[SirKnightly]

Here is the nightly for Windows on 14 Jul 2010 - Revision 6292

Group: SSE2
fso-WIN-SSE2-20100714_r6292.7z
MD5Sum

Group: Standard
fso-WIN-Standard-20100714_r6292.7z
MD5Sum

Group: Inferno_SSE
fso-WIN-Inferno_SSE-20100714_r6292.7z
MD5Sum

Group: Inferno_SSE2
fso-WIN-Inferno_SSE2-20100714_r6292.7z
MD5Sum

Code: [Select]

------------------------------------------------------------------------
r6292 | chief1983 | 2010-07-14 10:31:17 -0500 (Wed, 14 Jul 2010) | 1 line
Changed paths:
   M /trunk/fs2_open/code/ai/aicode.cpp

Fix an Assertion usage issue on *nix platforms.
------------------------------------------------------------------------



[chief1983]

In case anyone missed it, the importance of r6192 and r6192 is that it should finally seal up #1888.  So be sure to test autopilot stuff if you've been using it, as well as any of the mentioned AI type changes in the mantis issue.  See the autopilot test build thread for more info on the process it went through during finalization of this commit.

[Woolie Wool]

As far as I can tell, there are no remaining issues except for the autopilot cutscenes with fighters only (no capships) being slower, "shot" from a greater distance away, and less dynamic/WC-like than the original cinematics. Otherwise it works perfectly.

[Iss Mneur]

As far as I can tell, there are no remaining issues except for the autopilot cutscenes with fighters only (no capships) being slower, "shot" from a greater distance away, and less dynamic/WC-like than the original cinematics. Otherwise it works perfectly.

Is this different from before my patching?  Because I have not touched how the autopilot determines the camera positions and the movement speed of the ships.

I will take a look the code tonight, but if you can provide some examples that would be great.  Particularly what do you mean by "less dynamic/WC-like than the original cinematics"?

[Woolie Wool]

Currently the autopilot uses a fairly static camera angle that's a considerable distance from the player. The earlier autopilot used a closer camera angle where the player fighter flew past the camera in a dramatic manner and the camera swung around to follow it, like in Wing Commander: http://www.youtube.com/watch?v=Y-h8uxkxScQ

And yes, the version before your patching, when it worked, used the old camera angle.

EDIT: Curious, the camera system works with a wing of two fighters, but seems to break down with three.

[Iss Mneur]

Currently the autopilot uses a fairly static camera angle that's a considerable distance from the player. The earlier autopilot used a closer camera angle where the player fighter flew past the camera in a dramatic manner and the camera swung around to follow it, like in Wing Commander: http://www.youtube.com/watch?v=Y-h8uxkxScQ

And yes, the version before your patching, when it worked, used the old camera angle.

EDIT: Curious, the camera system works with a wing of two fighters, but seems to break down with three.

As I mentioned on IRC, the only place that I have seen this happen is in the mission that you give me.  I have not been able to reproduce the issue else where.  That being said, if Alpha 1 gets disabled in the mission, it seems the the problem goes away and I get the proper flyby.  Also, it seems like the camera gets further away during every AP run so I think I may have an idea what this issue is caused by.

Will look at it some more tomorrow.

[Woolie Wool]

In this case there's an Apollo flying with two Angels, and the Apollo is slower than the Angel, so maybe it sees ships with unequal speeds and goes into capship mode?

EDIT: The problem apparently comes from ships having different player orders. If they do not have identical player orders, the game uses the capship autopilot cinematics.

[Iss Mneur]

I think I may have fixed it.  Please try this build(debug SSE /w PDBs; SHA1 7098F923374B26A5159F76288E7053E14A797C2D).

[Woolie Wool]

That seems to do the trick. So what was causing it?

[Iss Mneur]

That seems to do the trick. So what was causing it?

It was being caused by the camera position being determined by the players position and not the flight leader like all of the other math was.  So, the further away the player was from the flight leader the further the camera was.

Code: [Select]

Index: code/autopilot/autopilot.cpp
===================================================================
--- code/autopilot/autopilot.cpp (revision 6294)
+++ code/autopilot/autopilot.cpp (working copy)
@@ -729,11 +729,11 @@
 
  tpos = *Navs[CurrentNav].GetPosition();
  // determine distance toward nav at which camera will be
- vm_vec_sub(&pos, &tpos, &Player_obj->pos);
+ vm_vec_sub(&pos, &tpos, &Autopilot_flight_leader->pos);
  vm_vec_normalize(&pos); // pos is now a unit vector in the direction we will be moving the camera
  //norm1 = pos;
  vm_vec_scale(&pos, 5*speed_cap*tc_factor); // pos is now scaled by 5 times the speed (5 seconds ahead)
- vm_vec_add(&pos, &pos, &Player_obj->pos); // pos is now 5*speed cap in front of player ship
+ vm_vec_add(&pos, &pos, &Autopilot_flight_leader->pos); // pos is now 5*speed cap in front of player ship
 
  switch (myrand()%24)
  // 8 different ways of getting perp points
@@ -1027,6 +1027,7 @@
  /* calcuate the speed that everyone is supposed to be going so that there
  is no need for anyone to accelerate or decelerate (most obvious with
  the player's fighter slowing down as it changes the camera pan speed). */
+ Assert( Ai_info[Ships[Autopilot_flight_leader->instance].ai_index].waypoint_speed_cap > 0 );
  vm_vec_scale(&velocity, (float)Ai_info[Ships[Autopilot_flight_leader->instance].ai_index].waypoint_speed_cap);
 
  // Find all ships that are supposed to autopilot with the player and move them