Flow 互動生成藝術 by Yuming Chen is a highly esteemed winner of the very prestigious 2025 A' Design Award 銀獎 in 生成式設計 Competition Category.

銀生成式· 2025

Flow 互動生成藝術

by Yuming Chen

A' Design Award 銀獎 Winner in 生成式 Category

174,960

Verified Award Winner

Verify Award Status for #162336

Visit Site View at A' Design Award Press Release Award Badge

Colors

The Design

About Flow 互動生成藝術

Flow係一件互動生成藝術作品，會回應文字提示同身體動作。參觀者輸入一段簡短嘅圖像描述，圖像就會即時喺螢幕上生成，唔使等待。攝影機會捕捉參觀者嘅動作，並將捕捉到嘅動態轉化為數據，用嚟改變生成圖像入面嘅粒子同其他視覺效果。快速嘅生成模型加上預設500乘500像素嘅解像度，令回應保持流暢。Flow令冇技術訓練嘅人都可以直接塑造人工智能圖像。

The Recognition

About the A' Design Award 銀獎

Flow 互動生成藝術是 A' Design Award 銀獎得主，這是授予頂級、華美而令人印象深刻的設計的極為崇高的認可。銀獎得主是極其卓越、極具創造力的作品，展現出其設計師傑出的專業技能、才華與敏銳。預計Flow 互動生成藝術將成為一件出色的典範之作，兼具非凡的專業性與深刻的創新性。這件贏得 A' Design Award 銀獎的設計是一項備受矚目、令人欽佩的發明，造福並拓展藝術、科學、設計和技術的界限。Flow 互動生成藝術是一件令人羨慕的創作，它喚起渴望，帶來積極的情感，並激發驚嘆與讚嘆。

Behind the Design

Press Kit

Credits, team members and technical details provided by the designer.

Unique Properties: "Flow" is a new attempt for interactive art in the field of AI, which realises real-time interaction between the audience and AI artworks by introducing large models of generative art and motion capture technology. After the audience inputs the description of the image, the AI artwork can be generated on the screen in real time without additional waiting, and the audience can influence the formed AI artwork through the change of body movement, so that it is no longer just a static picture.
Dimensions & Packaging: In order to realise real-time interaction between AI art and the viewer, large models with low resolution but fast generation speeds, such as SD turbo, need to be selected. At the same time, considering the smoothness of response and ensuring proper resolution, I preset the resolution of the generated image to 500×500 so that the FPS can be kept around 10 when the viewer interacts with the screen.
Tags: AI, Interactive Art, Digital Art, Generative Art
Production Technology: Plug-ins are made by writing Python code in Touch Designer to connect to the API port of the large model to get the big model. Create visual effects such as particles in Touch Designer to visually process the generated images. And access the camera port to capture the audience's movement in real time, and convert the movement into data in Touch Designer to influence the effect of the generated image.
Design Challenge: The challenge in the development of this project was how to combine the large model with the digital creative tool TouchDesigner, and how to make the generation of the images smoother and more interactive with the audience. In order to achieve the effect of real-time generation without waiting, the picture quality of the generated images needed to be sacrificed, but I used digital art techniques to secondary process to achieve a better visual effect.
Inspiration: Thinking about the limitations of current AI art generators. In addition to inputting cue words, in what other ways can we influence the generation of AI artworks? How can the general audience (non-professional practitioners) experience the fascination brought by AI-produced art? I achieve real-time interaction between viewers and AI artworks by introducing the fastest-generating AI macromodels into Touch Desiger, so that the experiencers can interact with the AI artworks through the camera.
Operation Flow: Enter prompts to generate images in real time, and wave your hands or any body movement to change and interact with the generated images.
Image Credits: Image #1: Creator Yuming Chen
Image #2: Creator Yuming Chen
Image #3: Creator Yuming Chen
Image #4: Creator Yuming Chen
Image #5: Creator Yuming Chen
Video Credits: Yuming Chen