You do not generate Geometry, you just have a funtion that maps a 3d Coordinate to a distance. This function defines the shape of you geometry.

Using this function you can have a ray march through space by checking the distance function on its current position you know how far you can progress. If you get for example a distance of 10 you that within 10 units of you current position there is a surface.
Then you also know you can advance 10 units without hitting anything.
Then you recheck you distance funtion. If the surface was in the direction of your ray the distance might now get close to 0. If it wasnt it might get bigger.
That way you can trace out very efficiently your geometry based on simple distance functions.

Once you get the hang of it you can do very cool stuff with as seen on the Website Slin posted.
Also you get Shadows and AO pretty much for free.
Maybe counterintuitive Texturing is also not that big of a deal.

EDIT: You can look up some standard distance functions here:
here

The biggest problem I see for the gamestudio is because the whole geometry is kinda procedurally generated on the graphics card you would have to somehow communicate your level in a new way to the graphics card, as you just rely on some basic shapes rather than models based on polygons.
Maybe Slin knows a little bit more about that than I do.

EDIT2:
On distance field shadows

It's a subtle difference in lighting but it's there, you just have to look closely. If you draw a line from the left big circle to the one on the right and divide the torus in half like this you will see that the upper part of this surface is slightly darker than the lower part.


Although I like the shadertoy example by Slin and find the concept interesting I see no reason at all to not go for a simple polygonal approach.

Thank you Puppeteer for the explanation. The site of Iñigo Quilez is also very informative and entertaining at the same time; I read his presentation: Rendering Worlds with Two Triangles with raytracing on the GPU in 4096 bytes, now I think I have understood it.

As Superku said, I think regular meshes are nicer for certain things. I want for example hard edges like in this image:



and I am not 100% sure if I can achieve this easily with distance fields.

Other than that, I see also the possibilities for soft shadow rendering and collision detection. Maybe I will try a distance field for shadow computation; otherwise I would have used raytracing instead. I would then calculate a distance field that matches the polygonal ring. In the pixel shader I would evaluate then the shadow factor.

It looks like you are running the demo on a laptop. Have you ever experienced similar issues with games on your system?

Currently I'm not able to test it on my Desktop-PC. But I will try it as soon as I can.

Any progress?

Yes - but only codewise. I am currently porting the Irrlicht XML reader to Lite-C for reading XML from file/WRS. For that I wrote an own String-lib, too, which costed me also some time, but it works now. Beside that, regular life, work, women and illness took their share, so my progress is not trackable with screenshots atm

Here are the latest screenshots:

Very nice. Can you pick the color directly in a rgb dialog with this "pick" button on the left window?

No, but from the screen directly!