Yeah... Don't get me wrong, this is not a "I wanna make a fullblown rpg adventure with minimum work using an engine someone else has created for me..." thread... :) I'm simply wondering what approach you would think is the best, and of course, if it's doable: 1) Massive tilearray for each scene, every tile is draw in z-order, or top to bottom, if u wish. pro's: lots of variation, random encounters are easy as hell... :) no big images, except tilesets etc... con's: map making will be a pain... extreme or custom detail cant be achieved easily.... (?) my preferred way: 2) Artist creates a fullresolution image (one or several 4096 sized textures) of the scene, creates a low poly model (no backsides on models etc) that will be used as an occlusion object. this allows extreme detail and areas can be custom tailored with all the bells and whistles that are needed! :) pro's: extreme detail! it's 3D, so all three dimensional benefits are there... con's: high memory usage, not easily moddable or "randomized"... massive array still needed for pathfinding, movement etc... What do you think? I'd go for number two unless someone posts reasons why not...? ;) questions: is there a way to disable the color buffer writes? i.e draw only the zbuffer with background, then slap the rendered texture on top, and then characters+effects...?

Hehe.. ok. :D

but what about this...

Quote:Original post by Rasmadrak
is there a way to disable the color buffer writes?
i.e draw only the zbuffer with background, then slap the rendered texture on top, and then characters+effects...?

That is pretty much crucial to number two... ;)

Ahh... simple as that, eh? :D

Thanks! Off to create wonderous worlds then!!


Cheers!

Quote: con's: not easily moddable or "randomized"...

how not so?
certainly easier than numA
you are gonna draw the tiles as squres arent u and not diamonds?

the thing with option two was that there wouldn't be any tiles... ;)
simply one big texture, possibly subdivided or something...

But I would probably need to mix a little A with B, since tree's and other foliage objects are quite complex and a lot of polygons would be wasted on these if I would use a 3D occlusion object/"map" ...

so I guess it would be:

1) Fill zbuffer

2) Slap map texture on top.

3) Draw characters

4) Draw foliage and other random things

5) Draw Effects


OR

If one could use the depth data created while rendering, that could be stored as the alpha channel in the map texture. that way one would have 100% per pixel control on occlusion... but I don't think that the depthdata would fit in a 256 "color" channel...

however, if one would "toonshade" the rendered depthbuffer, maybe then it would fit, but I guess the characters would be popping through walls etc.. :/