map_heights.tga understood

**alpaca** · November 19, 2006, 03:33 PM

Well for those of you that don't read the org regularly: I developed a new theory about map_heights.tga.
You can find it here:
http://forums.totalwar.org/vb/showth...54#post1310754

**Muizer** · November 19, 2006, 07:50 PM

Originally Posted by alpaca

Well for those of you that don't read the org regularly: I developed a new theory about map_heights.tga.
You can find it here:
http://forums.totalwar.org/vb/showth...54#post1310754

Eh, isn't that a rather complicated way to explain something fairly simple? As far as I know, to overlay a region size map on a ground type size map, you need to double its scale and shift it by half a ground type map pixel. I say "ground type" rather than height, because ground type pixels are tiles, whereas the height values are not (they are post heights of a triangulated network).

Now as for the "inaccessible settlement" problem, you'll have to explain what you mean by this. Is it a battlemap phenomenon? I suppose it must be. I'm not familiar with slope related access problems on the campaign map. Those are invariably ground type or feature problems.

**alpaca** · November 20, 2006, 07:26 AM

Originally Posted by Muizer

Eh, isn't that a rather complicated way to explain something fairly simple? As far as I know, to overlay a region size map on a ground type size map, you need to double its scale and shift it by half a ground type map pixel. I say "ground type" rather than height, because ground type pixels are tiles, whereas the height values are not (they are post heights of a triangulated network).

Now as for the "inaccessible settlement" problem, you'll have to explain what you mean by this. Is it a battlemap phenomenon? I suppose it must be. I'm not familiar with slope related access problems on the campaign map. Those are invariably ground type or feature problems.

Yes, that's pretty much what I said. But ground type pixels aren't tiles, I think each tile has 9 ground type pixels, same as on the height map, although, again, most are shared with neighboring tiles.
Remember, a tile is in my opinion a pixel on the region map.
And maybe my explanation is a bit more complicated, but it's exact and understandable (or at least I think it's understandable).

About the inaccessible settlements: Maybe you're right and the problem isn't the height map but the ground types map. Do the mountain ground types make a settlement inaccessible?

**Spurius** · November 20, 2006, 08:56 AM

Originally Posted by Muizer

I'm not familiar with slope related access problems on the campaign map. Those are invariably ground type or feature problems.

We're currently testing it further, but from earlier tests I know that if you 'up' the brightness of the pixels in heightsmap - land only, of course, this will go well only up to a certain point, and then the errors with 'unreachable settlements' etc will start pouring in - not changing ANYTHING in groundtypes at that time.

So we are certainly not discounting this info, but so far have not been able to confirm it, either.

Should become clear very shortly anyway, though. We're trying just about every imaginable combination here now.

**Muizer** · November 20, 2006, 10:13 AM

Hm, well it is true the coordinate system seems tied to the Region map pixel size, but ground type pixels interact on an "individual basis" with height posts and they exist independently on the battlemap. With three different overlaying patterns (height posts, region planes and ground type planes) I wonder if there's such a thing as a "tile".

High mountain and dense forest are highly impassable. You need a path of two passable ground type pixels wide through such terrain. Low mountains are slightly less critical. Of the two pixel wide path only half need be replaced with passable groundtypes (in a chessboard pattern).
Swamps are also at least partially impassable as are oceans of course.

About the height effect on access, it is new for me. I can see reasons why a warning should be given, but they are more to do with what happens on the battlefield and in general I find the game doesn't give any warnings of imminent battlemap weirdness. Is this ground type sensitive as well? And what about movement allowance?

**Spurius** · November 20, 2006, 11:31 AM

Originally Posted by Muizer

High mountain and dense forest are highly impassable. You need a path of two passable ground type pixels wide through such terrain. And what about movement allowance?

Dense forest, I believe - but already saw one exception to that high mountain rule - you can encircle Jeddah for instance with a 3 pixel wide mountain range, all high mountains and give it roads - these will not appear though, but no error either - so coasts (and islands) probably have slightly different rules already.

Although this could also just be a bug, of course.

**wilddog** · November 21, 2006, 01:51 PM

I haven't seen any slope related problems. Normally its just the ground types are causing the problem.

Alpaca I need to read through your post to see if you can work out the region size from the heights as that's the way the 'regions' map is generated. Got to admit I really liked having that corrected_regions_map created. I think trying to position some of the ports is going to be more time consuming.

**Spurius** · November 21, 2006, 02:23 PM

Coincidence?

http://forums.totalwar.org/vb/showth...68#post1313668

**alpaca** · November 21, 2006, 03:54 PM

Originally Posted by wilddog

I haven't seen any slope related problems. Normally its just the ground types are causing the problem.

Alpaca I need to read through your post to see if you can work out the region size from the heights as that's the way the 'regions' map is generated. Got to admit I really liked having that corrected_regions_map created. I think trying to position some of the ports is going to be more time consuming.

Since somebody from the .org showed me some quite old research data he gathered, I think it should be safe to assume that any tile that has a land center height pixel is land and therefore a safe location for harbors. However it might be that some pixels below it are water because harbors have a land and a water component. Will test that when I have time.

**wilddog** · November 21, 2006, 04:21 PM

Spurious

I looked at your link to the article by Spur (presumably you?). I'm not really following what you are saying or trying to show. Essentially a reduction of the heights map doesn't map to the regions map - that's the problem (or at least as I'm encountering). You can see this in the old RTW system by creating a heights map - then reduce to create the regions map) and then compare the result to the corrected regions and they don't align. Same is true in M2TW.

EDIT : See below as this was just an issue with me using Paintshop pro.

Alpaca
So a simple chequered board effect test should result in either all Sea or all land which would prove out your theory.

Spurious
I stand corrected.

I'd always created the map_regions by reducing the heights map in an old version of paintshop pro. I just used Photoshop elements and the nearest neighbour option for reducing and the reduced map seems to correspond correctly with the map_regions I need. Thanks for the link as it now means positioning the ports etc should now be straight forward again.

Merged triple post. Please use the edit button next time. - Trajan

**alpaca** · November 22, 2006, 02:42 PM

wilddog you're pretty sceptical eh.
Allright, here's a little experiment for you: Try replacing in any area you like (for example Cornwall) all pixels with odd x AND odd y coordinates with water and sending a character there. It's all inaccessible although there's a lot of land.
Then, invert it and make all these pixels land and the surrounding pixels water. You can access all of the pixels although there's a lot of water.
Convinced now?

**wilddog** · November 22, 2006, 04:22 PM

alpaca
I was a bit over tired yesterday. Spent ages trying to work out my old paintshop pro conversion and trying to see if the heights corresponded. Anyway made a mess of it so started using photoshop elements as per the note above (Only just bought the new elements 5 so haven't had much use out of it yet). Anyway my first problem was resolved and this realigned the map _region into what I wanted. Before that I couldn't see what was happening so thought your idea didn't fit.

Anyway tried just now just using a bit of Exmoor as it's a bit smaller

and you can see blow that you are of course right that tiles don't count as passable.

Should of known better than to question you