AI New Era: Imaging is the Most Interesting Breakthrough

Source: Portrait Photography Micro Platform

Interviewee: Joey Chou, AI visual creator, focusing on using AI technology for fashion art creation, Xiaohongshu: JC_Boss

Interviewer: Dong Dong

Q | **What impact or change do you think AI technology has brought to image creation? **

The rapid development of AI in the field of vision has provided creators with great convenience, making it easier to complete some ideas that were previously difficult or costly to achieve. For example, AI technology can be used to simulate characters and environments from different countries, different ages, and different cultures, reducing the cost of on-site shooting or renting props and clothing. In addition, it allows me to complete the work more quickly and efficiently, and can also use AI technology to automatically generate or optimize some details and elements, saving a lot of time and energy in post-processing.

I am full of expectations and concerns about the future development of AI technology. What is expected is that it will bring more possibilities and innovations, making image creation more diverse. The worry is that if creators blindly rely on AI, they will lose their own style, resulting in lack of creativity and convergence of aesthetics. **AI technology is more like a double-edged sword. As a creator, you should always be vigilant and actively cultivate your own independent aesthetics. **

Q | **When did you start to get in touch with AI creation? **

I started to get in touch with AI image generation a year ago, and I have also witnessed the rapid development of AI technology, from a simple plan to a picture that is closer to a real photo. I tried to combine different styles and themes with fashion elements, imitating some film and television dramas, such as "Blinders" and the works directed by Wes Anderson, and I also tried to use AI technology to create some imaginary scenes and fashionable characters.

Q | **In which fields do you think AI pictures will replace traditional photos? **

In some video content fields that require a large amount, fast or limited cost, AI pictures may have more advantages. Especially in advertising, education, scientific research, communication, communication and other fields. However, it is difficult to replace AI technology for portrait photos serving real people, because the whole shooting process is full of real experience and emotion, which cannot be obtained by simulating "shape".

Q | **Talk about your cooperation with international brands in the field of AI imaging. What kind of AI image needs do they have? **

It is a great honor to cooperate with some well-known brands at home and abroad. Their needs are usually to use AI technology to produce some unique and attractive video content for products or activities. In the early stage, I will be given some basic requirements, such as style and theme, and then I will add creativity and details based on this. For example, my latest group of creations, Versace and Barbie's joint summer clothing AI concept pictures, the creation process is very interesting, understand the styles of their respective brands, and then integrate them just right, with the help of AI's imagination, create a world that belongs to Barbie fashion.

/ Versace x Barbie x Dua Lipa /

Versace, Barbie, Dua Lipa joint summer series posters, a very creative cross-border collision in the summer of 2023. Versace's high-end fashion, Dua Lipa's music style and Barbie's sweet temperament are combined together, coupled with AI's unlimited imagination, to generate these surreal fashion images. I hope to express a deeper understanding and admiration for the series through these images, show the brand culture and fashion charm, and declare the infinite possibilities of AI art.

Q | **Where does your creative inspiration usually come from? **

From some of my favorite movies, TV series, music, and other art categories, I will learn some knowledge about shooting skills, clothing styles, artistic elements, etc. from them.

/Butteraks: Street Fashion/

This is a set of AI fashion images created based on the British historical crime drama "Bathers" as the mother. I like this drama very much, it brings out the aesthetics of violence to the fullest, and the characters, stories, and music all have unique charms. The shots of this drama use a lot of wide-angle close-ups and long-range scenes. The pictures are full of dynamics and have great visual impact. In terms of clothing, the designer used dark tones and low-saturation colors, as well as some offensive and symbolic accessories. The newsboy hat is a symbol of his gang and also has a unique British flavor. After watching the episode, a group of models walking on the streets of Britain unconsciously appeared in my mind, so I used AI technology to highlight the characters from the scene, and strengthened the fashion sense of the picture. Start with the shooting techniques, clothing style, and creative background, summarize the information and write out keywords, and then use AI software to generate pictures, continuously adjust and optimize details, and some even use Photoshop to finely integrate the characters and background, adjust the color, light, Contrast, etc., to make the picture look more coordinated.

/Wes Anderson: A Symmetrical Dopamine Journey/

This is also a group of fashion works. I like Wes Anderson's film style very much. The high saturation and warm colors bring bright, gorgeous and fairytale effects to the picture. Symmetrical composition, panning shots, overhead shots and other techniques bring a regular, coordinated, and flat vision to the picture. These lens languages make each picture like a finely crafted work of art. I combined the hot dopamine color scheme this summer with autumn and winter clothing, and incorporated Wes Anderson-style retro tones, trying to create a bright, jumping, and energetic autumn and winter retro style.

Q | **Talk about the software. Do you think there are any differences between different software after using it? **

At present, the two most commonly used AI image generation tools in China are Mid Journey and Stable Diffusion. Both tools can generate images according to text prompts. MJ needs to pay, SD is open source and free. MJ performed better in picture quality, detail and realism, especially in portrait generation. SD provides more options in terms of picture customization and diversity, and can adjust the size, strictness, seed value, sampler and other parameters of the picture. MJ's creativity is richer and more suitable for free play creation, while SD's controllability is better and it is more favored by e-commerce designers. The combination of the two is currently the most ideal way of drawing.

Q | **Using AI to create, how to ensure accurate realization of expectations? Is there any secret to giving keywords? **

AI technology itself has a certain degree of uncertainty and randomness, and each software has its own algorithm and model, so it is difficult to guarantee that the graph will fully meet expectations. But there are still some methods or techniques that can improve the quality and accuracy of the plot. Be as clear, specific, concise, and concise as possible, and avoid ambiguity and redundancy. For example, to generate a blonde girl with a big smile, use "blonde girl smiling" instead of "a young and beautiful girl with long blonde hair and blue eyes and a big smile on her face".

Q | **When combining real photos, how to make the fusion image more natural? **

I also tried to make AI pictures for relatives and friends, and let them be the characters in their favorite fairy tales or dramas. If you want to produce a natural and stable picture, you first need to choose a clear frontal photo, which can ensure that the software can better recognize and analyze the facial features and expressions of the person. The first generated picture may have some unsatisfactory aspects, such as color, light, contrast, and even the facial features will look distorted. You can try to adjust it a few times, or you can use the image editing software specially designed for human faces to make adjustments. The picture looks more similar to me.

Q | **If you expect it, in what direction do you think AI imaging technology will develop? **

It will be more popular and convenient, and more personalized and differentiated services will be provided. AI imaging technology may appear in more mobile device applications or software, allowing users to use it anytime, anywhere.

Q | **Which style of creation do you want to try next? **

I want to try some sci-fi or fantasy styles, combined with fashion elements, to create a concept of future clothing. It will also continue to create themes based on film and television drama scenes, and tell stories of adventure and adventure.