Since Shinen pointed sometime ago that they could be using tesselation on their next game(Fast Racing Neo) I thought that it was worth the try by asking them. Didndt get a yes or no for an answer, but at least got an interesting comment that might point to that possibility

https://twitter.com/ShinenGames/statuses/512491838012272641





Very interesting isnt it?
After reading this I wondered how would they preserve the models as intact as possible,texture compression has a limit of 6x and introduces artifacts and since we are talking about 1GB of storage per mesh the texture compression would not be enough since wii u has just 1GB of RAM for games, thats when I recalled that tesselation does some kind of geometry compression and so I decided if there could be a connection and in fact seems that way


Example
http://www.academia.edu/2696670/Displacement_Mapping
"
Domain surface and a scalar displacement map

One application of displacement mapping that will be looked at more closely is displaced subdivision surfaces (Lee et al, 2000). It is used for describing surface of highly detailed models that are generated by 3D scanning devices. This method greatlyreduces memory requirements for the model, as it requires only a relatively simpledomain surface and a scalar displacement map.

Tessellation
Most surfaces can be approximated with a large number of small polygons. This processof converting a 3D object to a collection of polygons is called tessellation.

4 DISPLACED SUBDIVISION SURFACES
As was said in the chapter 3.1, one reason for using displacement mapping is it’scompact representation. In this chapter we will look at a one very efficient way of doingit, displaced subdivision surfaces, which was introduced by Lee et al (2000).The idea of displaced subdivision surfaces is very simple. Basically it represents thedetailed surface as a scalar-valued displacement over a smooth domain surface.Example of displaced subdivision surface is illustrated in Figure 6

4.2 Benefits
Highly detailed polygon models, such as those generated by 3D scanning technologies,can be effectively modeled using only a domain surface and a displacement map.Memory requirements are smaller as a scalar valued displacement map requires vastlyless storage space than a vector valued polygon mesh.Displacement subdivision surfaces are a convenient way for representing animation. It’smore efficient to calculate animation calculations on simple control mesh than on highlycomplex polygon mesh.Surface detail can be easily edited by modifying the scalar displacement value

5 ADAPTIVE VIEW DEPENDENT TESSELLATION
Previous chapter presented a compact way to model detailed surfaces that adapts to thesurface complexity. In this chapter another approach, adaptive view dependenttessellation of displacement maps (Doggett et Hirche, 2000), is presented. It is moredynamic and also suitable for hardware rendering.When software based methods are used a large number of triangles are required to present a displacement mapped surface. If hardware rendering pipeline could beextended to handle displacement maps as geometric primitives, it would offer significant savings in memory use. It would also lessen the transfer of vertex data fromthe CPU to the graphics processor, which easily becomes a bottleneck for hardware
"

And well, here is an example of how much you can reduce the memory requirements if you use tesselation with dispalcmeents



I also found that in 3D scanning is very common to use the STL(Standard Tessellation Language) and here are some documents and definitions


Here is an example application of STL
https://www.solidconcepts.com/news-releases/solid-concepts-inc-now-offers-3d-scanning-services/
"
Solid Concepts Inc. Now Offers 3D Scanning Services
Solid Concepts’ 3D scanning technology is known as laser triangulation scanning. Directed light reflects back from the object scanned to record the object as millions of polygonal triangles, or a fine mesh. The scanner records the polygonal triangles to comprise a full three dimensional object, which is then translated into an STL file ready for immediate replication with 3D printing. While many designers model their 3D designs from scratch using special (and often expensive) 3D CAD programs, when replicating a part that is already in existence it is exponentially more helpful to begin with a scanned object to eliminate unnecessary time modeling an object from the ground up.
3D printing begins with an STL file. STL – Standard Tessellation Language – files are the industry standard for rapid prototype or 3D Printing. STL files record and save information from 3D CAD data as hundreds of polygons, or facets, defined by vectors and vertices. These polygons instruct 3D Printers during the additive layer manufacturing process. Solid Concepts provides scanned objects with tolerance integrity ready for replication.
"

http://www.x2metrology.com/2014/09/20/science-3d-scanning-quickly-get-buy-offs/
"
The Science of 3D Scanning, How to Quickly get Buy Offs.
September 20, 2014/0 Comments/in 3D Scanning /byinfo@x2metrology.comThe Science of 3D Scanning using a optical 3D scanner. Optical 3D Scanning is based on triangulation, it projects a fringe pattern (structured light) with two cameras on your part just as easy taking a picture from a camera phone. After the scan is complete the software then calculates millions of points within seconds with the final output a clean STL (Standard Tessellation Language).

"

There are some firends that have said to notice the use of Texture Syntesis in the first image that shinen provided, seems pretty noticeable on the terrain rendering; they suggest that at least for the terrain in that image they could have used normal
maps with texture syntesis, of course that using dispalcement maps in texture synthesis is possible, so what do you think?