Click the blue text to follow us
What Can AI Art Do in Public Security?

Gu Xiao, a young police officer in the Data Department of the Hongkou Branch of the Shanghai Public Security Bureau, part-time instructor at the branch, has won the title of excellent WeChat editor from the municipal bureau. In his spare time, he enjoys photography, graphic design, and 3D design, and is skilled in Photoshop, 3D modeling, special effects production, and AI art. Below, he will discuss the applications of AI art in public security work, hoping to resonate with more young police officers.

Recently, the topic of AI has become increasingly popular, seemingly sparking a technological revolution. For example, we are familiar with ChatGPT and Baidu’s Wenxin Yiyan. Many friends are very interested in its “text-to-image” functionality.

What is Text-to-Image?
Text-to-image refers to generating corresponding images from text input.

Editor
Text-to-image, which is AI art, has actually existed for a long time. With the recent launch of the latest V5 model of the AI painting software MIDJOURNEY, its capabilities have reached a new height.




Today, the editor will introduce two mainstream AI art tools: MIDJOURNEY and STABLE DIFFUSION, as well as their applications in public security work.
They operate on the same principle, generating images from text input instead of traditional painting methods. Their advantages lie in high efficiency and quality.

Editor
AI art can create various styles and content, based on neural network learning from diffusion models, rather than searching for existing images online. This means you can create any content you want.

MIDJOURNEY
Advantages:
1. Excels at generating commercial-type images such as illustrations, posters, UI, etc.
2. Web or app-based, cloud computing, convenient and fast, no hardware barriers.
Disadvantages:
Free users are limited to 25 times per day; a membership (30 USD/month) is required to lift restrictions (shared accounts can be purchased on Taobao to save costs).

STABLE DIFFUSION
Advantages:
1. Excels at creating anime and realistic style images.
2. Local deployment, offline use, completely free.
Disadvantages:
Requires GPU computation, with certain requirements for graphics cards (recommended NVIDIA 20 series or higher).


Editor
Next, the editor will discuss the applications of these two AI art software in relation to public security needs. The installation methods will be mentioned at the end, so those in need can start learning~~
1. MIDJOURNEY

▲ Open the MIDJOURNEY application interface, just input the code /imagine and add descriptive words to get started.

Editor
Descriptive words are the descriptions of the required images, guiding the AI on what to draw, and must be in English.
1
Decorate your text with illustrations
Illustration-style works can be used for posters, PPTs, video production, etc.
Descriptive words:
man in the lake on a surf table, summer sunny day, lake, panoramic view, flat illustration, hd
(Translation) A man on a surfboard in a lake, on a sunny summer day, with a panoramic view, flat illustration, HD.


Combine with FLAT ILLUSTRATION to add the content you need, and you can complete high-quality illustrations. Of course, you can also leave it out and see what you get.

For example, if you want to create a PPT on urban governance
Descriptive words:
smart city
(Translation) Smart city.

Adding this to your PPT will enhance its quality significantly, and if combined with AE, dynamic effects can also be created.
Of course, AI art has infinite possibilities, which requires you to use more precise descriptive words to achieve.
Descriptive words:
Design a logo, police car, flat illustration, minimalistic, graphic design
(Translation) Design a logo, police car, flat illustration, minimalistic, graphic design.

2
Illustrate your story
Want to tell a story but lack illustrations? Try MIDJOURNEY.

When you want to tell a story, such as responsible pet ownership
Descriptive words:
an Asian teenager girl, walking her black Scottish terrier dog in the park, Flat illustration, detailed.
(Translation) An Asian girl walking her black Scottish terrier in the park, flat illustration, detailed.

Descriptive words:
a boy standing on the street, a red wallet on the ground, flat design, flat illustration.
(Translation) A boy standing on the street, a red wallet on the ground, flat design, flat illustration.


If you find flat illustrations a bit childish, you can also try different styles.
Descriptive words:
two person from china looking at mobile phone, subway, comic.
(Translation) Two people looking at a mobile phone in a subway station, comic.


Stories can be used for WeChat promotions, publicity, and various other aspects. Remember, AI creation is limitless; keep trying different content and styles.
3
Let’s try some realistic designs.
Generate some realistic works to add color to your poster designs.
Descriptive words:
cinematic lighting, back of a Chinese police wearing black uniform, bladerunner, horror, 4k, movie poster, dark, unreal engine, realistic, cyberpunk, depth of field.
(Translation) Cinematic lighting, back of a Chinese police officer wearing a black uniform, bladerunner, horror, 4K, movie poster, dark, Unreal Engine, realistic, cyberpunk, depth of field.



4
Or, you want to design a mascot for the Youth League Committee.

Descriptive words:
graphic, photorealistic, Portrait, cat in Police suit.
(Translation) Graphic design, photorealistic, portrait, cat in police uniform.

5
Design UI interfaces for your PPT and applications.


Descriptive words:
create management system ui/ux website landing page :: police car :: powerpoint.
(Translation) Create management system UI interface, website landing page, police car, PPT.

Descriptive words:
Website, a minimalistic clean splash page with simple blue, white and black, tech theme ::1.3, modern ui, fullscreen background.
(Translation) Website, a minimalistic clean splash page with simple blue, white and red, police theme, modern UI, fullscreen background.

6
Alternatively, you might want to try different styles of police images.


Descriptive words:
pixar movie poster concept ::1.5 policeman from china ::1.3 street, sunny.
(Translation) Pixar style movie poster, Chinese policeman, street, sunny.


7
Any content, it’s up to you.


Descriptive words:
a detailed police robot::1.5 with white and black metals and blue red alarm lamp, view of a cyberpunk city, ultra photorealistic, ultra detailed, flair, intricate details, 16k, volumetric lighting, shot with Canon EOS R3 Mirrorless Digital Camera, Octane render, ray tracing, chaos 5.
(Translation) A detailed police robot made of white and black metals with blue and red alarm lights, overlooking a cyberpunk city, ultra photorealistic, ultra detailed, flair, intricate details, 16K, volumetric lighting, shot with a Canon EOS R3 mirrorless digital camera, Octane render, ray tracing, chaos 5.

8
Finally, let me share some tips.
When you lack inspiration for scene descriptions, you can ask ChatGPT to provide you with scene descriptions, for example, depicting a telecom fraudster making a phone call:


Copy it into MIDJOURNEY, and you can also ask ChatGPT to modify it based on your ideas.
Descriptive words:
villain in the office ::1.3, black trench coat, shadow-covered face, cunning eyes, old wooden chair, antique telephone, rotary dial, tangled phone cord, dim room, yellow light, irregular shadows, weathered walls, Chinese ink painting, messy desk, newspapers, maps, cigarette, green porcelain vase, withered flowers.
(Translation) A villain in the office, wearing a black trench coat, with a shadow-covered face, cunning eyes, sitting in an old wooden chair, using an antique telephone with a rotary dial, tangled phone cords, in a dim room with yellow light, irregular shadows, weathered walls, with Chinese ink painting, a messy desk, newspapers, maps, a cigarette, a green porcelain vase, and withered flowers.


STABLE DIFFUSION produces very detailed scene descriptions, yielding excellent results.
2
In the Year of the Rabbit, try creating a police rabbit.
Descriptive words:
Detailed portrait photo of (cute and fluffy white bunny) wearing police clothing (black), gray background, full body, cyberpunk, super detailed, 8k, fantasy, dreamlike, surrealism, super cute, ambient warm tones, full body.
(Translation) A detailed portrait photo of a cute and fluffy white bunny wearing black police clothing, on a gray background, full body, cyberpunk, super detailed, 8K, fantasy, dreamlike, surrealism, super cute, warm ambient tones, full body.


3
Try creating an illustration.


4
Or generate some components for promotional materials.





With the unique CONTROLNET plugin of STABLE DIFFUSION, you can create controllable image content by reading sketches, bone diagrams, depth maps, etc.
Descriptive words:
best quality, masterpiece, a girl running on the street, sunlight, high buildings, cloudy, plants.
(Translation) Best quality, masterpiece, a girl running on the street, sunlight, high buildings, clouds, plants.



Although my drawing skills are lacking, the final results are still good. You can choose between MIDJOURNEY or STABLE DIFFUSION based on your needs.
3. About Installation

MIDJOURNEY is web-based. Users only need to register and log in to DISCORD (similar to QQ channels in China) at https://discord.com/ to use it. Members can generate unlimited images (30 USD/month), and shared accounts can be purchased on Taobao to save costs.
Related installation tutorial:
https://www.bilibili.com/video/BV1kL411D7UY

STABLE DIFFUSION is a PC application that can be downloaded from GITHUB or as an integrated package.
Related installation tutorial:
https://www.bilibili.com/video/BV1NX4y1Q7MH/

In the age of AI, those who master AI tools will lead the way. Interested friends should give it a try!



Longqing Public Security WeChat Official Account Issue 2759
Material: WeChat Official Account Hongqing Listening
Editor: Fei Yufan
Proofreader: Guo Jun
Reviewer: Yan Lijuan
