A harsh premise First of all, let’s face the fact that we are dealing with a strictly geographical impossibility: an exact cartographic representation of the world is impossible, as it is impossible to project a 3D object on 2D, while correctly mapping both distances and angles. This is why we like the term "Unexhausted Precision", as it best indicates our utopic endeavour towards geometrical exactness, coupled with our painful awareness that it is not attainable in absolute terms. As for photometric rendering, things don't look any more malleable: let's imagine a glorious sunny day, where the luminous intensity ratio between shade and light is in the order of the hundreds of thousands. It would simply be ridiculous to assess that it can all be expressed through a grey scale from 0 to 255, the renowned 16 Million Colours! The only meagre consolation here being that digital cameras are hindered by the selfsame limitation. An extremely complicated model Building anatomically correct models ought to be considered as a difficult goal in itself, rather than the first stage in building synthetic actors. Some sort of compromise is therefore necessary: in order to show a character paling with fear, we will not sit back and wait for the completion of a fully-fledged simulation of the human circulatory apparatus. On the other hand, we could not even get started at our task, were we not able to simulate breathing, saccadic eye movement, and other involuntary movements. Measuring faces and face measuring tools The difficulty in modelling such a complex object as the human face, - a system of bones, muscles, fat, skin, hair, down - not only has to do with geometry, but also with geography and measurement. We maintain that a systematic mapping must be used, a cylindrical system of coordinates, based on quadrilateral meshes. Such a model is technically complicated to achieve, if one starts from a laser - or any optical - scan. If one aims for accuracy, and starts from a photographic support, the risks is great of ending up entangled in a formidably delicate and difficult process: 6 shots are required, (corresponding to the 6 orthogonal sides of a cube), all taken at the very same time, with identical cameras whose positioning must be accurate to one tenth of a millimetre, and lighting must be diffuse and uniform. Yet the most important steps are still to be accomplished: the removal of lens aberration, inverse perspective, and pixel correlation from the different projections. Suddenly, an illumination: let's build casts, measure them with accuracy, set the resulting mesh in the framework of the Cartesian space: the rest we will solve later. After drawn out purgatorial brainstorming, we came to the conclusion that only a mechanical system could provide an acceptable cylindrical mapping. Therefore we built the "Mercatore Toucher". Sometimes, to get to the data, you have to come up with your own measurement tools. Sometimes, to reach the desired productivity, you have to build your own ultra-fast computers and tie them together into renderfarms. Experimentation Experimentation in Codenrama is currently focused on the simulation of human facial expressions. We hold the current commercial software offerings in this field to be inadequate, and consider the state of the art, as put forth by the special effects film industry, not really satisfactory yet. There's a very logical reason for this situation: all of us, we are machines specialized in human face processing; our analysis and evaluation capabilities, when it comes to human faces, go far beyond assessing the realism of the model: we are - and expect to be -- able to collect semantic information, medical information and so forth. Therefore "fooling the eye" proves to be a tremendously difficult task. Motion Capture techniques applied to faces produce a huge amount of non-structured data, therefore virtually un-editable, and to e be used "as is"; and this situation, all of a sudden, has taken us right back to "analog". It is no secret that Hollywood has failed in its endeavour to create a synthetic Marilyn Monroe, and that they had to fall back on humanoids (Gollum in The Lord of the Rings), aliens, dragons and hairy puppets. To portray a known figure from the past, in fact, it would not be enough, indeed, to measure an imitator: a software would be needed capable of extracting the performer's acting style through the analysis of available reels. This process would require artificial vision techniques coupled with comparative statistical analysis. A real piece of cake!