This quite common and frequently used technique to photograph subjects with and without specular/reflectivity turned up on a couple websites recently, with discussion seemingly focused on how new, useful and technically challenging it could be. Its not complicated or hard at all and so i thought id do a brief blog posting rather than confront the boastful, misguided egos that populate forums!
The technique of cross polorisation has been around for probably longer than 20 years or so - used alot by wildlife photographers/general photographers to minimise the reflectivity elements in scenes caused by flash photography or just highly relfective surfaces in general...such as wet or shiny things. Its also a common technique to analyse surface defects and so on in more scientific based cases.
I first used it about 15 years ago when doing photography (old skool chemicals style) at college to do closeup portraits whilst minimising the greasyness of the skin (particularly cos we were all teenagers at the time). People generally think it just makes things look weird. Well yes - it can do with general photography - as it can remove the specular highlights and visual cues you expect to see in pictures and how you see things in the real world.
It has a great use in CGI however and is used by pretty much every VFX house going - particularly for digital-double work. I recall reading about it in one of Paul Debevecs papers about the reflectivity of the human face from 2000 and have used it ever since to do texture photography (when possible). A progression of this work from 2000 was released in 2007 and demonstrates further relatively simple ways of abstracted data from arrayed/controlled photographs.
On page 7 of the paper - some images and figures briefly describe how to seperate the diffuse and specular components of the skin.
Cross polorisation means there is both a poloriser filter on both the camera and the light sources. If you take a picture of the face with the filter on the camera perpedicular to that of the light sources - you will recover a diffuse only element. A second picture with all filters parralell will yield a picture with reduced diffuse and no include specular components.
A simple photoshop operation of difference between the two will recover the specular component - which is a great start to having a detailed reflection/specular map. These two maps alone will get you more than half way - the flat diffuse and reflectivity.
But wait a sec....so how do we take these pics?! Thats the part people neglect to mention like its some secrest knowledge or something. Well - you will need a polorisation filter for your camera...available in any camera shop anywhere pretty much...or online of course.
You will also need some polorisation film - which can be bought in various sizes/gages from optical companies pretty cheaply. You just cut a piece off and tape it over your camera flashes (with the correct/matching orientation) and your all set. Uh oh - i may of intentionally upset some people now by mentioning how simple it is.....nevermind - do a quick google for cross polorisation and you will find some other pages/blogs talking about it from about 5 years ago. Plus its mentioned in numerous Cinefex issues throughout the past decade.
Heres a quick example of what you end up with. Now this isnt an ideal lighting setup ive used here - and there is some spill from a lightsouce elsewhere in the room that isnt polorised...but as you can see - its a great base for the colour/diffuse map and spec maps.