In the Candycore - Sweet Dreamland collection, I chose knit materials to complement the sweet visuals and gentle fantasy of the design. The soft, tactile quality of knits mirrors the dreamy, comforting essence of childhood nostalgia, inspired by Pip & Pop’s whimsical installation art that I encountered at the 2022 Meow Wolf exhibit in New Mexico. This collection combines delicate candy hues, layered textures, and sculptural knit embellishments to create a sensory experience, inviting wearers to indulge in the warmth and innocence of youthful dreams with a sense of playful sophistication.
Installation art, Pip & Pop, Meow Wolf exhibit, 2022, New Mexico
This is a collection of fashion design sketches I created when I was 9, inspired by candy and candy wrappers. I want to breathe new life into these old designs using new technologies like AI.
AI-generated designs based on sketches
I believe that for fashion design, showing only one side of a design is insufficient. Therefore, I experimented with using AI to generate multiple views of the same design for a more comprehensive display.
The video demonstrates my workflow, built with Comfy UI, for transforming 2D images into 3D animations.
The result of the 2D to 3D workflow above
Take a step further, I built a AI-generated line-sketch workflow to create front and back views technical flats, essential for garment production, based on different angel images of a design generated from the 2D to 3D workflow above.
The flowing video demonstrates how the line-sketch workflow operates.
In this process, I found that the AI-generated line-sketch workflow couldn’t perfectly convey garment structure in the way a human-drawn technical flat would. This remains a key issue to address and improve. Therefore, I made adjustments to the AI-generated sketches, producing the following technical flats to be used in future garment production.
Conclusions & Future Directions
In this project, I aimed to design a full-process workflow that transforms a simple 2D sketch into 3D models, capturing every detail of the garment from multiple angles and then generating technical flats essential for fashion production. The project yielded impressive results in two key areas:
-
Converting a simple sketch into a detailed 2D garment rendering.
-
Using AI to turn static 2D images into 3D animations.
However, there is room for improvement—particularly in enhancing AI-generated line sketches to convey garment structure as effectively as human-drawn technical flats.
Looking forward, I have two primary areas of exploration:
-
AI Transformation from 2D to Fully Functional 3D Models: This involves creating 3D models that can be directly used in virtual fashion software like Clo3D and Blender, not just animations. This would make it more efficient to transition from design to physical garment by allowing designers to adjust details in virtual environments. Some training may still be needed for designers to proficiently use tools like Clo3D.
-
Bringing 3D Digital Models into the Physical World: Building on the success of the 3D modeling process, I plan to research and experiment with techniques that transform digital designs into real-world garments. This could include:
-
3D knitting and weaving
-
3D printing with materials engineered to mimic fabric qualities, focusing on finding materials and printing techniques that produce flexible, breathable structures.
-
These concepts require further research and testing to determine the best methods for achieving fabric-like properties in 3D-printed materials and creating garments directly from digital models.