Monday, August 25, 2014


Mass defect

I'va basically found the way to procedurally create mountain ranges interleaved with valleys, thus ensuring some realism while guaranteeing some simplicity in actual map drawing.
Then I felt a sort of compulsion to play Civilization II, it's a couple of days I don't do much else, but I think that tomorrow I'll be back on serious things.

Sunday, August 17, 2014


Billy 7

I turns out I'm working on a procedural terrain creation system, although a mixed one.
What I'm trying to do is to place small pieces of pre-rendered procedurally generated elements in a meaningful way, in order to create the impression of a big, continuous procedurally element created from the ground.
For example working on the placement of forests I created a fractal, without even realizing it until I saw the  results.
I'll now work on mountains, which pose some difficulty since you must provision valleys. I thought that the best starting point are mountains passes, which define connection between valleys. Once defined these two elements I'll fill the voids with mountains.
I started thinking to create a complete procedural terrain generator on my own, it always happens in these cases, the phases are always the same: I have an idea --> I aim to the fullest implementation --> I search on the web articles about it --> after reading the first math formula or code syntax I close the browser.
In this case, in particular, I'd need to learn hlsl. But what would be the advantage? I already have a powerful terrain generator which is wm, why should I reinvent the wheel? I'll simply take advantage of wm, while creating more or less algorithms for terrain placement.

Tuesday, August 12, 2014


Cygnet Committee

I was a bit depressed yesterday since giving up with the wm-centric approach really changed everything. I started thinking to drop quest3d altogether, since it has a poor support for landscapes, and finally switching to unity.
Then I had a sort of aha! moment (in my bed, as it usually happens) and I realized how redundant and exaggerated was to have wm build every single inch of terrain, when I simply could provision a hybrid system, a bit tile-based, a bit continuous terrain. In fact I could pre-render portions of terrains and then place them here and there on the map.
The resulting terrain would not be seamless but it will still be totally acceptable, especially because the single tiles will not be square nor hexagonal but will be ellipses, and also, by creating several versions of the terrain types and randomly using them I can create some variety.

Monday, August 11, 2014


I Got You Babe

I'm considering a major change in the overall game architecture: so far the plan was to rely on world machine to achieve beautiful visuals, but I'm seriously thinking to drop it, as it's not scalable at all, in fact I would have to set up a centralized server that should contain hundreds of thousands of images, to be used by clients dynamically. Since I'm not google I can't affort to set up this system, so I must trade the beautiful maps that I could create with wm for a more workable solution. It's not a tragedy, I must be ready to these major changes.
So now everything changes, because with wm my concern was to set up all the small individual tiles that provided the land textures, the heightmaps, and also the terrain code types to the game client. Without these I have to rely to 3d shapes, which can be very quickly referenced between client and server, but I must also provide a decent way to texture those shapes. For sure the maps will be crappier than planned.

Tuesday, August 05, 2014


Cielo irregolarmente nuvoloso al mattino. Dal pomeriggio aumento della nuvolosità sui rilievi con temporali che in serata potranno toccare la pianura.

Pressione al suolo: 1012 hpa
Temperature: Minime in aumento, massime stazionarie.
Venti: In pianura deboli da Ovest, in montagna deboli da Nord.
Zero Termico: A circa 3700 metri.

I'm having a bad time with the map maker system.
The solutions I've found are either quick and easy but imprecise, leading to unacceptable artifacts in the final map, or terribly complicated and definetely not scalable.
The latest solution involves a mechanism by which the user will broadly define the various terrain types, leaving the program to draw all the details. I don't find any other reasonable way to build a map that spans an equivalent amount of 4096 km2.
I'll now spend probably another 3 months in building this tool, which basically creates the shapes which are then referenced by world machine to build the various terrain types.
If it is impossible to provide an easy way to create maps I'll have to skip the map making process and deny the users the pleasure of creating their own scenarios.

This page is powered by Blogger. Isn't yours?