Add some comments explaining issues/todo's with the implementation

[xonotic/xonotic-data.pk3dir.git] / qcsrc / client / particles.qc

diff --git a/qcsrc/client/particles.qc b/qcsrc/client/particles.qc
index 077a76acb27667c5d7f35ea4335a4977ff390de5..e4002d2edd178a391b961811b7d2897446ad56ef 100644 (file)
--- a/qcsrc/client/particles.qc
+++ b/qcsrc/client/particles.qc
@@ -489,6 +489,30 @@ void Draw_ArcBeam()
        float i;
        for(i = 1; i <= segments; ++i)
        {
+               // WEAPONTODO (server and client):
+               // Segment blend and distance should probably really be calculated in a better way,
+               // however I am not sure how to do it properly. There are a few things I have tried,
+               // but most of them do not work properly due to my lack of understanding regarding
+               // the mathematics behind them.
+
+               // Ideally, we should calculate the positions along a perfect curve
+               // between wantdir and self.beam_dir with an option for depth of arc
+
+               // Another issue is that (on the client code) we must separate the
+               // curve into multiple rendered curves when handling warpzones.
+               
+               // I can handle this by detecting it for each segment, however that
+               // is a fairly inefficient method in comparison to having a curved line
+               // drawing function similar to Draw_CylindricLine that accepts
+               // top and bottom origins as input, this way there would be no
+               // overlapping edges when connecting the curved pieces.
+
+               // WEAPONTODO (client):
+               // In order to do nice fading and pointing on the starting segment, we must always
+               // have that drawn as a separate triangle... However, that is difficult to do when
+               // keeping in mind the above problems and also optimizing the amount of segments
+               // drawn on screen at any given time. (Automatic beam quality scaling, essentially)
+
                // calculate this on every segment to ensure that we always reach the full length of the attack
                float segmentblend = bound(0, (i/segments) + self.beam_tightness, 1);
                float segmentdist = vlen(beam_endpos_estimate - last_origin) * (i/segments);