Five texture mapping algorithms were developed for simulating textures of physical objects on a CRT. Twenty-one thread winding cards with different colours were captured by a digital camera for generating texture profiles which were used to describe the statistical colour distributions of images. A pair comparison method was used to assess the performance of various texture mapping algorithms. A total of 42000 judgements were made by a panel of 10 normal colour vision observers. The results showed that realistic appearance can be achieved by mapping a single set of tristimulus values onto a texture profile specified by the colour distribution only in luminance factor channel.