Whole Earth tomography is one of the most quantitative of the Earth sciences, involving about 10 million parameters, thousands of seismograms and time-consuming state of the art calculations. Yet, the methods of presenting results are primitive, and misleading to non-specialists. The pioneers of tomography developed methods to communicate their results to the broader geological community via color images. The effort backfired. We have all seen dramatic cross-sections of blue slabs diving from the surface to the core-mantle boundary, red plumes traversing the whole mantle and color-saturated funnel-shaped features claimed to extend to more than 400 km, or even to the core-mantle boundary, under Iceland. These spectacular images have been reproduced in Science, GSAToday, Science News, Discover Magazine, and most current textbooks and have colored our views of the workings of our planet. These images are artifacts (sometimes devised to support preconceived notions)and do not represent the structure or the behavior of the Earth's interior. Yet they have been used by modelers to guide their calculations and by geochemists to revise their standard model. Problems include along ray smearing, coverage, robustness, color saturation, parallax, background or reference models, smoothing, and selection, orientation and trimming of cross-sections. The neglect of source-side (slab anisotropy) introduces other artifacts into deep images. Quantitative and statistical analysis of tomographic models yield quite different conclusions about slab penetration, the existence of plumes and layered convection than visual inspection of selected cross-sections. Tomographic maps at various depths have fewer of these problems and are easily interpreted by geologists. There is a wealth of information in tomographic models and they are valuable for testing hypotheses.