With regard to the generations..
If you imagine a car maker making a model of Car, like (to use a UK example) the Ford Escort. Ford started off with a Mark 1. The model remains basically the same, but there are variations of engine, seat covers and paintwork. Then when Ford had gone as far as they could with the first design (which wasn't very good as they were learning new things with the new model), they started again making the Mark 2, then Mark 3 etc.
With the figures, the software manufacturer made an initial generation, Generation 1. It really wasn't very flexible at the joints, had limited detail and limited options to customise the figure into unique characters. To improve this, they went to generation 2 which was better, but as it was early in the technology cycle, it still wasn't terribly good. Then Generation 3 came out. It was more customisable, so it was possible to morph the mesh (the collection of little flat surfaces [facets] used to define the model) into different shapes to get different characters. The two figures in the image above use different modifications of the same Generation 3 mesh. The mesh had a lot of facets to obtain the detail (around 72,000 in the characters above), but the bending at the elbows, hip & knee wasn't terribly human in the way it happened (with the elbow and buttocks bulging outwards at large bends and the knee looking like it folded along a crease line. But it was a good start.
When Generation 4 came out things improved and the number of facets increased to 80,000+. However at that number of facets, the computers of the time started to struggle with memory, so a less detailed mesh was created for the 5th generation, but more thought put into the way the joints happened, and the ease of creating different characters.
These trends of fewer facets, more customisation and better joint bends continued to the most recent generation, generation 8, where it's rare to recognise a character as there are thousands. The number of facets in the mesh in the latest generation has reduced to 17000, but additional information can be called upon to contour the surface when extra close up detail is needed, and each facet can be broken up into a maximum of 256 individual mini-facets. This way the 17000 facet model can render with the level of detail that a 4.7million facet model can, but not overload the PC at other times.
With regards to printing.
I'm not sure which type of printing you mean.
The output of a render is a 2 dimensional image that can be printed on a piece of paper just like every other picture. There is a software limit of 10,000 pixels square, so at full Laser printer resolution it could create an image 16.5" square at ultimate printer resolution, or more reasonably at poster quality, about 6ft square.
As for the other type of printing, 3D printing. Yes, the characters can be printed. This is a trial 3D print about 95mm high (3.75" high) of the character on the left above when wearing a bikini as opposed to a dress. I'm sure it could be improved with more experience, a less transparent resin and more trials. I have not gone much further with 3D printing because I am allergic to that resin, even when it is in the solid state - makes my skin sting after touching it and the effect lasts for 24 hours.
Regards,
Richard