[Freeciv-Dev] Re: (PR#2369) iso-view eye candy

[Jason Dorje Short <vze49r5w@xxxxxxxxxxx>]

Raimar Falke wrote:
> On Thu, Nov 21, 2002 at 07:19:50AM -0500, Jason Dorje Short wrote:
> While automatic blending is nice it can't produce the same results as
> if it was done by hand. Look at
> http://www.civ3.de/www/content/misc/images/terrain-original.jpg. Look
> how the hills around the mine are blended. Look at how the south coast
> is blended. Especially at the coast you can't tell which tile is ocean
> and which is land. This is a good thing IMHO.
This _is_ a good thing.  But I am fairly positive it was not done by 
drawing 2000 different tiles.
> While I don't have any clue about how does the graphic engine of civ3
> works the contents of this modpack
> http://www.civfanatics.net/uploads2/cordy-terrain.zip reveals that
> they also work with a system which is similar to our old one. Leaving
> the extra height out of this the tiles of Mountains-snow.pcx are very
> similar to our 16 tiles. In the x* files they blend three terrain
> types. And yes these are a lot of tiles.
>
> I think that the results show that they got this right. While I'm not
> for coping their graphics/approach 1:1 I want to know why we don't do
> the old way in the iso case. And the only answer I have is: because it
> requires less graphics and so it was a solution which brought results
> (i.e. a working iso-mode) faster.
The 'new' method is the civ2 method.  I assume when the iso code was 
written, they wanted to use the civ2 method since it was clearly 
superior to the current method (which I will tentatively call the 'civ1' 
method).  From doc/HACKING:
[The current tile system should be converted to something like civ2's.
-They get away with drawing way less tiles, and it looks better. -PU]

That the civ3 method is better than either really doesn't affect this.

Note that the civ2 method does have multiple tiles for hills, mountains, 
and forests.  And that ocean tiles are drawn in 9 parts to accomodate 
all the different coastlines.
What I would like to see in the long run is a separation from 
civ1==non-iso and civ2==iso, and a modularization of the tileset 
systems.  Then we can think about introducing a 'civ3' method, if we can 
figure out how their drawing was done.
But as I said, this is a long-term goal.

jason