The fundamental object used to store vertex information in Panda is the GeomVertexData. This stores a list of vertices, organized conceptually as a table, where each row of the table represents a different vertex, and the columns of the table represent the different kinds of per-vertex data that may be associated with each vertex. For instance, the following table defines four vertices, each with its own vertex position, normal vector, color, and texture coordinate pair:

Vertices are always numbered beginning at 0, and continue to the number of rows in the table (minus 1).

Not all GeomVertexData objects will use these same four columns; some will have fewer columns, and some will have more. In fact, all columns, except for "vertex", which stores the vertex position, are optional.

The order of the columns is not meaningful, but the column names are. There are certain column names that are reserved for Panda, and instruct Panda what the meaning of each column is. For instance, the vertex position column is always named "vertex", and the lighting normal column, if it is present, must be named "normal". See GeomVertexFormat for the complete list of reserved column names.

You can define your own custom columns. If there are any columns that have a name that Panda does not recognize, Panda will not do anything special with the column, but it can still send it to the graphics card. Of course, it is then up to you to write a vertex shader that understands what to do with the data in the column.

It is possible to break up a GeomVertexData into more than one array. A GeomVertexArray is table of vertex data that is stored in one contiguous block of memory. Typically, each GeomVertexData consists of just one array; but it is also possible to distribute the data so that some columns are stored in one array, while other columns are stored in another array:

You might want to do this, for instance, if you have certain columns of data that are always the same between different blocks of vertices; you can put those columns in a separate array, and then use the same array within multiple different GeomVertexData objects. There is no limit to the number of different arrays you can have within one GeomVertexData; you can make each column a separate array if you like. (There may be performance implications to consider. Some graphics drivers may work better with one block of contiguous data--one array--while others may prefer many different arrays. This performance difference is likely to be small, however.)