#
https://mastodon.social/users/j5v/statuses/116184521520492140
The textures use UV coordinates. There are no problems with them. Also no normal mapping.
Personal Blog about anything - mostly programming, cooking and random thoughts
https://mastodon.social/users/j5v/statuses/116184521520492140
The textures use UV coordinates. There are no problems with them. Also no normal mapping.
#gamedev progress:
The normals of the animated mesh still look wrong. Not yet sure why.
#gamedev progress:
#gamedev progress:
Implemented a procedural version of the book shelves. The geometry is created by simple inset and extrusion operations on quads.
#gamedev progress:
I've spent the last couple of evenings implementing a #GLTF 2.0 parser to support animations. Today I've achieved the first step towards that goal and can now load meshes from .glb files.
#gamedev progress:
I've implemented a new algorithm to create level layouts. The previous approach added on chamber after the other only connecting it at one suitable location. This lead to "tree like" layout.
The new approach creates a Delaunay triangulation where each chambers center is a vertex. Then it discards edges as long as the graph stays connected. The remaining edge are used to as corridors.
small step forward: added shelves to rooms.
Worked a bit on my project which currently acts as a dungeon/level generator. Added the ability to use multiple textures for different parts of the room.
I made a small improvement to the corridor curve sampling to create an equidistant sampling.
let t_samples = 100;
// determine length of path
let path_len = {
let mut l = 0.0;
let mut prev_point = control_a;
for t in 1..=t_samples {
let t = t as f32 / t_samples as f32;
let point = corridor.sample(t);
l += point.pos.dist(&prev_point);
prev_point = point.pos;
}
l
};
// determine spacing based on a desired spacing
// E.g. if we want 1.0 spacing but the curve is 1.1 long,
// a naive approach would result in a 0.1 segment at the end.
// Instead we want a 1.1 segement or a 0.55 segment
let desired_spacing = 1.0;
let spacing = {
let needed_steps = path_len / desired_spacing;
let low = needed_steps.floor().max(1.0);
let high = needed_steps.ceil();
let low_step = path_len / low;
let high_step = path_len / high;
if (low_step - desired_spacing).abs() <= (high_step - desired_spacing).abs() {
low_step
} else {
high_step
}
};
let mut dist = 0.0;
let mut prev_dist_point = control_a;
for t in 1..t_samples {
let t = t as f32 / t_samples as f32;
let point = corridor.sample(t);
dist += point.pos.dist(&prev_dist_point);
prev_dist_point = point.pos;
if dist >= spacing {
// create sample point
}
}
I've added the possibility that new rooms can also be connected to existing corridors.
Additionally added round rooms to the generation. They use my previously implemented Delauney triangulation. I guess there are better ways to generate them but these are just placeholders for the moment.