Hello, i'm building a small scale 2d smooth scrolling game..currently i'm working on the map editor and am curious to see if my map format is as good as it could be..currently I have it like this And that's basically my map format as of now..my tiles are all located on a bitmap, and are all numbered I'm not sure if I have too many layers to be able to make a smooth scrolling game...or will that be to much for the computer to handle that fast? My drawing routine maybe a little shaky as well..it goes similar to this..

for(int x=0; x<=12; x++)
{
for(int y=0; y<=12; y++)
{
DrawLayer1(x, y);
DrawLayer2(x, y);
DrawLayer3(x, y);
DrawCharacters(x,y);
DrawLayer4(x, y);
DrawLayer5(x, y);
DrawRoof(x, y);
}
}

so Yeah, I would appreciate it if you guys told me what you think of that, or any enhancements or corrections or ideas you might have..Thanks :-) ArchG

Thank you for your responses...

Quote:
How many tiles do you have in each layer?

Hmm, I'm not sure if I understand this question right..but with the fixed size of 128x128 there would be about 16k tiles in every layer..(if they were filled)

I will have no background bitmap or anything, so everything is made completely in tiles

Quote:
Why do you need 3 background layers?

I guess I just thought that with 3 background layers I could make the maps look very nice...for example in layer 1 (BG1) I could have my grass...then in layer 2 (BG2) I could have some rocks or mushrooms or flowers or something over top of the grass..and then in BG3 I could put the bottom of a tree down..making it look like a tree with some flowers and grass behind it..
3 maybe too many though, i'll most likely take out a BG layer..

Quote:
Does your tiles have transparency?

All layers have transparency except for the ground layer, the very bottom layer..making it so you can stack things and kindof emulate some 3d i guess...

So yea, if that is to many layers and to many tiles and such, I might have to get rid of some of them, I would like to be able to maintain a good 30-35 fps and still have some left over throttle incase I still wanna do some other stuff

I also like the idea of using chars instead of ints..but if the map is smaller than 1 mb I didn't think it would really be a big deal anyway,as it would all be loaded into memory at runtime..but if that becomes an issue (it probably will) I was thinking about using chars or maybe bit fields with my structs

Thanks alot for the input guys

I've just written a 2D top-down tile rendering engine similar to what Endurion suggests, using linked lists.

I started off in a similar fashion to you and had an array that was filled with ground tiles, if I wanted a flower then my bitmap would contain tiles that had the ground and the flower. THis led to loads of tiles and my artistic ability let me down. I then wanted an OVER layer so the characters could walk behind them. However, I didnt want an OVER layer on every tile, meaning that most of my OVER layer was redundant and just taking up space.

Using linked lists completely circumvents redundant tiles and also allows flexibility in the amount of graphical data on each tile. There is an article somewhere in the Resource section that completely describes this technique, here is a brief summary in case you are not familiar:

You start with an array (lets say 128x128, as in your case) but instead of the array holding graphical data it holds data on the properties of the tile (attribute, zone etc.) and it also holds a linked list of all the tile graphics to display on that tile. In a simple form:

Map[0][0] ---> List Header ----> List Node ----> List Node ----> NULL;
Map[0][1] ---> List Header ----> NULL;

In this example, the array elements (at the start) hold property data and a pointer to the List Header. The List Header is nothing more than a list node and there is always a header as there is always a background layer (BG1 in your case). Each node ONLY holds a pointer to a 'tile graphic' structure, which describes how the tile shoudl be drawn to the screen (i.e. texture, size, masking etc.).
Each array element will always have one node (BG1) and can then contain any number of subsequent nodes. Meaning that some tiles can have 10-20 (!) layers whereas others may only have 1 and you are using only as much memory as is required.
In this example, Map[0][0] could contain a grass floor, a flower and a tree, whereas Map[0][1] only contains a grass floor.


If your ability with lists is good then I would definitely adopt this strategy, if your knowledge of lists is shaky then learn and adopt this technique!
I've just moved onto this from where you are now and I guarantee the nuts and bolts are identical, its just that this method slashes your memory usage to almost a bare minimum for your needs.

If you cant find the article then I have it on disk somewhere and could dig it out and send it to you if necessary.

Hope this helps, sorry for being long-winded if this theory is known to you.

The article is tile based techniques by Jason McIntosh