06.2024 Office Talk
A New Tool in the Workplace - AI Drawing Magic
Group Integrated Efficiency and Retail Planning Headquarters / Digital Marketing and Public Relations Group
 The work "Space Opera House" completed by AI artificial intelligence image generator in 2022 defeated a real person to win the championship of the American Art Expo, causing a global uproar. AI automatic image generation has since received high attention, which also proves that the future world is reserved for people who know how to use AI. Therefore, the Group's Comprehensive Effects and Retail Planning Headquarters held an "AI Drawing Magician Training Class" and invited Mr. Qiu Weitian, the champion of the 2023 Street VS AI Midjournal Image Competition, as a lecturer to enhance the skills of the group's colleagues in using AI.
AI makes it easy for design novices to get started
 In February 2023, a well-known Taiwanese writer mistakenly referred to "AI drawing" as "computer drawing" and was criticized, unexpectedly opening the door to the mass market for AI drawing platforms. Whether it's web literacy, TV celebrities, or graphic experts, they are all passionately explaining the differences between the two. People are also curious and starting to try AI drawing, generating many supernatural images of teeth arranged too tightly, fingers deformed, and hair growing in the air.
However, in less than a year, fans or forums dedicated to discussing and researching AI graphics have long seen these supernatural image products, replaced by professional images that are both fake and genuine. Recently, at the Renaissance Business and Industry Student Art Exhibition, the champion of the "Digital Drawing Group" was even kicked out of using AI drawing. Due to not complying with the competition rules, the prize was cancelled, which also surprised the speed of AI drawing development!
For many non design undergraduate people, the threshold for using computer software for drawing is quite high, but now it can be easily solved through AI. Users only need to provide specific keywords related to the image requirements, and AI can automatically calculate images that match the keywords. Moreover, the more detailed the input words, the more beautiful the AI can produce. If you are not satisfied with the results, you can also provide AI with more words to extend and change, and you can adjust the size or change various styles as needed. In other words, as long as you master the communication skills with AI, you can quickly produce the required images.
Four Spell Generation Methods for AI Drawing
There are a variety of AI drawing platforms on the market, each with different strengths. For example, Dall-E 3 has the best Chinese understanding ability and can be used for free in Bing; Midjournal consistently ranks first in terms of artistry, but users must pay for it; Stable Diffusion is an open resource platform with a high technical threshold and the highest controllability. Its expansion and functionality are highly diverse; Adobe Firefly uses its own Adobe Stock for training, so there is no copyright dispute. Its ability to edit or expand images is very strong, but its ability to make things out of nothing is relatively weak. You may choose a suitable platform based on your own needs and goals.
However, regardless of which platform is used, the first step in the operation requires a spell to initiate a conversation with the AI. The lecturer suggests treating oneself as a client, AI is an outsourced designer, and image requirements are spells, which can be expressed in natural language. In order to make the AI images closer to the imagination in the brain, the instructor categorized the spells into four categories for the convenience of students to follow.
1. Keyword Narrative Method: Only provide keywords, and leave the rest to AI to freely play. For example, "Cat, Sleep, Red Sofa" is suitable for those who have no requirements for atmosphere or composition, are unsure of what they want, or want to see what creativity AI can bring.
2. Flat and straightforward narration: Use simple sentences to outline the desired image. For example, "An orange cat sleeps on a red sofa.". Suitable for simple themes and compositions without setting any style, it can quickly verify whether AI can practice its own ideas.
3. Complete narrative method: describe the scene, character details, environment, and style completely and carefully. For example, in a quiet room, an orange kitten sleeps on a red sofa, with the cat on the left and many pillows on the right, in Van Gogh's oil painting style. Suitable for having a clear composition and style in mind, allowing for precise control of objects or environments.
4. AI robot Prompt method: Discuss with ChatGPT/Copilot the desired composition to generate and have it produce appropriate spells. Especially when AI drawing platforms are not yet mature in understanding traditional Chinese and have difficulty expressing themselves in English, this method can quickly solve problems.
In addition, the instructor also reminded in the course that the biggest weakness of AI drawing is that the longer the narration, the more elements are ignored, especially in the description of "style". Due to the excessive narration of "scenes" and "characters", it may be easy to be ignored. Therefore, how to achieve "balance" is a challenge that creators must face. It is recommended that beginners start their spells with short sentences (no more than 30 words), as AI has a higher chance of fully absorbing instructions.
If you have no ideas or need to extract more inspiration, you can also refer to other people's spells and their generated images on an AI drawing platform. The lecturer also reminds that currently, some paid versions of platforms produce images that can be used for commercial purposes, but it is still necessary to understand the legal terms of each platform before using them.
The students of this "AI Drawing Magician Training Class" include art designers, marketers, and e-commerce business colleagues from retail and other related enterprises of the group. The course is arranged to use the promotional visual aids of retail promotion periods (such as anniversaries and Ghost Festivals) for practical operations. Laughter and discussion echoed throughout the process, and within just 15 minutes, the students had completed their preliminary work. Everyone was satisfied with the results of their first attempt and recognized and looked forward to AI becoming a good helper in the workplace.
 AI Empowerment Leaps Competitiveness Forward
The question of whether AI will replace or assist human work is still highly discussed. Li Feifei, Section Chief of the AI Research Center at Stanford University in the United States, once pointed out that AI technology is the foundation that can "enhance" humans rather than replace them. The strength of generative AI technology is to assist in creativity, conceptualization, and simplifying tedious tasks. After saving a lot of time in existing work, it is even more important to strengthen one's own value. Overall, technology remains people-oriented, and innovation still needs to return to humanity, insight into consumer needs and pain points, and cultivate problem-solving abilities. In the future, it is not AI that can defeat humanity, but people who make good use of AI!
According to a research report by PwC, global GDP may increase by 14% due to AI by 2030, approximately USD 15.7 trillion. Among them, retail, financial services, and healthcare are expected to be the main beneficiary industries, which can be combined with artificial intelligence to improve productivity, product quality, and consumption. Given the rapid development of technology and the ever-changing digital applications, the Group's Comprehensive Efficiency and Retail Planning Headquarters will continue to assist retail enterprises in keeping up with trends, making good use of technology and artificial intelligence to collaborate and leverage their core advantages, and mastering the key competitiveness of future workplaces and enterprises.
#
AI makes it easy for design novices to get started
 In February 2023, a well-known Taiwanese writer mistakenly referred to "AI drawing" as "computer drawing" and was criticized, unexpectedly opening the door to the mass market for AI drawing platforms. Whether it's web literacy, TV celebrities, or graphic experts, they are all passionately explaining the differences between the two. People are also curious and starting to try AI drawing, generating many supernatural images of teeth arranged too tightly, fingers deformed, and hair growing in the air.
However, in less than a year, fans or forums dedicated to discussing and researching AI graphics have long seen these supernatural image products, replaced by professional images that are both fake and genuine. Recently, at the Renaissance Business and Industry Student Art Exhibition, the champion of the "Digital Drawing Group" was even kicked out of using AI drawing. Due to not complying with the competition rules, the prize was cancelled, which also surprised the speed of AI drawing development!
For many non design undergraduate people, the threshold for using computer software for drawing is quite high, but now it can be easily solved through AI. Users only need to provide specific keywords related to the image requirements, and AI can automatically calculate images that match the keywords. Moreover, the more detailed the input words, the more beautiful the AI can produce. If you are not satisfied with the results, you can also provide AI with more words to extend and change, and you can adjust the size or change various styles as needed. In other words, as long as you master the communication skills with AI, you can quickly produce the required images.
Four Spell Generation Methods for AI Drawing
There are a variety of AI drawing platforms on the market, each with different strengths. For example, Dall-E 3 has the best Chinese understanding ability and can be used for free in Bing; Midjournal consistently ranks first in terms of artistry, but users must pay for it; Stable Diffusion is an open resource platform with a high technical threshold and the highest controllability. Its expansion and functionality are highly diverse; Adobe Firefly uses its own Adobe Stock for training, so there is no copyright dispute. Its ability to edit or expand images is very strong, but its ability to make things out of nothing is relatively weak. You may choose a suitable platform based on your own needs and goals.
However, regardless of which platform is used, the first step in the operation requires a spell to initiate a conversation with the AI. The lecturer suggests treating oneself as a client, AI is an outsourced designer, and image requirements are spells, which can be expressed in natural language. In order to make the AI images closer to the imagination in the brain, the instructor categorized the spells into four categories for the convenience of students to follow.
1. Keyword Narrative Method: Only provide keywords, and leave the rest to AI to freely play. For example, "Cat, Sleep, Red Sofa" is suitable for those who have no requirements for atmosphere or composition, are unsure of what they want, or want to see what creativity AI can bring.
2. Flat and straightforward narration: Use simple sentences to outline the desired image. For example, "An orange cat sleeps on a red sofa.". Suitable for simple themes and compositions without setting any style, it can quickly verify whether AI can practice its own ideas.
3. Complete narrative method: describe the scene, character details, environment, and style completely and carefully. For example, in a quiet room, an orange kitten sleeps on a red sofa, with the cat on the left and many pillows on the right, in Van Gogh's oil painting style. Suitable for having a clear composition and style in mind, allowing for precise control of objects or environments.
4. AI robot Prompt method: Discuss with ChatGPT/Copilot the desired composition to generate and have it produce appropriate spells. Especially when AI drawing platforms are not yet mature in understanding traditional Chinese and have difficulty expressing themselves in English, this method can quickly solve problems.
In addition, the instructor also reminded in the course that the biggest weakness of AI drawing is that the longer the narration, the more elements are ignored, especially in the description of "style". Due to the excessive narration of "scenes" and "characters", it may be easy to be ignored. Therefore, how to achieve "balance" is a challenge that creators must face. It is recommended that beginners start their spells with short sentences (no more than 30 words), as AI has a higher chance of fully absorbing instructions.
If you have no ideas or need to extract more inspiration, you can also refer to other people's spells and their generated images on an AI drawing platform. The lecturer also reminds that currently, some paid versions of platforms produce images that can be used for commercial purposes, but it is still necessary to understand the legal terms of each platform before using them.
The students of this "AI Drawing Magician Training Class" include art designers, marketers, and e-commerce business colleagues from retail and other related enterprises of the group. The course is arranged to use the promotional visual aids of retail promotion periods (such as anniversaries and Ghost Festivals) for practical operations. Laughter and discussion echoed throughout the process, and within just 15 minutes, the students had completed their preliminary work. Everyone was satisfied with the results of their first attempt and recognized and looked forward to AI becoming a good helper in the workplace.
 AI Empowerment Leaps Competitiveness Forward
The question of whether AI will replace or assist human work is still highly discussed. Li Feifei, Section Chief of the AI Research Center at Stanford University in the United States, once pointed out that AI technology is the foundation that can "enhance" humans rather than replace them. The strength of generative AI technology is to assist in creativity, conceptualization, and simplifying tedious tasks. After saving a lot of time in existing work, it is even more important to strengthen one's own value. Overall, technology remains people-oriented, and innovation still needs to return to humanity, insight into consumer needs and pain points, and cultivate problem-solving abilities. In the future, it is not AI that can defeat humanity, but people who make good use of AI!
According to a research report by PwC, global GDP may increase by 14% due to AI by 2030, approximately USD 15.7 trillion. Among them, retail, financial services, and healthcare are expected to be the main beneficiary industries, which can be combined with artificial intelligence to improve productivity, product quality, and consumption. Given the rapid development of technology and the ever-changing digital applications, the Group's Comprehensive Efficiency and Retail Planning Headquarters will continue to assist retail enterprises in keeping up with trends, making good use of technology and artificial intelligence to collaborate and leverage their core advantages, and mastering the key competitiveness of future workplaces and enterprises.
#