This paper proposes a compact description method for 3D gamut shell. The real 3D image-to-device gamut mapping is a best way to make the appearance matching in between display and print images suppressing the source color information losses. The image gamut surface is shaped as polygon meshes by extracting the most outside points from the random color distributions of natural image. The gamut shape data should be attached to image in compact format. In our proposal, first, the image gamut shape is given by the 2D color distance array, called r-image, which is measured as radius distance r(, Φ) in discrete polar angles from the image color center. Next, the r-image is transformed into svd-image by SVD (Singular Value Decomposition). Finally, the compact parameters (the major singular values neglecting the small values and the orthogonal eigen vectors) are delivered to the users. The r-image can be approximately restored from the reduced SVD parameters and the image gamut surface is reconstructed and used for the users to map the image data into the inside of the device gamut referencing to the image gamut shape. The paper discusses how the r-image is compressed and the gamut shape could be well recovered with its surface colors. The SVD description method is compared with DCT based compaction in spatial frequency domain.