Mirrored as part of a study of Minetest events of 2010 to 2019 and people involved, and in connection with a related book, events in 2017 to 2018, in particular, conferring upon host legal rights related to Fair Use.

More about the long range terrain approximation test

The main reason I tried this out is that this allows the player to better choose where to go without huge data transfers, huge memory consumption and huge burden on the GPU. I am aiming for the low-end machines, after all.

I think it already gives a lot better feel to the environment, and if I used this, it would allow me to create a lot more varied maps because the player wouldn’t need to randomly choose where to start travelling even if the destination is very far away. And it would make sense to have some really large-scale features. Currently there is no point in having a mountain 400 blocks high, because you couldn’t see it anyway, just a small part of the slope. I love it that it actually makes sense to climb up a hill to see better into the distance when using this.

It currently is completely unoptimized which causes the lower FPS. This is not a problem when it is optimized.

Also, because it currently simply refers to the map generation noise functions, it doesn’t represent what there actually currently exists. The generator does some things that are not possible to determine from the noise functions at all, and small terrain features which are not representable with the small polygon count.

Also, on an old world, the algorithms might have changed a bit. So, you should create a new world for testing this out and getting the right feel to it.

It is possible to make the server calculate simple properties for the surface from what there actually is, and make it send the data to the client instead of the client using the noise functions. In addition to ground height, ground type and plant content, there could for example be flags telling if there is artificial light in some place or a small building or a large building.

The client would be able to light up the torch-illuminated places at night and draw some simple building mock-ups, which would make it easy to spot player-built structures from the distance. You wouldn’t see how they actually look, but you would see that there is something interesting and worth visiting at.

4 Responses to “More about the long range terrain approximation test”

  1. Lurler Says:

    What if someone already changed terrain there? Then the sid has nothing to do with actual terrain.

  2. electronics, etc. » Blog Archive » Long range terrain approximation test Says:

    [...] info in the next post This entry was posted on Saturday, June 18th, 2011 at 15:41 and is filed under Uncategorized. [...]

  3. Holger Schurig Says:

    Do you know the Blog “Sea of Memes”, where the Author described his approaches to draw a minecraft-like landscape with various LOD techniques?

    He’s even using OpenGL, so the code runs on Win/MacOSX/Linux. Unfortunately there only with the proprietary graphics drivers, as he doesn’t use the Mesa-supported OpenGL 2.1, but a higher version.

    The video http://www.youtube.com/watch?v=SQPwqq7FSos is somewhat the culmination of his current efforts. The blog is at http://www.sea-of-memes.com/

  4. George Says:

    This is awesome. Also, are these very tall mountains generated by the current algorithm?

Leave a Reply