControlNet)
I have organized all the materials used in this tutorial and put them in the network disk link at the end of the article, so you don't need to spend a lot of time and energy looking for them.
I've spent a lot of time putting together these and hope they'll be helpful to you.
Let's get straight to the point:
General practice
Make Text Basemap
Meituxiuxiu, PS can be used, I use the unprofessional Meituxiuxiu here
Black text on white background for poster text and embedded text
Hidden text and light and shadow text use black and white text
Open Meitu Xiuxiu = "Picture Beautification =" Make Wallpaper = "Choose Background Color = "Next (text editing, but more introduction)
Select a large model:
Use real pictures: majicmixRealistic, chilloutmix
Other uses: revAnimated, realisticVision, deliberate
All the model materials used in this article will be packaged and organized for everyone, and placed in the network disk link at the end of the article
If you have energy and time, you can try more, and you may get unexpected results.
1. Light and shadow text practice:
1.1 Make a base map (white text on a black background):
This picture looks kind of scary
1.2 Select the large model:
Choose a realistic model (majicRealistic or chilloutmix)
1.3 Write keywords (Prompt):
Common words for urban architecture:
Best quality, ultra high definition, masterpiece, ultimate details, 8K,
Urban night view, buildings, cars, headlights, streets, roads, water reflection
Best Quality, Ultra HD, Masterpiece, Extreme Detail, 8K,
City night, buildings, cars, headlights, street, road, water reflection
Common characters:
The highest quality, masterpiece, ultimate detail,
1 girl, beautiful, full body, long skirt, standing with hands in pockets
Highest quality, masterpiece, extreme detail,
1 girl, beautiful, full body, long skirt, standing, hands in pockets
Keyword classification query of StableDiffusion
① Write the picture quality first
② Then write the appearance description of the young lady
③It is better to add some words such as "sunshine", "light and shadow", "chiaroscuro", and add a little weight
sunlight, light and shadow, chiaroscuro
Positive keywords:
((masterpiece)),(highres:1.3), (((best quality))), ((ultra-detailed)), ((illustration)),(photorealistic:1.3),(realistic:1.4),(Cinematic Lighting:1.3),
aynamic angle,in the winter,(snow white forest:1.1), hazy mist,(on the hot springs:1.3) ,(many thin detailed snowflakes:1.3), (bloom), (shine), (twinkling stars:), (snowing), many white tree, detailed snowflakes,
cowboy shot,
(1 girl),(((solo))),running,(between legs:1.3),
(chinese girl:1.1),
hanfu,hanfugirl, beautiful detailed eyes,(black eyes:1.2),(pale firered hair:1.3),light smile,
(big breasts:1.1),
,(beautiful detailed face:1.3),(cute), blushing, light smile, makeup ,lipstick,detailed clothes,(pale red transparent tight floral qipao dress:1.35),
(white fishnet thighhighs:1.25).
lora:add_detail:0.5,
sunlight, light and shadow, chiaroscuro
Negative keywords:
(cleavage:1.4),(Multiple people),(((simple background))),monochrome ,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, lowres, bad anatomy, bad hands, text, error, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, ugly,pregnant,vore,duplicate,morbid,mut ilated,tran nsexual, hermaphrodite,long neck,mutated hands,poorly drawn hands,poorly drawn face,mutation,deformed,blurry,bad anatomy,bad proportions,malformed limbs,extra limbs,cloned face,disfigured,gross proportions, (((missing arms))),(((missing legs))), (((extra arms))),(((extra legs))),pubic hair, plump,bad legs,error,legs,username,blurry,bad feet ,(cleavage:1.4)
1.4 Important ControlNet parameter settings
I use the tile model
Special attention should be paid to the control weight (Control Weight) and the boot termination timing (Ending Control Step)
Control Weight : set between 0.35~0.55
Boot termination timing (Ending Control Step) : 0.7~1, the closer to 1, the clearer the word
There are no fixed parameters, and different characters need to be adjusted separately, so you can only try more by yourself
1.5 Final effect:
2. Embedded text operation:
2.1 Make a base map (black text on a white background):
2.2 Select a large model
The choice here is: deliberate, other revAnimated, realisticVision can also try
2.3 Write keywords (Prompt)
Plum blossom, red, petals, winter, snow white background, old tree trunk, bright blue sky, tough, vivid, vitality
Plum Blossom, Red, Petals, Winter, Snowy Backdrop, Ancient Trunk, Bright Blue Sky, Resilient, Vibrant, Dynamic
Positive keywords:
Best quality, ultra high definition, masterpiece, ultimate details, 8K,
Plum Blossom, Red, Petals, Winter, Snowy Backdrop, Ancient Trunk, Bright Blue Sky, Resilient, Vibrant, Dynamic
Negative keywords:
I found a negative word about the scenery here:
(normal quality), (low quality), (worst quality), paintings, sketches,fog,signature,soft, blurry,drawing,sketch, poor quality, uply text,type, word, logo, pixelated, low resolution.,saturated,high contrast, oversharpened,dirt,
2.4 Important ControlNet parameter settings
Then there is the choice of controlnet, here we must choose canny, lineart, softedg, scribble and other line draft models
In my actual test, the effects of canny and scribble are not bad. You can try these two models first.
Reduce the weight so that the text can better blend with the scene
You can set the weight between 0.65 and 0.85 , and increase the weight if you want the text to be more prominent
2.5 Final effect
3. Hidden text operation:
black and white
3.1 Make a base map (white text on a black background):
3.2 Select a large model
The realisticVision I chose here, others can also try: deliberate, revAnimated
3.3 Write keywords (Prompt):
关键词: 英文:Oriental Pearl Tower, Dazzling Light, Night Sky, Spherical Structure, Skyline, Skyscrapers, Huangpu River, Bright Pearl, Shanghai.
Chinese: Oriental Pearl, Dazzling Light, Night Sky, Spherical Structure, Skyline, Skyscrapers, Huangpu River, Bright Pearl, Shanghai.
Positive keywords:
Best quality, ultra high definition, masterpiece, ultimate details, 8K,
Urban night view, buildings, cars, headlights, streets, roads, water reflection
Oriental Pearl Tower, Dazzling Light, Night Sky, Spherical Structure, Skyline, Skyscrapers, Huangpu River, Bright Pearl, Shanghai
Negative keywords:
(normal quality), (low quality), (worst quality), paintings, sketches,fog,signature,soft, blurry,drawing,sketch, poor quality, uply text,type, word, logo, pixelated, low resolution.,saturated,high contrast, oversharpened,dirt,
3.4 Important ControlNet parameter settings
Note: The parameter setting of controlnet is more important
Here we need to download a new controlnet model
The model can be downloaded from the network disk at the end of the article
Model storage location: sd directory\models\ControlNet
No preprocessor is required, the model will be our newly downloaded one
Special attention should be paid to the control weight, as long as it is between 0.45 and 0.65
To control the weight, you can go back and forth to try the effect. If the words are clearer, the weight should be larger.
3.5 Final effect
The effect is not very good, you can adjust the parameters slowly.
4. Poster text practice:
4.1 Make a base map (black text on a white background):
4.2 Select large model
What I choose here is: revAnimated
4.3 Write keywords (Prompt):
Description: This painting depicts a pool of vibrant lotus flowers. In the center of the picture, a lotus flower unfolds in the sun, the pink petals are as smooth as silk, exuding a faint fragrance. The green lotus leaves unfold like an umbrella, floating on the tranquil pool water. A few dragonflies shuttle among the lotus leaves, adding to the warmth of summer. In the distance, windswept reeds and a pale blue sky make the perfect backdrop.
Keywords: English: Lotus, Pink Petals, Sunshine, Green Leaves, Pond, Dragonflies, Reeds, Light Blue Sky, Serene, Summer.
Chinese: lotus, pink petals, sunlight, green lotus leaves, pond, dragonfly, reeds, light blue sky, serenity, summer.
Positive keywords:
Best quality, ultra high definition, masterpiece, ultimate details, 8K,
Lotus, Pink Petals, Sunshine, Green Leaves, Pond, Dragonflies, Reeds, Light Blue Sky, Serene, Summer,
Negative keywords:
I still use the above keywords here:
(normal quality), (low quality), (worst quality), paintings, sketches,fog,signature,soft, blurry,drawing,sketch, poor quality, uply text,type, word, logo, pixelated, low resolution.,saturated,high contrast, oversharpened,dirt,
4.4 Important ControlNet parameter settings
The preprocessor selects invert
Model selection depth
Knowledge Supplement :
If it is to generate liquid material text, such as milk, you can use canny, lineart, softedg, scribble these few line drafts
Then slightly lower the control weight of the controlnet (between 0.6 and 0.8)
In this way, the text will have the feeling of liquid splash
4.5 Final effect
Network disk data package
Read the original text to get the information package of AI painting creative text: The whole process of AI painting creative text is revealed, your ultimate text art practice book