The paramer mismatch-based spectral gamut mapping framework is an approach which optimizes the spectral reproduction colorimetrically for multiple viewing conditions. Unfortunately, due to the pixelwise nature of this method, almost similar neighboring pixels might be mapped to completely different colorants which yield disturbing banding artifacts. The previous proposed solution for this problem adds some noise to the a* and b* channels of the input images prior to calculating the separation image. Even though this procedure solves the problem of banding artifacts, it adversely affects the graininess of the final print. In this paper, we propose an approach based on both colorimetric and spatial criteria to reduce banding artifacts of the final print. To our knowledge, the proposed method is the first attempt of joint spatio-spectral gamut mapping and separation. It leads to smoother spectral separations by preserving image edges but is still not completely free of artifacts.