• Journal of Bio-Environment Control
• /
• v.33 no.2
• /
• pp.120-128
• /
• 2024

Since ChatGPT was released in 2022, the generative artificial intelligence (AI) industry has seen massive growth and is expected to bring significant innovations to cognitive tasks. AI-based image generation, in particular, is leading major changes in the digital world. This study investigates the technical foundations of Midjourney, Stable Diffusion, and Firefly-three notable AI image generation tools-and compares their effectiveness by examining the images they produce. The results show that these AI tools can generate realistic images of tomatoes, strawberries, paprikas, and cucumbers, typical crops grown in greenhouse. Especially, Firefly stood out for its ability to produce very realistic images of greenhouse-grown crops. However, all tools struggled to fully capture the environmental context of greenhouses where these crops grow. The process of refining prompts and using reference images has proven effective in accurately generating images of strawberry fruits and their cultivation systems. In the case of generating cucumber images, the AI tools produced images very close to real ones, with no significant differences found in their evaluation scores. This study demonstrates how AI-based image generation technology can be applied in agriculture, suggesting a bright future for its use in this field.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2024.01a
• /
• pp.293-296
• /
• 2024

현재 생성형 AI가 활발히 연구되고 있는 가운데, 대부분의 이미지 생성 AI는 프롬프트를 기반으로 한 Text-To-Image 방식을 주로 사용하고 있다. 하지만, 프롬프트 기반의 생성 AI는 실제 서비스에 도입하기 어려운 점이 많다. 여러 이미지 중, 하늘 이미지는 메타버스 등 가상 공간에서 매우 자주 사용되는 이미지 중 하나이면서 여러 입력값에 의해 이미지가 달라진다. 이 논문에서는 GAN을 활용해 기상 조건에 적합한 하늘 이미지를 생성하는 프로그램을 설계 및 구현한다.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2023.07a
• /
• pp.569-572
• /
• 2023

최근 하루가 다르게 발전하고 있는 생성형 AI가 창작과 예술 분야에 어떤 영향을 미칠 수 있는지, 새롭게 등장하고 있는 다양한 분야에서 활용 가능한 획기적인 기능 등을 살펴보고 이를 바탕으로 새로운 창작 방향을 제시할 수 있는 방법들을 살펴보려 한다. 최근, 작곡가와 소설가들은 물론, 디지털 아티스트들까지도 생성형 AI를 활용하여 독특한 음악, 글, 그리고 이미지를 창조하는데 성공했다는 사례들이 속속 드러나고 있고 영상, 게임, 웹툰 등 많은 산업현장에서 직접적인 활용방법에 대한 연구결과가 등장하고 실제 적용 사례도 늘어나고 있다. 이미지 생성기인 미드저니와 스테이블디퓨전 같은 도구들은 혁신적인 방법으로 빠르게 높은 퀄리티의 이미지를 생성하고 다양한 아이디어를 제공 받을 수 있는 도구로 창작과 예술 분야에서 큰 관심을 받고 있다. 이러한 발전은 창작과 예술 분야에서 생성형 AI의 무한한 가능성을 보여주는 한편, 인간의 창의성 침해와 예술가들의 노력 희석에 대한 비판적 시각을 불러일으키기도 한다. 본 연구는 이런 다양한 관점에서 창작·예술 분야의 생성형 AI 활용을 깊이 있게 탐구한다. 그 과정에서 여러 생성형 AI 도구들, 특히 이미지 생성기 미드저니와 스테이블디퓨전의 기능과 활용 방안, 그로 인한 사회적, 윤리적 측면을 분석하며, 창작·예술 분야에서의 생성형 AI 활용의 적절한 방향성과 미래 전망을 제시해 보고자 한다.

• The Journal of the Convergence on Culture Technology
• /
• v.9 no.6
• /
• pp.237-244
• /
• 2023

Today, AI is being applied to various industrial fields, leading to a paradigm shift in the overall industry. In the fashion industry, AI is also used to predict trends and provide various services for consumers, and in particular, AI image creation tools have the potential as a tool for fashion design development. This study investigated the possibilities and limitations of using Midjourny for fashion design development by creating images using Midjourney among AI image creation tools and identifying its characteristics. The characteristics of images created in Midjourney are as follows. First, it has the intuitiveness to create images by intuitively applying or combining images corresponding to commands. Second, there is randomness in which different images are generated when the same command is entered at different times. Third, when using existing images and commands together, the image created in Midjourney is more dependent on the existing image than the command. In conclusion, Midjourny's various image creation functions and the ability to change images according to commands can be helpful in developing original fashion designs. However, it is important to note that fashion designs that cannot be worn or made are sometimes presented. It is expected that the results of this study will serve as basic data for the use of AI image creation tools for fashion design development.

• The Journal of the Convergence on Culture Technology
• /
• v.10 no.3
• /
• pp.199-207
• /
• 2024

Today, AI image creation technology is being expanded and utilized across industries. Accordingly, various AI image creation programs optimized for the fashion industry are being developed and commercialized. In this study, we compared and analyzed the visual characteristics of fashion images created by AI image creation programs such as Playground, Midjourney, and The New Black to identify the characteristics of each program and point out areas where each program can be used and problems. The results are as follows: First, while Playground and Midjourney intuitively applied the contents of the command to create images that were different from actual fashion trends, Dannew Black created images that were relatively similar to fashion trends. Second, while Playground separates or combines images corresponding to the command content, Midjourny tends to create new images by adding and fusing various details. Third, in Playground, colors not included in the command appear randomly, and in The New Black, colors not included in the command appear coordinated, and Midjourney generates the color specified in the command relatively accurately. In conclusion, Midjourney can be used when seeking inspiration for developing unique and creative fashion designs, and The New Black can be helpful in referencing fashion trends or fashion styling. On the other hand, playgrounds can be somewhat confusing when it comes to color creation, so this is something to be careful about. It is expected that AI image creation tools can be used more efficiently in fashion design development.

• The Journal of the Convergence on Culture Technology
• /
• v.10 no.5
• /
• pp.397-406
• /
• 2024

Today, in the field of contemporary fashion design, AI is being actively utilized as a new design tool, leading to a new paradigm of collaboration between designers and AI. This study is about a method for developing integrated fashion design through collaboration between human designers and AI. The purpose of this study is to analyze the visual and formative characteristics and changes of fashion design images generated using the AI generation program Midjourney, thereby expanding the understanding and utilization methods of AI image generation programs in fashion design development. The results of this study are as follows. First, Midjourney has the characteristic of relying more on the characteristics of the existing image used rather than the command when creating the image. It also creates new images by distributing and applying the design through an eclectic interaction between the costume and the image background. By excluding the names of fashion items from the commands, you can generate images that can give you more diverse ideas. Second, Midjourney initially expressed clothing colors using colors used in fashion schematics in color creation, and gradually expanded to various color series. Third, there is a kind of compromise between color and design when Midjourney creates an image, and accordingly, by specifying and limiting the image background and clothing colors, more diverse and advanced fashion design images can be created.

• The Journal of the Convergence on Culture Technology
• /
• v.10 no.2
• /
• pp.343-348
• /
• 2024

Today, AI image creation programs are optimized for various and specialized purposes such as fashion product advertising, customized fashion style suggestions, and design development, and are actively utilized in the fashion industry. Meanwhile, color is a powerful formative element and plays an important role in expressing images for suggesting products or fashion styles. This study seeks to expand understanding of the use of Midjourney by identifying the characteristics of color combinations that appear in clothing images created using Midjourney among AI image creation tools. The results of this study are as follows. First, the initial image created in Midjourney reflects the existing image color used to create the image more than the color specified in the command. Second, the color combinations that appear in the clothes of the images created in Midjourney are divided into separate and mixed colors. The ratio of colors expressed in a separate color scheme is affected by the color order specified in the command. The number of colors combined in a mixed color scheme appears as a combination of fewer colors than the total number of colors of clothing in the existing image used to create the image in Midjourney and the number of colors specified in the command. Third, caution is needed because changes in background color can affect the user's color perception of the clothes in the image and the formation of the costume image. It is hoped that the results of this study will be helpful in fashion design education and practice.

• The Journal of the Convergence on Culture Technology
• /
• v.9 no.4
• /
• pp.667-679
• /
• 2023

The advent of image generative AI has made it possible for people who are not experts in art and design to create finished artworks through text input. With the increasing availability of generated images and their impact on the art industry, there is a need for research on how users perceive the process of co-creating with AI. In this study, we conducted an experimental study to investigate the expected and experienced processes of image generative AI creation among general users and to find out which processes affect users' sense of creative agency. The results showed that there was a gap between the expected and experienced creative process, and users tended to perceive a low sense of creative agency. We recommend eight ways that AI can act as an enabler to support users' creative intentions so that they can experience a higher sense of creative agency. This study can contribute to the future development of image-generating AI by considering user-centered creative experiences.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2024.01a
• /
• pp.77-78
• /
• 2024

본 논문에서는 O2O 상점의 자동화에 필수적인 객체 인식 모델의 성능 향상을 목표로, 생성 AI 기술을 이용한 데이터 증강 방법을 제시한다. 제안하는 방법은 텍스트 프롬프트를 활용하여 진열대 상품 이미지를 포함한 다양한 고품질 이미지를 생성할 수 있음을 보인다. 또한, 실제에 더 가까운 상세한 이미지를 생성하기 위한 최적화된 프롬프트를 제안하고, Stable-Diffusion과 DALL-E2의 생성 결과를 통해 비교 분석한다. 이러한 접근 방법은 객체 인식 모델의 성능 향상에 영향을 미칠 것으로 기대된다.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2024.01a
• /
• pp.481-484
• /
• 2024

본 논문에서는 레고(LEGO®) 조립 설명서를 생성하기 위해 딥러닝을 이용한 조립 및 설명서 생성 시스템을 제안한다. 이 시스템은 사용자가 제공한 단일 이미지를 기반으로 레고 조립 설명서를 자동 생성한다. 해당 시스템은 딥러닝 기반 이미지 분할 기술을 활용하여 물체를 배경으로부터 분리하고 이를 통해 조립 설명서를 생성하는 과정을 포함하며, 조립을 위한 알고리즘을 새로 설계하였다. 이 시스템은 기존 레고 제품의 한계를 극복하고, 사용자에게 주어진 부품으로 다양한 모델을 자유롭게 조립할 수 있게 한다. 또한, 복잡한 레고 조립 과정을 간소화하고, 조립의 장벽을 낮추는 데 도움을 준다.