With simple color offset drawings, we can guide generative diffusion models to come up with pseudo plausible solutions for textile outfits. With proper combination of drawing style and prompt, this is a pretty versatile way to create intriguing visualizations, that differ from stereotypical AI output.

It is not yet possible to extract plausible instructional patterns or material lists but it shows a plain direct way to bring formal ideas to plausible photo and / or video format.

yarn based body outfit

PROMPT

realistic documentary photo of

androgyn adult person with realisitic skin in a contemporary airy yarn textile outfit.

The oufit is made from different textile yarns and cords and incooperates different contemporary and traditional textile techniques like contemporary crochet, knitting, lace and macrame and DIY improvised style.

The materials range from conventional yarn, recycled yarns to biomaterials.

plain white backdrop.

perspective interpolations

With the help of current state of the art image generators, we are capable to interpolate multiple perspectives from only the frontal view of our drawn/generated outfit. This opens up the possibility to explore the outfit sketch furthermore. The following interpolations are generated using the Nano Banana Model by Google.

Conclusion: On first sight, the interpolations look pretty plausible – but with an eye on the detail, there are inconsistencies between each perspective, as there is no common metadata, that binds all diffusions together. With proper reasoning, this should be a solvable issue.