top of page

AIGC

In multimedia design work, conceptualization and proposal are often required in the realm of visual design. Currently, I frequently use AIGC to generate different compositions and forms for the same theme, providing me with more reference examples for proposal. When it comes to frontend programming, AIGC helps me write code that meets the requirements in a short amount of time, and assists with debugging, greatly enhancing work efficiency.

Danbi_A_bright_white_Taiwan_temple_in_the_style_of_futurist_mec_1757b7de-e0ab-491e-8246-39

Taiwan Temple Reimagined

Software: Midjourney

Taiwanese temples reimagined. This is a topic for personal creation. Growing up in a rural area in Kaohsiung, I was surrounded by many grand temples. I witnessed the construction of these temples from scratch. However, I always felt puzzled about the final step of painting the temples. Due to the poor air quality in southern Taiwan, the exterior of the temples easily accumulates a layer of dark gray dust, resulting in a cheap-looking texture when overlapped with the diverse colors of the temples. Therefore, I hope to rethink the color schemes in temple design to meet current environmental and aesthetic standards.

Danbi_A_bright_white_Taiwan_temples_dragon_pillar_in_the_style__fc04f570-f276-4eb5-ac02-15
Danbi_Taiwan_Temple_in_the_style_of_ingrid_baars_superheroes_hy_12e2cc2d-78e1-4536-85d6-73
Danbi_Create_a_visual_representation_of_a_Chinese_man_embodying_00920299-0abc-4b68-a419-86

When using AI image generation to create the attire of a Taoist priest, inputting keywords related to Taiwanese folk beliefs often yields unsatisfactory results. Instead, keywords related to traditional Chinese religious clothing or ethnic minority attire are required to generate a somewhat suitable appearance. However, this often results in many symbols on the clothing being incorrect, leading to cultural misappropriation issues. Additionally, it's often more difficult to generate male figures than female figures, and even when facial expressions are generated, there's typically little variation.

Danbi_chinese_qing_dynasty_fireball_in_hong_kong_fire_dances_an_c60cf6e2-01d0-4775-97aa-0b

As for Mj v5, it is now capable of generating images based on a piece of text, allowing for more intricate depictions of the interaction between images and scenes. By specifying the camera specifications and image styles, a highly refined representation can be obtained.

​

Prompt: chinese qing dynasty, fireball in hong kong, fire dances and fireballs, in the style of tokina opera, balinese art, traditional poses, epic portraiture, skeletal, sony fe 12-24mm f/2.8 gm, 32k uhd, --ar 16:9 --V5

Touch Penel Content for CES

Software: VScode with ChatGPT

One of my tasks at work is to create interactive touch screen content for an international trade show. Due to the tight deadline, I need to quickly write the code. With the assistance of ChatGPT, all I need to do is write the HTML and create the webpage framework. It can quickly generate the JS code I need for interactivity and debug my code, making the process much more efficient.

Picture1-min.png

Global Event Keyvisions Design

Softwares: Midjourney, Blender

When creating the main visual for an international trade show, I use Midjourney as my source of inspiration for composition. It allows me to collaborate with different departments to determine which composition works best, providing me with a more specific direction when creating models in Blender. However, I used version 4 of Mj at the time, and the visuals I needed were relatively abstract and difficult to express in words. Therefore, when using AI to generate images, they often did not meet my expectations. Thus, I believe that Midjourney is better suited for depicting concrete images that can be described in detail.

2023_0417_Web_Interview_presentation-min.jpg
bottom of page