In this paper, we present an image mosaic application and investigate the color rendering performances obtained with various color spaces, among which the nonlinear LUX transform, that is based on a logarithmic image processing model and on biological evidence about the retina processing within the human visual system. We focus on the color matching step for automatic forming of the mosaic image (without any user interaction nor color correction post-processing). We derive a semi-normalized version of the LUX transform, and test its benefits for the mosaic color rendering. The LUX nonlinear color space exhibits a nice behavior of adaptation to light context (logarithmic compression), which is in agreement with the biological processing made by the cones in the eye retina. We compare its behavior for color discrimination and color adaptation with other models proposed in the literature. We show qualitative and quantitative results that demonstrate the improvement achieved by the LUX transform (in terms of color contrast and peak signal to noise ratio PSNR) for building the mosaic.