Have any questions about Textura? Find your answers here.

Kenji Takabayashi
Written by Kenji Takabayashi

What textiles do not perform properly in Textura

Textura AI works by analysing the microstructure of the fabrics using state-of-the-art neural networks techniques to predict their optical behaviour.


Traditionally, it's been difficult to estimate texture maps from physical material samples without using complex optical devices. These devices take multiple images of the fabric under different illuminations because fabrics are anisotropic materials that reflect light unevenly when the viewpoint or light changes.

How fabrics reflect light is determined by their micro-structure, which is related to how they are fabricated. At Textura AI, we use state-of-the-art neural network techniques to analyze the microstructure of fabrics and predict their optical behavior. We have a huge database of fabric materials captured at the microscopic level using a complex optical device, which means that our limitations are the same as those of such devices.

However, common PBR material models (texture maps) are unsuitable for modeling volumetric effects such as fur and fleece. To correctly reproduce these types of materials, a special render engine and material model are required.

At Textura, we are continually working on expanding the range of available materials for digitalization. However, our current technology has some limitations that we would like to describe below.

Fur and fleece

Fur and fleece are particularly challenging to digitize because they require a special render engine that can reproduce the individual yarns and filaments present in these types of fabrics. Our current solution returns PBR texture maps that are not optimized for such materials.

Suedes and velvets

Our current material model does not return the maps necessary to render suedes and velvets properly. Therefore, the digital version of these types of fabrics is not guaranteed to be realistic.

Meso-structured materials

Our current solution does not support 3D mesostructures, and some fabrics, particularly those used in footwear, interior, and upholstery design, have a significant volume effect. This volume may have been created by natural folds, molding, or just the way the fabric has been built, and the scanner lid may deform this effect or cause it to disappear. However, if the 3D effect is not too high and the scanner is not deformed by the lid, mild mesostructures in the material will be properly represented.

Categories:
Tags: