PDA

View Full Version : GLSL Strides



Septimra
07-09-2015, 06:59 PM
I know that all information given is in vec4 so I suppose that has something to do with what I am experiencing.

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216

This my array converted to bytes passed into a uniform buffer

uniform Skinning{
//vertex offset into the weights array
int start_position[216];
};

however it seems every component in start_position is 4 off
start_position[0] = 0
start_position[0] = 4
start_position[0] = 8

I now understand that its taking it as an ivec4, and my values are packed away in the x,y,z,w components.
So the new question is how to get around losing space like this in an array in glsl? Is there a different layout to use?

I tried std140, packed, shared just to see--but no effect.


EDIT. This is good. I think I might've pretty much solved this one, but any insight would be appreciated.

Alfonse Reinheart
07-09-2015, 08:04 PM
OK, your skinning matrices should be uniforms. But the weights (and the indices for the matrices) should be per-vertex attributes.

The standard way this is done is to have a (u)vec4 for the indices and a vec4 for the weights; the XYZW components represent up to 4 bones that are weighted to that vertex. If a vertex in your mesh is affected by more than 4 bones, then you just use only the 4 bones with the highest weights (recomputing the weights so that they add up to 1). If a vertex is affected by fewer than 4 bones, then the extra slots in the vec4 have a weight of 0.

You use the index input vector to fetch the matrix from the uniform array, and you use the weight input vector to compute the weight for that bone.

As for the specifics of your question:


I now understand that its taking it as an ivec4, and my values are packed away in the x,y,z,w components.
So the new question is how to get around losing space like this in an array in glsl? Is there a different layout to use?

There's no way to guarantee that.

The only layout for UBOs that has a well-defined arrangement is `std140`. And with that layout, the smallest possible array stride (the number of bytes from one element to the next) is the size of a vec4. Implementations could offer more forgiving layouts, but they would be implementation dependent (accessible through `shared` and `packed`).

`std430` doesn't have this restriction, but it is only available for SSBOs.

Basically, you have to live with it. Or work with it, by declaring an array of `ivec4`'s, then accessing the elements like this:



for(int i = 0; i < 216; ++i)
{
int j = i % 4;
unif_array[i / 216][j]; //Access individual element.
}