A novel method to characterize the printer color output with few sparse samples is presented. Color measurements previously obtained on other substrates and stored in the printer are used to increase the accuracy of measurements in a new target media, thus reducing the number of samples needed. A geometrical warp is applied to the color space to adapt the differences between the two media. The warping is built with a small set of measurements on the target media and extended to the entire color. We tested the method on a HP T1100 ink-jet printer, at different levels of sampling –from 27 to 512 points– on uniform and sparse data points, and with seven different substrate families. For media as the only varying factor, mean estimation error below 1dE is obtained for less than 100 uniformly spaced color patches or 50 sparse ones. In conclusion, color space warping is proven to be an effective method to reduce the needed color samples by using previously characterized media.