Colour prediction for sustainable fibre blending

The blending of coloured fibre is explored as a sustainable method of colouration when coupled with sustainable fibre and dyeing choices such as spun-dyed Lenzing Viscose Austria. It was found that a selection of spun-dyed colours (primaries) can be used to create homogenous 4-colour blends when mixed in specific groups. The use of 4-colour blends ensures that the optimal amount of colours within a gamut are produced with the lowest possible number of primaries depending on the acceptable mean colour difference of the 4-colour blends. The acceptable mean colour difference of a blend (measured by averaging each pair of colour differences between the primaries in a blend) can be derived using example 4-colour blends and participant observations at a set viewing distance. Using MATLAB, a method of estimating the number of primaries required to fill a given gamut in CIELAB colour space was developed. Primaries can be distributed across CIELAB colour space and grouped into tetrahedral groups of four for blending. The mean colour difference of the tetrahedral 4-colour blends can be increased or decreased by varying the number of primaries within a gamut. It was also found that the maximum mean colour difference of blends in order for them to appear solid (when viewed at a specific viewing distance) was transferable to blends in knit form. Comparisons of existing blend prediction models with the prediction possibilities of a standard neural network and novel neural network were undertaken using data gathered from 333 blended samples. The results showed that neural networks outperformed the existing prediction models and can be successfully used to predict the colour of blends to an industry standard. The investigations of this thesis have shown that a sustainable colouration system using spun-dyed viscose blends is possible and that accurate colour predictions of these blends can be made