Are shadows really this slow in jME? When I stand next to trees in the rpg toolkit demo the FPS plummets, and when I turn them off the fps goes up to 60.
If I need a tile grid (not plotting of individual tiles) can jME:s heightmaps be used? E.g. is it easy to find which tile was clicked and where to place a model at specific x,y tile?
Can jME:s procedural texturing be used to accomplish layers? E.g. a dirt path on a grass slope?
2) Are shadows really this slow in jME? When I stand next to trees in the rpg toolkit demo the FPS plummets, and when I turn them off the fps goes up to 60.
3) If I need a tile grid (not plotting of individual tiles) can jME:s heightmaps be used? E.g. is it easy to find which tile was clicked and where to place a model at specific x,y tile?
Shadows CAN be VERY slow in jME, although there is hope in jME 2.0. You may have to do some serious limiting on what gets shadowed and how.
jME support both object picking and triangle picking.
But probably better just to create your own tile structure (maybe extend node) then each tile could be a node and be added to an collection of nodes.
That would get you the ability to cast a selected tile to your Node structure and manipulate it anyway you want ;)
Ok, won't tiles as individual nodes be slower than a generated terrain? Creating height maps with individual tile nodes, also seems a bit tricky. Both in regards to their intersections, but also the seams of the terrain textures and possible stretching that could occur?
A node can be part of a subset of nodes. Basically, each tile is a Node that can contain it own textures, and then added to anouther node (call it the parent). A single texture could be used on the parent and the subnodes could each have a seperate texture state to apply a multi-texture.
I would have to wipe up a quick proof of concept, but I believe that would work.
As far as efficiency goes that would have to be tested, since I don't have the answer. I would think they would be comparable though (just a logic slow down).
One HUGE benefit would be utilizing the power of the jME node though.
madlion said: 1) Do you contact the author? Site seems to be down today. Perhaps he so isn't working on this anymore.
The link works for me.
4) Do you know http://wwwhomes.uni-bielefeld.de/krabien/cgi/output.py?JMESplatting or http://www.jmonkeyengine.com/wiki/doku.php?id=hardware_texture_splatting ?
Thanks, now I know.
3) So what do you mean? Instead of using single tile meshes using a terrain with heighmap?
Ok... so can't you simply calculate the tile from a specific 3d point!?
You have x,y,z ignore y (up vector) and you know the size (width, height) of your tile... just as you have a 2d map. Perhaps you have to take the terrain center point into count.
