This function permits you to use a custom data type for specific vertex format attributes as part of the new vertex format being created. The available values to use are defined by the data type constant that you choose, listed below:
Constant Description vertex_type_float1 A single floating point value vertex_type_float2 Two floating point values vertex_type_float3 Three floating point values vertex_type_float4 Four floating point values vertex_type_colour Four component values (r, g, b, a) vertex_type_ubyte4 Four component unsigned byte values (from 0 to 255)
The use that these constants will be put too also needs to be defined so that the values can be "bound" properly within the shader being created. This is necessary due to the fact that DX and OpenGL have different requirements so if you don't bind them properly, they won't come through right in the shader. The available usage constants that you can choose are listed below and those you use will depend on the specifics of the shader being created:
Constant Description vertex_usage_position position values (x, y, z) vertex_usage_colour colour values (r, g, b, a) vertex_usage_normal vertex normal values (nx, ny, nz) vertex_usage_textcoord UV coordinates (u, v) vertex_usage_blendweight the blendweight of the input matrix (for skeletal animation, for example) vertex_usage_blendindices the indices of the matrices to use (for skeletal animation, for example) vertex_usage_depth vertex depth buffer value vertex_usage_tangent tangent values vertex_usage_binormal binormal values vertex_usage_fog fog values vertex_usage_sample sampler index
There are some important things to note when using custom formats like these:
- The vertex_format_add_custom() function only supports vertex_usage_position, vertex_usage_colour, vertex_usage_normal and vertex_usage_textcoord when using GLSL shaders. These will map to the shader attributes in_Position, in_Colour[0 - ...], in_Normal respectively (anything that is not one of these three attributes - eg: texture coordinates - can be mapped to any attribute you define).
- In general you should use vertex_usage_textcoord for all extra parameters where possible, as types like vertex_usage_blendweight and vertex_usage_tangent are close to deprecated in most shader languages, and probably won't convert properly. Instead use vec, vec2, vec3 or vec4 types vertex_usage_textcoord and everything should work fine.
- GLSL ES does not support integer attributes, so passing in ivec4's does not work (this type is usually used when passing in vertex_usage_blendindices). What you need to do is pass in texture coordinates and then in the shader, convert them to ivec4 like this:
attribute vec3 in_Position;
attribute vec4 in_BlendIndices;
attribute vec4 in_BlendWeights;
varying vec4 v_vColour;
varying mat4 v_mat;
gl_Position = gm_Matrices[MATRIX_WORLD_VIEW_PROJECTION] * vec4( in_Position.xyz, 1.0);
v_vColour = in_BlendWeights;
ivec4 t = ivec4(in_BlendIndices);
v_mat = gm_Matrices[ t.x ];
Blend weights are usually stored in an array and then accessed using blend indices, but you can see here that instead of defining in_BlendIndices as an ivec4 attribute, it's a vec4, then cast to an ivec4 in the code. This is then used index the array created using the gm_Matrix (you can only access an array using an INT value - not a float).
|type||The data type that this custom vertex data will hold (see the type constants listed below).|
|usage||The use that the data will get(see the usage constants listed below).|
my_format = vertex_format_end();
The above code will create a new vertex format with just texture and 3 custom floating point values for position. It is then stores the format id in the variable "my_format".