CODEWITHSUNDEEP
openglverticesuv-mappingbuffer-objects
user11494366
user11494366

Reputation:

OpenGL 4 - UV Coordinates for Triangle Strip Cube

I have a cube made with a triangle strip, and I am trying to find the UV coordinates for it.

vert = new VBO<Vector3>(new Vector3[] {
            new Vector3(1, 1, 1), 
            new Vector3(0, 1, 1), 
            new Vector3(1, 1, 0),
            new Vector3(0, 1, 0),
            new Vector3(1, 0, 1),
            new Vector3(0, 0, 1),
            new Vector3(0, 0, 0),
            new Vector3(1, 0, 0)
       });
ind = new VBO<uint>(new uint[] { 3, 2, 6, 7, 4, 2, 0, 3, 1, 6, 5, 4, 1, 0 }, BufferTarget.ElementArrayBuffer);

Does anyone know what they would be?

Upvotes: 0

Views: 1126

Answers (1)

Texedo
Texedo

Reputation: 3

Short Answer: You can assign any value to the UV coordinates, even if they overlap ( albeit, this isn't usually desirable ). So long as you create a UV coordinate for every vertex coordinate. If you're Ok with overlaps, you could just declare 8 Vector2(s) as your UV coordinates and assign them with any value between -1 and 1.

Long answer:

This all depends on the way you index your coordinates.

UV coordinates tell you how to map a 2D polygonal region of a 2D texture to your 3D model geometry. There should be a UV coordinate for every vertex coordinate, if your UV(s) and vertices use the same indices ( which doesn't seem optimal for your vertex coordinates as they are ).

Indices designate which of your coordinates ( 3 indices for triangles, 4 for squares ) correlate to a 2D (texture) or 3D (model) polygon. The way your vertex coordinates are defined, unless you duplicated every vertex in a way that every 3 vertices defines a triangle, you'd have to use indexing to indicate which of your 8 vertices is a polygon. For example, the indices { 0, 1, and 3 } indicate the top-right-rear ( rear here, meaning further on the positive Z axis ) triangle on top of your cube.

An issue comes with using the same index array for your vertices and UV(s). If you indexed your model as is, your model faces wouldn't have any problems but some of your UV faces would overlap with other previously defined UV faces. This is because the vertices of some faces will be shared with other vertices on the the other side of your texture space. Think about your cube as if it were a cross and to put it together, you would wrap the base back around to the top. You can't do that if your cube's geometry only exists in 2 dimensions ( as it would in your UV coordinates ).

The seemingly best solution in this case would be to use cube projection, which I don't know how to do yet. So, I'll recommend what I understand is the next best solution:

Duplicate any vertices that would cause the UV faces to wrap over one another ( the base of the cross ) and optionally vertices that would cause too much distortion in the way the texture would be applied to the vertex coordinates; the 2 outer vertices of the head, "hip"(?), and arms of the cross would be further spaced out, requiring distortion in the texture to produce the desired outputs.

Doing so should result in you having 10 vertex coordinates, 10 UV coordinates, and 36 indices, where every 3 indices defines a triangle ( 12 triangles ).

Keep in mind, there multiple ways of achieving what you're asking so deeper research is recommended.

A visual representation of the previously described coordinate and indexing alignment. ( Fixed the Z axis )

This represents duplication of the vertex and UV coordinates at index 0 and 1 to indices 8 and 9. Vertex coordinates 8 and 9 hold the same 3D location value as vertex 0 and 1, whereas the UV coordinates 8 and 9 are located lower on the Y axis than coordinates 6 and 7.

I forgot to put this in the example image but the indices in the example would be:

int indices[] = {
    0, 1, 2,
    1, 2, 3,
    2, 3, 4,
    3, 4, 5,
    0, 2, 4,
    0, 4, 6,
    1, 3, 5,
    1, 5, 7,
    4, 5, 6,
    5, 6, 7,
    6, 7, 8,
    7, 8, 9
}

This will give you 12 modelspace triangles and 12 UV triangles, where every 3 indices is a triangle.

EDIT: As per the link provided by @Rabbid76, 14 vertex and UV coordinates would be better as you wouldn't get the distortion. The way I mentioned is just another way of doing it that has its ups and downs( more distortion, slightly less memory usage ).

Upvotes: 0

Related Questions