## Re: [Algorithms] ellipse inflated geometry

 Re: [Algorithms] ellipse inflated geometry From: Martin Gladnishki - 2002-06-29 12:48:28 ```Oh, sorry. I just got struck by the idea of having oriented ellipsoids as bounding volumes. Has anyone done this before? It seems pretty convenient to me. Cheers, Martin ----- Original Message ----- From: "Jeremiah Zanin" To: Sent: Friday, June 28, 2002 6:32 PM Subject: Re: [Algorithms] ellipse inflated geometry > My point was that if you allow an ellipsoid to rotate and you want the first > time of contact (the ellipsoid makes a continuous volume as it rotates) like > with a swept sphere, it's hard :-) I'm sure there's an algebraic solution > out there, but i've haven't run across it. > > jeremiah > > ----- Original Message ----- > From: "Martin Gladnishki" > To: > Sent: Friday, June 28, 2002 12:11 PM > Subject: Re: [Algorithms] ellipse inflated geometry > > > > Well, as Gil said, if the ellipsoid-to-sphere transform A is precomputed > in > > a matrix, I think the same goes to rotation that is, a reverse-rotation > > matrix B is precomputed and multiplied by A, which gives you the final > > transform C=AB for the oriented ellipsoid. When testing a triangle against > > the ellipsoid, you transform the tri-vertices to sphere space and test > > there. And since in sphere space the center is at the origin your test > > simplifies to x*x+y*y+z*z<=r*r. > > > > Sorry, if any of this is nonsense but I'm new to ellipsoids. Can someone > > bring more light on the subject? > > > > Cheers, > > Martin > > ----- Original Message ----- > > From: "Jeremiah Zanin" > > To: > > Sent: Friday, June 28, 2002 4:23 PM > > Subject: Re: [Algorithms] ellipse inflated geometry > > > > > > > When you say "replace", does that mean your collision geometry is > actually > > > made up of static-sized spheres, capsules, etc? How do you handle > > different > > > sized swept spheres? Do you store multiple versions of the geometry? > > > > > > For sphere to edge checks, we turn the edge into a capsule (in code) > with > > > the radius of the sphere and the swept sphere becomes just a line > segment > > > (line segment to capsule check). Is this what you mean? > > > > > > As pointed out, you can transform your geometry such that the ellipsoid > is > > a > > > unit sphere through scaling. The problem with ellipsoids is when they > > > rotate around an axis other than the major axis...how do you do a swept > > > ellipsoid that's rotating? But if you don't have to worry about that > then > > I > > > think an ellipsoid might be a good choice. I think a capsule would also > > be > > > a good choice and I think would be cheaper since you don't have to > > transform > > > the geometry or the ray, but it has the same problem with rotating. We > > > currently use two spheres for our character, which does 'ok', but I'd > like > > > to try a capsule someday...... > > > > > > jeremiah > > > > > > ----- Original Message ----- > > > From: "Chris Campbell" > > > To: > > > Sent: Thursday, June 27, 2002 11:58 PM > > > Subject: [Algorithms] ellipse inflated geometry > > > > > > > > > > > > > > Hi, > > > > My current swept sphere vs triangle-mesh has a few annoying little > > > hiccups, > > > > so I am about to replace it with the method discussed a while back on > > this > > > > list, namely: > > > > > > > > > We replace vertices/edges/triangles in the > > > > > environment with spheres/cylinders/prisms, and perform a ray test > > > > > against the resulting collection of features. > > > > > > > > I am pretty sure I can get this working no problem, but I was thinking > > it > > > > would be nice if instead of a sphere, I could have an ellipse twice as > > > > 'high' as it is 'wide' to better fit a standing human. The problem is > > then > > > > the ray vs squashed-cylinder intersection. Maybe there is a well known > > way > > > > to mathematically describe a squashed cylinder, but I can't figure it > > out. > > > > Has anyone done this? Do people just use two spheres instead, or > > transform > > > > the ray or something...? > > > > > > > > Actually two spheres seems like perhaps best way since both rays can > > test > > > > against the same inflated triangle, any arbitrary rotation of my human > > is > > > no > > > > problem, hmmm and to crouch him I just bring the upper sphere down > > etc.... > > > > > > > > ChrisC > > > > > > > > > > > > ------------------------------------------------------- > > > > This sf.net email is sponsored by:ThinkGeek > > > > Bringing you mounds of caffeinated joy. > > > > http://thinkgeek.com/sf > > > > _______________________________________________ > > > > GDAlgorithms-list mailing list > > > > GDAlgorithms-list@... > > > > https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list > > > > Archives: > > > > http://sourceforge.net/mailarchive/forum.php?forum_id=6188 > > > > > > > > > > > > ------------------------------------------------------- > > > This sf.net email is sponsored by:ThinkGeek > > > Caffeinated soap. No kidding. > > > http://thinkgeek.com/sf > > > _______________________________________________ > > > GDAlgorithms-list mailing list > > > GDAlgorithms-list@... > > > https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list > > > Archives: > > > http://sourceforge.net/mailarchive/forum.php?forum_id=6188 > > > > > > > > > > > ------------------------------------------------------- > > This sf.net email is sponsored by:ThinkGeek > > Caffeinated soap. No kidding. > > http://thinkgeek.com/sf > > _______________________________________________ > > GDAlgorithms-list mailing list > > GDAlgorithms-list@... > > https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list > > Archives: > > http://sourceforge.net/mailarchive/forum.php?forum_id=6188 > > > > ------------------------------------------------------- > This sf.net email is sponsored by:ThinkGeek > Caffeinated soap. No kidding. > http://thinkgeek.com/sf > _______________________________________________ > GDAlgorithms-list mailing list > GDAlgorithms-list@... > https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list > Archives: > http://sourceforge.net/mailarchive/forum.php?forum_id=6188 > ```

 [Algorithms] ellipse inflated geometry From: Chris Campbell - 2002-06-28 03:51:30 ```Hi, My current swept sphere vs triangle-mesh has a few annoying little hiccups, so I am about to replace it with the method discussed a while back on this list, namely: > We replace vertices/edges/triangles in the > environment with spheres/cylinders/prisms, and perform a ray test > against the resulting collection of features. I am pretty sure I can get this working no problem, but I was thinking it would be nice if instead of a sphere, I could have an ellipse twice as 'high' as it is 'wide' to better fit a standing human. The problem is then the ray vs squashed-cylinder intersection. Maybe there is a well known way to mathematically describe a squashed cylinder, but I can't figure it out. Has anyone done this? Do people just use two spheres instead, or transform the ray or something...? Actually two spheres seems like perhaps best way since both rays can test against the same inflated triangle, any arbitrary rotation of my human is no problem, hmmm and to crouch him I just bring the upper sphere down etc.... ChrisC ```
 Re: [Algorithms] ellipse inflated geometry From: Jeremiah Zanin - 2002-06-28 14:22:55 ```When you say "replace", does that mean your collision geometry is actually made up of static-sized spheres, capsules, etc? How do you handle different sized swept spheres? Do you store multiple versions of the geometry? For sphere to edge checks, we turn the edge into a capsule (in code) with the radius of the sphere and the swept sphere becomes just a line segment (line segment to capsule check). Is this what you mean? As pointed out, you can transform your geometry such that the ellipsoid is a unit sphere through scaling. The problem with ellipsoids is when they rotate around an axis other than the major axis...how do you do a swept ellipsoid that's rotating? But if you don't have to worry about that then I think an ellipsoid might be a good choice. I think a capsule would also be a good choice and I think would be cheaper since you don't have to transform the geometry or the ray, but it has the same problem with rotating. We currently use two spheres for our character, which does 'ok', but I'd like to try a capsule someday...... jeremiah ----- Original Message ----- From: "Chris Campbell" To: Sent: Thursday, June 27, 2002 11:58 PM Subject: [Algorithms] ellipse inflated geometry > > Hi, > My current swept sphere vs triangle-mesh has a few annoying little hiccups, > so I am about to replace it with the method discussed a while back on this > list, namely: > > > We replace vertices/edges/triangles in the > > environment with spheres/cylinders/prisms, and perform a ray test > > against the resulting collection of features. > > I am pretty sure I can get this working no problem, but I was thinking it > would be nice if instead of a sphere, I could have an ellipse twice as > 'high' as it is 'wide' to better fit a standing human. The problem is then > the ray vs squashed-cylinder intersection. Maybe there is a well known way > to mathematically describe a squashed cylinder, but I can't figure it out. > Has anyone done this? Do people just use two spheres instead, or transform > the ray or something...? > > Actually two spheres seems like perhaps best way since both rays can test > against the same inflated triangle, any arbitrary rotation of my human is no > problem, hmmm and to crouch him I just bring the upper sphere down etc.... > > ChrisC > > > ------------------------------------------------------- > This sf.net email is sponsored by:ThinkGeek > Bringing you mounds of caffeinated joy. > http://thinkgeek.com/sf > _______________________________________________ > GDAlgorithms-list mailing list > GDAlgorithms-list@... > https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list > Archives: > http://sourceforge.net/mailarchive/forum.php?forum_id=6188 ```
 Re: [Algorithms] ellipse inflated geometry From: Martin Gladnishki - 2002-06-28 15:12:33 ```Well, as Gil said, if the ellipsoid-to-sphere transform A is precomputed in a matrix, I think the same goes to rotation that is, a reverse-rotation matrix B is precomputed and multiplied by A, which gives you the final transform C=AB for the oriented ellipsoid. When testing a triangle against the ellipsoid, you transform the tri-vertices to sphere space and test there. And since in sphere space the center is at the origin your test simplifies to x*x+y*y+z*z<=r*r. Sorry, if any of this is nonsense but I'm new to ellipsoids. Can someone bring more light on the subject? Cheers, Martin ----- Original Message ----- From: "Jeremiah Zanin" To: Sent: Friday, June 28, 2002 4:23 PM Subject: Re: [Algorithms] ellipse inflated geometry > When you say "replace", does that mean your collision geometry is actually > made up of static-sized spheres, capsules, etc? How do you handle different > sized swept spheres? Do you store multiple versions of the geometry? > > For sphere to edge checks, we turn the edge into a capsule (in code) with > the radius of the sphere and the swept sphere becomes just a line segment > (line segment to capsule check). Is this what you mean? > > As pointed out, you can transform your geometry such that the ellipsoid is a > unit sphere through scaling. The problem with ellipsoids is when they > rotate around an axis other than the major axis...how do you do a swept > ellipsoid that's rotating? But if you don't have to worry about that then I > think an ellipsoid might be a good choice. I think a capsule would also be > a good choice and I think would be cheaper since you don't have to transform > the geometry or the ray, but it has the same problem with rotating. We > currently use two spheres for our character, which does 'ok', but I'd like > to try a capsule someday...... > > jeremiah > > ----- Original Message ----- > From: "Chris Campbell" > To: > Sent: Thursday, June 27, 2002 11:58 PM > Subject: [Algorithms] ellipse inflated geometry > > > > > > Hi, > > My current swept sphere vs triangle-mesh has a few annoying little > hiccups, > > so I am about to replace it with the method discussed a while back on this > > list, namely: > > > > > We replace vertices/edges/triangles in the > > > environment with spheres/cylinders/prisms, and perform a ray test > > > against the resulting collection of features. > > > > I am pretty sure I can get this working no problem, but I was thinking it > > would be nice if instead of a sphere, I could have an ellipse twice as > > 'high' as it is 'wide' to better fit a standing human. The problem is then > > the ray vs squashed-cylinder intersection. Maybe there is a well known way > > to mathematically describe a squashed cylinder, but I can't figure it out. > > Has anyone done this? Do people just use two spheres instead, or transform > > the ray or something...? > > > > Actually two spheres seems like perhaps best way since both rays can test > > against the same inflated triangle, any arbitrary rotation of my human is > no > > problem, hmmm and to crouch him I just bring the upper sphere down etc.... > > > > ChrisC > > > > > > ------------------------------------------------------- > > This sf.net email is sponsored by:ThinkGeek > > Bringing you mounds of caffeinated joy. > > http://thinkgeek.com/sf > > _______________________________________________ > > GDAlgorithms-list mailing list > > GDAlgorithms-list@... > > https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list > > Archives: > > http://sourceforge.net/mailarchive/forum.php?forum_id=6188 > > > > ------------------------------------------------------- > This sf.net email is sponsored by:ThinkGeek > Caffeinated soap. No kidding. > http://thinkgeek.com/sf > _______________________________________________ > GDAlgorithms-list mailing list > GDAlgorithms-list@... > https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list > Archives: > http://sourceforge.net/mailarchive/forum.php?forum_id=6188 > ```
 Re: [Algorithms] ellipse inflated geometry From: Jeremiah Zanin - 2002-06-28 16:31:06 ```My point was that if you allow an ellipsoid to rotate and you want the first time of contact (the ellipsoid makes a continuous volume as it rotates) like with a swept sphere, it's hard :-) I'm sure there's an algebraic solution out there, but i've haven't run across it. jeremiah ----- Original Message ----- From: "Martin Gladnishki" To: Sent: Friday, June 28, 2002 12:11 PM Subject: Re: [Algorithms] ellipse inflated geometry > Well, as Gil said, if the ellipsoid-to-sphere transform A is precomputed in > a matrix, I think the same goes to rotation that is, a reverse-rotation > matrix B is precomputed and multiplied by A, which gives you the final > transform C=AB for the oriented ellipsoid. When testing a triangle against > the ellipsoid, you transform the tri-vertices to sphere space and test > there. And since in sphere space the center is at the origin your test > simplifies to x*x+y*y+z*z<=r*r. > > Sorry, if any of this is nonsense but I'm new to ellipsoids. Can someone > bring more light on the subject? > > Cheers, > Martin > ----- Original Message ----- > From: "Jeremiah Zanin" > To: > Sent: Friday, June 28, 2002 4:23 PM > Subject: Re: [Algorithms] ellipse inflated geometry > > > > When you say "replace", does that mean your collision geometry is actually > > made up of static-sized spheres, capsules, etc? How do you handle > different > > sized swept spheres? Do you store multiple versions of the geometry? > > > > For sphere to edge checks, we turn the edge into a capsule (in code) with > > the radius of the sphere and the swept sphere becomes just a line segment > > (line segment to capsule check). Is this what you mean? > > > > As pointed out, you can transform your geometry such that the ellipsoid is > a > > unit sphere through scaling. The problem with ellipsoids is when they > > rotate around an axis other than the major axis...how do you do a swept > > ellipsoid that's rotating? But if you don't have to worry about that then > I > > think an ellipsoid might be a good choice. I think a capsule would also > be > > a good choice and I think would be cheaper since you don't have to > transform > > the geometry or the ray, but it has the same problem with rotating. We > > currently use two spheres for our character, which does 'ok', but I'd like > > to try a capsule someday...... > > > > jeremiah > > > > ----- Original Message ----- > > From: "Chris Campbell" > > To: > > Sent: Thursday, June 27, 2002 11:58 PM > > Subject: [Algorithms] ellipse inflated geometry > > > > > > > > > > Hi, > > > My current swept sphere vs triangle-mesh has a few annoying little > > hiccups, > > > so I am about to replace it with the method discussed a while back on > this > > > list, namely: > > > > > > > We replace vertices/edges/triangles in the > > > > environment with spheres/cylinders/prisms, and perform a ray test > > > > against the resulting collection of features. > > > > > > I am pretty sure I can get this working no problem, but I was thinking > it > > > would be nice if instead of a sphere, I could have an ellipse twice as > > > 'high' as it is 'wide' to better fit a standing human. The problem is > then > > > the ray vs squashed-cylinder intersection. Maybe there is a well known > way > > > to mathematically describe a squashed cylinder, but I can't figure it > out. > > > Has anyone done this? Do people just use two spheres instead, or > transform > > > the ray or something...? > > > > > > Actually two spheres seems like perhaps best way since both rays can > test > > > against the same inflated triangle, any arbitrary rotation of my human > is > > no > > > problem, hmmm and to crouch him I just bring the upper sphere down > etc.... > > > > > > ChrisC > > > > > > > > > ------------------------------------------------------- > > > This sf.net email is sponsored by:ThinkGeek > > > Bringing you mounds of caffeinated joy. > > > http://thinkgeek.com/sf > > > _______________________________________________ > > > GDAlgorithms-list mailing list > > > GDAlgorithms-list@... > > > https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list > > > Archives: > > > http://sourceforge.net/mailarchive/forum.php?forum_id=6188 > > > > > > > > ------------------------------------------------------- > > This sf.net email is sponsored by:ThinkGeek > > Caffeinated soap. No kidding. > > http://thinkgeek.com/sf > > _______________________________________________ > > GDAlgorithms-list mailing list > > GDAlgorithms-list@... > > https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list > > Archives: > > http://sourceforge.net/mailarchive/forum.php?forum_id=6188 > > > > > > ------------------------------------------------------- > This sf.net email is sponsored by:ThinkGeek > Caffeinated soap. No kidding. > http://thinkgeek.com/sf > _______________________________________________ > GDAlgorithms-list mailing list > GDAlgorithms-list@... > https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list > Archives: > http://sourceforge.net/mailarchive/forum.php?forum_id=6188 ```
 Re: [Algorithms] ellipse inflated geometry From: Martin Gladnishki - 2002-06-29 12:48:28 ```Oh, sorry. I just got struck by the idea of having oriented ellipsoids as bounding volumes. Has anyone done this before? It seems pretty convenient to me. Cheers, Martin ----- Original Message ----- From: "Jeremiah Zanin" To: Sent: Friday, June 28, 2002 6:32 PM Subject: Re: [Algorithms] ellipse inflated geometry > My point was that if you allow an ellipsoid to rotate and you want the first > time of contact (the ellipsoid makes a continuous volume as it rotates) like > with a swept sphere, it's hard :-) I'm sure there's an algebraic solution > out there, but i've haven't run across it. > > jeremiah > > ----- Original Message ----- > From: "Martin Gladnishki" > To: > Sent: Friday, June 28, 2002 12:11 PM > Subject: Re: [Algorithms] ellipse inflated geometry > > > > Well, as Gil said, if the ellipsoid-to-sphere transform A is precomputed > in > > a matrix, I think the same goes to rotation that is, a reverse-rotation > > matrix B is precomputed and multiplied by A, which gives you the final > > transform C=AB for the oriented ellipsoid. When testing a triangle against > > the ellipsoid, you transform the tri-vertices to sphere space and test > > there. And since in sphere space the center is at the origin your test > > simplifies to x*x+y*y+z*z<=r*r. > > > > Sorry, if any of this is nonsense but I'm new to ellipsoids. Can someone > > bring more light on the subject? > > > > Cheers, > > Martin > > ----- Original Message ----- > > From: "Jeremiah Zanin" > > To: > > Sent: Friday, June 28, 2002 4:23 PM > > Subject: Re: [Algorithms] ellipse inflated geometry > > > > > > > When you say "replace", does that mean your collision geometry is > actually > > > made up of static-sized spheres, capsules, etc? How do you handle > > different > > > sized swept spheres? Do you store multiple versions of the geometry? > > > > > > For sphere to edge checks, we turn the edge into a capsule (in code) > with > > > the radius of the sphere and the swept sphere becomes just a line > segment > > > (line segment to capsule check). Is this what you mean? > > > > > > As pointed out, you can transform your geometry such that the ellipsoid > is > > a > > > unit sphere through scaling. The problem with ellipsoids is when they > > > rotate around an axis other than the major axis...how do you do a swept > > > ellipsoid that's rotating? But if you don't have to worry about that > then > > I > > > think an ellipsoid might be a good choice. I think a capsule would also > > be > > > a good choice and I think would be cheaper since you don't have to > > transform > > > the geometry or the ray, but it has the same problem with rotating. We > > > currently use two spheres for our character, which does 'ok', but I'd > like > > > to try a capsule someday...... > > > > > > jeremiah > > > > > > ----- Original Message ----- > > > From: "Chris Campbell" > > > To: > > > Sent: Thursday, June 27, 2002 11:58 PM > > > Subject: [Algorithms] ellipse inflated geometry > > > > > > > > > > > > > > Hi, > > > > My current swept sphere vs triangle-mesh has a few annoying little > > > hiccups, > > > > so I am about to replace it with the method discussed a while back on > > this > > > > list, namely: > > > > > > > > > We replace vertices/edges/triangles in the > > > > > environment with spheres/cylinders/prisms, and perform a ray test > > > > > against the resulting collection of features. > > > > > > > > I am pretty sure I can get this working no problem, but I was thinking > > it > > > > would be nice if instead of a sphere, I could have an ellipse twice as > > > > 'high' as it is 'wide' to better fit a standing human. The problem is > > then > > > > the ray vs squashed-cylinder intersection. Maybe there is a well known > > way > > > > to mathematically describe a squashed cylinder, but I can't figure it > > out. > > > > Has anyone done this? Do people just use two spheres instead, or > > transform > > > > the ray or something...? > > > > > > > > Actually two spheres seems like perhaps best way since both rays can > > test > > > > against the same inflated triangle, any arbitrary rotation of my human > > is > > > no > > > > problem, hmmm and to crouch him I just bring the upper sphere down > > etc.... > > > > > > > > ChrisC > > > > > > > > > > > > ------------------------------------------------------- > > > > This sf.net email is sponsored by:ThinkGeek > > > > Bringing you mounds of caffeinated joy. > > > > http://thinkgeek.com/sf > > > > _______________________________________________ > > > > GDAlgorithms-list mailing list > > > > GDAlgorithms-list@... > > > > https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list > > > > Archives: > > > > http://sourceforge.net/mailarchive/forum.php?forum_id=6188 > > > > > > > > > > > > ------------------------------------------------------- > > > This sf.net email is sponsored by:ThinkGeek > > > Caffeinated soap. No kidding. > > > http://thinkgeek.com/sf > > > _______________________________________________ > > > GDAlgorithms-list mailing list > > > GDAlgorithms-list@... > > > https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list > > > Archives: > > > http://sourceforge.net/mailarchive/forum.php?forum_id=6188 > > > > > > > > > > > ------------------------------------------------------- > > This sf.net email is sponsored by:ThinkGeek > > Caffeinated soap. No kidding. > > http://thinkgeek.com/sf > > _______________________________________________ > > GDAlgorithms-list mailing list > > GDAlgorithms-list@... > > https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list > > Archives: > > http://sourceforge.net/mailarchive/forum.php?forum_id=6188 > > > > ------------------------------------------------------- > This sf.net email is sponsored by:ThinkGeek > Caffeinated soap. No kidding. > http://thinkgeek.com/sf > _______________________________________________ > GDAlgorithms-list mailing list > GDAlgorithms-list@... > https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list > Archives: > http://sourceforge.net/mailarchive/forum.php?forum_id=6188 > ```
 Re: [Algorithms] ellipse inflated geometry From: Charles Bloom - 2002-06-28 16:48:04 ```The ellipsoid should be aligned with your character's "z" (up) axis, and it should never rotate. IMHO primitives that are rotationally symmetric around the character's primary axis of rotation are the only way to go. You never want to forbid a character from rotating because of collision! At 12:32 PM 6/28/2002 -0400, Jeremiah Zanin wrote: >My point was that if you allow an ellipsoid to rotate and you want the first >time of contact (the ellipsoid makes a continuous volume as it rotates) like >with a swept sphere, it's hard :-) I'm sure there's an algebraic solution >out there, but i've haven't run across it. > >jeremiah ------------------------------------------------------- Charles Bloom cb@... http://www.cbloom.com ```
 Re: [Algorithms] ellipse inflated geometry From: Jeremiah Zanin - 2002-06-28 17:48:25 ```I tend to agree and that's why I said in my original post if you don't care about rotation (I guess meaning it was symmetric) then ellipsoids are good (along with capsules). ----- Original Message ----- From: "Charles Bloom" To: "Jeremiah Zanin" ; Sent: Friday, June 28, 2002 12:47 PM Subject: Re: [Algorithms] ellipse inflated geometry > > The ellipsoid should be aligned with your character's "z" (up) > axis, and it should never rotate. IMHO primitives that are > rotationally symmetric around the character's primary axis of > rotation are the only way to go. You never want to forbid a > character from rotating because of collision! > ```