Hi I have a problem with my 3d walls…
I would like to have som tiled textures on my walls but every time i add a texture it seems to stretch the texture instead. I have no idea of how to make this texture tiled. Could it be because I use boxes for my walls??
If it is what should I use instead??
Two ways that I know of:
-
Create a new “tiled” image from the original image, using basic Image stuff (Canvas.drawImage). Then use the new image as your texture. This only works if you are not going to be changing your geometry once your wall has been created.
-
Use the same set of texture coordinates for each “tile”, while traversing over your geometry. This requires you to add new vertices to your wall geometry for each tile, since each texture coordinate must map to a vertex.
The first is probably more likely to give you good results, but uses more texture memory. The second will useless memory, and gives better texture re-use, and is programatically cooler, but might not give you the results you want.
Sorry I missed this question earlier.
Texturing is effected by a number of different classes. The first is the boundary mode: This says whether to clamp the texture, or to wrap it. A wrapping texture will repeat, but a clamped texture will not - it will take the border pixels and stretch just those pixels to the edge of your geometry.
The second influence is the texture coordinates themselves. A texture has a local coordinate system that is independent of the geometry. If you bring the texture up in an image editor, the 0,0 point is the lower left corner. The upper right corner is 1,1 in texture coordinate space. What this means is that for each integer unit in texture coordinate space, you move exactly one width/height of the image over. Turning this around the other way, if you change the texture coordinate on a vertex from 1.0 to 2.0 you will now get two widths of the image there rather than a single width.
The thid influence is the TextureAttributes class and the Transform3D object that can be applied there. The transform is a matrix just like it can be for the vertex coordinates. However, it operates on the texture coordinates, not the vertex coordinates. With it, you can scale, translate, shear etc the texture that is applied to the object, without playing directly with the texture coordinates.
How these effect then appears on the object is dependent on that boundary mode. If I tell it to clamp, then scaling the texture coordinates down to half the size will just shrink the core of the image and leave a lot of room to drag those border pixels out to the edge of the object. If the border mode is in wrap mode, then you’ll see the image tiled across that dimension. The mode can be set differently on each dimension of the texture, so you could have wrap across the horizontal axis, and clamp along the vertical.
Hopefully that will help you sort out the problem.
And one last tip - unless you are going to be seeing both sides of the wall, a box is a bad primitive to use. There’s 6 times too many triangles than what you really need. Once you get up to large models, that can be a severe performance impact.
D’oh! I forgot about WRAP mode! 