SYNTHETIC SEWING PATTERNS
👁 57 |
| 2024-12
With generative tools, we can visualize almost any idea imaginable. However, when it comes to transforming these ideas into tangible matter, numerous material and craft-related questions arise. This experiment consists of two phases:
The synthetic creation of convincing outfit designs inspired by traditional rug patterns and cellular automata.
The interpolation of potential sewing patterns from the rendered images, using training data of digital outfits and their corresponding 2D sewing patterns.
This approach is highly experimental and should be viewed as the beginning of a feasibility study. It departs from traditional sewing craftsmanship, as it does not incorporate in-depth knowledge of materials, sewing techniques, or construction processes. These steps still require skilled human expertise.
Instead, this experiment opens up possibilities for how generative tools can expand the scope of DIY, promote the integration of recycled fabrics, support creative repair, and explore new ways of engaging with craft and design.
generative fashion outfits
amateur photo of oversized puffy bw casual haute couture outfit made from recycled textile material with traditional highly detailed used weaved textile carpet rug patterns with bitmap and ascii art art included like an eccentric sculpture with complex draperies and ruffles inspired by eccentric Oskar Schlemmer designs with extensions, throws, hood, recycled textile material, foil, trash bags, huge textile spheres, parachute, integrated cushions and textile tunnels and thick black fringes. two giant toy shoes instead of feet like humongous very huge surreal orange thick plastic buckets on the floor made from used deformed matte neon orange plastic material replacing shoes. Adult aged fat old ugly androgyn asian male person with realistic skin and slight wrinkles and long straight grey hair in casual pose and neutral face and eyes closed. plain urban scene with natural diffuse light and plain frontal distressed concrete wall with monchrome graffiti in the background white smoke clouds in background. desaturated colors, film grain and vignette and lens distortion.
a specifically trained Lora
With a custom LoRa specifically trained on both schematic outfit preview images and corresponding sewing patterns in 2D we can come up with a simple trick to get sewing patterns from whatever outfit photo.
We just place a random pixel noise field to the right of the outfit photo, so the diffusion kernel has high freedom to build up a sewing pattern from trained data in relation to the guiding image to the left.
With generative tools, we can visualize almost any idea imaginable. However, when it comes to transforming these ideas into tangible matter, numerous material and craft-related questions arise. This experiment consists of two phases:
This approach is highly experimental and should be viewed as the beginning of a feasibility study. It departs from traditional sewing craftsmanship, as it does not incorporate in-depth knowledge of materials, sewing techniques, or construction processes. These steps still require skilled human expertise.
Instead, this experiment opens up possibilities for how generative tools can expand the scope of DIY, promote the integration of recycled fabrics, support creative repair, and explore new ways of engaging with craft and design.
generative fashion outfits
a specifically trained Lora
With a custom LoRa specifically trained on both schematic outfit preview images and corresponding sewing patterns in 2D we can come up with a simple trick to get sewing patterns from whatever outfit photo.
We just place a random pixel noise field to the right of the outfit photo, so the diffusion kernel has high freedom to build up a sewing pattern from trained data in relation to the guiding image to the left.