網(wǎng)上有很多關(guān)于pos機用日文怎么說,為什么使用四元數(shù)的知識，也有很多人為大家解答關(guān)于pos機用日文怎么說的問題，今天pos機之家(www.rcqwhg.com)為大家整理了關(guān)于這方面的知識，讓我們一起來看下吧!

本文目錄一覽：

1、pos機用日文怎么說

pos機用日文怎么說

為了回答這個問題，先來看看一般關(guān)于旋轉(zhuǎn)（面向）的描述方法－歐拉描述法。它使用最簡單的x,y,z值來分別表示在x，y，z軸上的旋轉(zhuǎn)角度，其取值為0-360(或者0-2pi），一般使用roll，pitch，yaw來表示這些分量的旋轉(zhuǎn)值。需要注意的是，這里的旋轉(zhuǎn)是針對世界坐標系說的，這意味著第一次的旋轉(zhuǎn)不會影響第二、三次的轉(zhuǎn)軸，簡單的說，三角度系統(tǒng)無法表現(xiàn)任意軸的旋轉(zhuǎn)，只要一開始旋轉(zhuǎn)，物體本身就失去了任意軸的自主性，這也就導致了萬向軸鎖（Gimbal Lock）的問題。還有一種是軸角的描述方法（即我一直以為的四元數(shù)的表示法），這種方法比歐拉描述要好，它避免了Gimbal Lock，它使用一個3維向量表示轉(zhuǎn)軸和一個角度分量表示繞此轉(zhuǎn)軸的旋轉(zhuǎn)角度，即(x,y,z,angle)，一般表示為(x,y,z,w)或者(v,w)。但這種描述法卻不適合插值。那到底什么是Gimbal Lock呢？正如前面所說，因為歐拉描述中針對x,y,z的旋轉(zhuǎn)描述是世界坐標系下的值，所以當任意一軸旋轉(zhuǎn)90°的時候會導致該軸同其他軸重合，此時旋轉(zhuǎn)被重合的軸可能沒有任何效果，這就是Gimbal Lock，這里有個例子演示了Gimbal Lock，點擊這里下載。運行這個例子，使用左右箭頭改變yaw為90°，此時不管是使用上下箭頭還是Insert、Page Up鍵都無法改變Pitch，而都是改變了模型的roll。那么軸、角的描述方法又有什么問題呢？雖然軸、角的描述解決了Gimbal Lock，但這樣的描述方法會導致差值不平滑，差值結(jié)果可能跳躍，歐拉描述同樣有這樣的問題。什么是四元數(shù)四元數(shù)一般定義如下：q=w+xi+yj+zk其中w是實數(shù)，x,y,z是虛數(shù)，其中:i*i=-1j*j=-1k*k=-1也可以表示為：q=[w,v]其中v=(x,y,z)是矢量，w是標量，雖然v是矢量，但不能簡單的理解為3D空間的矢量，它是4維空間中的的矢量，也是非常不容易想像的。四元數(shù)也是可以歸一化的，并且只有單位化的四元數(shù)才用來描述旋轉(zhuǎn)（面向），四元數(shù)的單位化與Vector類似，首先||q|| = Norm(q)=sqrt(w2 + x2 + y2 + z2)因為w2 + x2 + y2 + z2=1所以Normlize(q)=q/Norm(q)=q / sqrt(w2 + x2 + y2 + z2)說了這么多，那么四元數(shù)與旋轉(zhuǎn)到底有什么關(guān)系？我以前一直認為軸、角的描述就是四元數(shù)，如果是那樣其與旋轉(zhuǎn)的關(guān)系也不言而喻，但并不是這么簡單，軸、角描述到四元數(shù)的轉(zhuǎn)化：w = cos(theta/2)x = ax * sin(theta/2)y = ay * sin(theta/2)z = az * sin(theta/2)其中(ax,ay,az)表示軸的矢量，theta表示繞此軸的旋轉(zhuǎn)角度，為什么是這樣？和軸、角描述到底有什么不同？這是因為軸角描述的“四元組”并不是一個空間下的東西，首先（ax,ay,az）是一個3維坐標下的矢量，而theta則是級坐標下的角度，簡單的將他們組合到一起并不能保證他們插值結(jié)果的穩(wěn)定性，因為他們無法歸一化，所以不能保證最終插值后得到的矢量長度（經(jīng)過旋轉(zhuǎn)變換后兩點之間的距離）相等，而四元數(shù)在是在一個統(tǒng)一的4維空間中，方便歸一化來插值，又能方便的得到軸、角這樣用于3D圖像的信息數(shù)據(jù)，所以用四元數(shù)再合適不過了。關(guān)于四元數(shù)的運算法則和推導這里有篇詳細的文章介紹，重要的是一點，類似與Matrix的四元數(shù)的乘法是不可交換的，四元數(shù)的乘法的意義也類似于Matrix的乘法－可以將兩個旋轉(zhuǎn)合并，例如：Q=Q1*Q2表示Q的是先做Q2的旋轉(zhuǎn)，再做Q1的旋轉(zhuǎn)的結(jié)果，而多個四元數(shù)的旋轉(zhuǎn)也是可以合并的，根據(jù)四元數(shù)乘法的定義，可以算出兩個四元數(shù)做一次乘法需要16次乘法和加法，而3x3的矩陣則需要27運算，所以當有多次旋轉(zhuǎn)操作時，使用四元數(shù)可以獲得更高的計算效率。為什么四元數(shù)可以避免Gimbal Lock在歐拉描述中，之所以會產(chǎn)生Gimbal Lock是因為使用的三角度系統(tǒng)是依次、順序變換的，如果在OGL中，代碼可能這樣：glRotatef( angleX, 1, 0, 0)glRotatef( angleY, 0, 1, 0)glRotatef( angleZ, 0, 0, 1)注意：以上代碼是順序執(zhí)行，而使用的又是統(tǒng)一的世界坐標，這樣當首先旋轉(zhuǎn)了Y軸后，Z軸將不再是原來的Z軸，而可能變成X軸，這樣針對Z的變化可能失效。而四元數(shù)描述的旋轉(zhuǎn)代碼可能是這樣：TempQ = From Eula(x,y,z)FinalQ =CameraQ * NewQtheta, ax, ay, az = From (FinalQ)glRotatef(theta, ax, ay, az);其中(ax,ay,az)描述一條任意軸，theta描述了繞此任意軸旋轉(zhuǎn)的角度，而所有的參數(shù)都來自于所有描述旋轉(zhuǎn)的四元數(shù)做乘法之后得到的值，可以看出這樣一次性的旋轉(zhuǎn)不會帶來問題。這里有個例子演示了使用四元數(shù)不會產(chǎn)生Gimbal Lock的問題。關(guān)于插值使用四元數(shù)的原因就是在于它非常適合插值，這是因為他是一個可以規(guī)格化的4維向量，最簡單的插值算法就是線性插值，公式如：q(t)=(1-t)q1+t q2但這個結(jié)果是需要規(guī)格化的，否則q(t)的單位長度會發(fā)生變化，所以q(t)=(1-t)q1+t q2 / || (1-t)q1+t q2 ||如圖: 盡管線性插值很有效，但不能以恒定的速率描述q1到q2之間的曲線，這也是其弊端，我們需要找到一種插值方法使得q1->q(t)之間的夾角θ是線性的，即θ(t)=(1-t)θ1+t*θ2，這樣我們得到了球形線性插值函數(shù)q(t)，如下：q(t)=q1 * sinθ(1-t)/sinθ + q2 * sinθt/sineθ如果使用D3D，可以直接使用D3DXQuaternionSlerp函數(shù)就可以完成這個插值過程。第二篇:四元數(shù)入門 四元數(shù)常?？梢栽?D的書上看到。但我的那本3D圖形學書上，在沒講四元數(shù)是干什么的之前，就列了幾張紙的公式，大概因為自己還在上高中，不知道的太多，看了半天沒看懂。。。終于，在gameres上看到了某強人翻譯的一個“4元數(shù)寶典 ”（原文是日本人寫的。。。），感覺很好，分享下?！镄D(zhuǎn)篇：我將說明使用了四元數(shù)（si yuan shu, quaternion）的旋轉(zhuǎn)的操作步驟（１）四元數(shù)的虛部，實部和寫法所謂四元數(shù)，就是把4個實數(shù)組合起來的東西。4個元素中，一個是實部，其余3個是虛部。比如，叫做Q的四元數(shù)，實部t而虛部是x,y,z構(gòu)成，則像下面這樣寫。Q = (t; x, y, z) 又，使用向量 V=(x,y,z)，Q = (t; V) 也可以這么寫。正規(guī)地用虛數(shù)單位i,j,k的寫法的話，Q = t + xi + yj + zk 也這樣寫，不過，我不大使用（２）四元數(shù)之間的乘法虛數(shù)單位之間的乘法 ii = -1, ij = -ji = k (其他的組合也是循環(huán)地以下同文) 有這么一種規(guī)則。（我總覺得，這就像是向量積（外積），對吧） 用這個規(guī)則一點點地計算很麻煩，所以請用像下面這樣的公式計算。A = (a; U) B = (b; V) AB = (ab - U?V; aV + bU + U×V)不過，“U?V”是內(nèi)積，「U×V」是外積的意思。注意：一般AB<>BA所以乘法的左右要注意?。?）3次元的坐標的四元數(shù)表示如要將某坐標(x,y,z)用四元數(shù)表示，P = (0; x, y, z) 則要這么寫。 另外，即使實部是零以外的值，下文的結(jié)果也一樣。用零的話省事所以我推薦。（４）旋轉(zhuǎn)的四元數(shù)表示以原點為旋轉(zhuǎn)中心，旋轉(zhuǎn)的軸是(α, β, γ)（但 α^2 + β^2 + γ^2 = 1）， （右手系的坐標定義的話，望向向量(α, β, γ)的前進方向反時針地） 轉(zhuǎn)θ角的旋轉(zhuǎn)，用四元數(shù)表示就是，Q = (cos(θ/2); α sin(θ/2), β sin(θ/2), γ sin(θ/2)) R = (cos(θ/2); -α sin(θ/2), -β sin(θ/2), -γ sin(θ/2)) (另外R 叫 Q 的共軛四元數(shù)。） 那么，如要實行旋轉(zhuǎn)，則 R P Q = (0; 答案) 請像這樣三明治式地計算。這個值的虛部就是旋轉(zhuǎn)之后的點的坐標值。 （另外，實部應該為零。請驗算看看） 例子代碼/// Quaternion.cpp /// (C) Toru Nakata, toru-nakata@aist.go.jp /// 2004 Dec 29 #include <math.h> #include <iostream.h> /// Define Data type typedef struct { double t; // real-component double x; // x-component double y; // y-component double z; // z-component } quaternion; //// Bill 注：Kakezan 在日語里是 “乘法”的意思quaternion Kakezan(quaternion left, quaternion right) { quaternion ans; double d1, d2, d3, d4; d1 = left.t * right.t; d2 = -left.x * right.x; d3 = -left.y * right.y; d4 = -left.z * right.z; ans.t = d1+ d2+ d3+ d4; d1 = left.t * right.x; d2 = right.t * left.x; d3 = left.y * right.z; d4 = -left.z * right.y; ans.x = d1+ d2+ d3+ d4; d1 = left.t * right.y; d2 = right.t * left.y; d3 = left.z * right.x; d4 = -left.x * right.z; ans.y = d1+ d2+ d3+ d4; d1 = left.t * right.z; d2 = right.t * left.z; d3 = left.x * right.y; d4 = -left.y * right.x; ans.z = d1+ d2+ d3+ d4; return ans; } //// Make Rotational quaternion quaternion MakeRotationalQuaternion(double radian, double AxisX, double AxisY, double AxisZ) { quaternion ans; double norm; double ccc, sss; ans.t = ans.x = ans.y = ans.z = 0.0; norm = AxisX * AxisX + AxisY * AxisY + AxisZ * AxisZ; if(norm <= 0.0) return ans; norm = 1.0 / sqrt(norm); AxisX *= norm; AxisY *= norm; AxisZ *= norm; ccc = cos(0.5 * radian); sss = sin(0.5 * radian); ans.t = ccc; ans.x = sss * AxisX; ans.y = sss * AxisY; ans.z = sss * AxisZ; return ans; } //// Put XYZ into quaternion quaternion PutXYZToQuaternion(double PosX, double PosY, double PosZ) { quaternion ans; ans.t = 0.0; ans.x = PosX; ans.y = PosY; ans.z = PosZ; return ans; } ///// main int main() { double px, py, pz; double ax, ay, az, th; quaternion ppp, qqq, rrr; cout << "Point Position (x, y, z) " << endl; cout << " x = "; cin >> px; cout << " y = "; cin >> py; cout << " z = "; cin >> pz; ppp = PutXYZToQuaternion(px, py, pz); while(1) { cout << "\Rotation Degree ? (Enter 0 to Quit) " << endl; cout << " angle = "; cin >> th; if(th == 0.0) break; cout << "Rotation Axis Direction ? (x, y, z) " << endl; cout << " x = "; cin >> ax; cout << " y = "; cin >> ay; cout << " z = "; cin >> az; th *= 3.1415926535897932384626433832795 / 180.0; /// Degree -> radian; qqq = MakeRotationalQuaternion(th, ax, ay, az); rrr = MakeRotationalQuaternion(-th, ax, ay, az); ppp = Kakezan(rrr, ppp); ppp = Kakezan(ppp, qqq); cout << "\Anser X = " << ppp.x << "\ Y = " << ppp.y << "\ Z = " << ppp.z << endl; } return 0; } 又一篇在以前涉及的程序中，處理物體的旋轉(zhuǎn)通常是用的矩陣的形式。由于硬件在紋理映射和光柵化上的加強，程序員可以將更多的CPU周期用于物理模擬等工作，這樣將使得程序更為逼真。 一，原文出處： http://www.gamasutra.com/features/19980703/quaternions_01.htm 二，摘錄： 有三種辦法表示旋轉(zhuǎn)：矩陣表示，歐拉角表示，以及四元組表示。矩陣，歐拉角表示法在處理插值的時候會遇到麻煩。There are many ways to represent the orientation of an object. Most programmers use 3x3 rotation matrices or three Euler angles to store this information. Each of these solutions works fine until you try to smoothly interpolate between two orientations of an object. 矩陣表示法不適合于進行插值(在轉(zhuǎn)動的前后兩個朝向之間取得瞬時朝向)。矩陣有9個自由度（3×3矩陣），而實際上表示一個旋轉(zhuǎn)只需要3個自由度（旋轉(zhuǎn)軸3）。Rotations involve only three degrees of freedom (DOF), around the x, y, and z coordinate axes. However, nine DOF (assuming 3x3 matrices) are required to constrain the rotation - clearly more than we need. Another shortcoming of rotation matrices is that they are extremely hard to use for interpolating rotations between two orientations. The resulting interpolations are also visually very jerky, which simply is not acceptable in games any more. 歐拉角表示法You can also use angles to represent rotations around three coordinate axes. You can write this as (q, c, f); simply stated, "Transform a point by rotating it counterclockwise about the z axis by q degrees, followed by a rotation about the y axis by c degrees, followed by a rotation about the x axis by f degrees." 歐拉角表示法的缺陷：把一個旋轉(zhuǎn)變成了一系列的旋轉(zhuǎn)。并且，進行插值計算也不方便。 However, there is no easy way to represent a single rotation with Euler angles that corresponds to a series of concatenated rotations. Furthermore, the smooth interpolation between two orientations involves numerical integration 目錄[隱藏]? 1 四元組表示法 ? 2 四元組的基本運算法則 ? 3 用四元組進行旋轉(zhuǎn)的公式 ? 4 旋轉(zhuǎn)疊加性質(zhì) [編輯]四元組表示法[w, v]其中v是矢量，表示旋轉(zhuǎn)軸。w標量，表示旋轉(zhuǎn)角度。所以，一個四元組即表示一個完整的旋轉(zhuǎn)。There are several notations that we can use to represent quaternions. The two most popular notations are complex number notation (Eq. 1) and 4D vector notation (Eq. 2). w + xi + yj + zk (where i2 = j2 = k2 = -1 and ij = k = -ji with real w, x, y, z) (Eq. 1) [w, v] (where v = (x, y, z) is called a "vector" and w is called a "scalar") (Eq. 2) 4D空間以及單元四元組：Each quaternion can be plotted in 4D space (since each quaternion is comprised of four parts), and this space is called quaternion space. Unit quaternions have the property that their magnitude is one and they form a subspace, S3, of the quaternion space. This subspace can be represented as a 4D sphere. (those that have a one-unit normal), since this reduces the number of necessary operations that you have to perform. [編輯]四元組的基本運算法則Table 1. Basic operations using quaternions. Addition: q + q′ = [w + w′, v + v′] Multiplication: qq′ = [ww′ - v ? v′, v x v′ + wv′ +w′v] (? is vector dot product and x is vector cross product); Note: qq′ ? q′q //為什么是這樣？－－定義成這樣滴，木有道理可以講Conjugate: q* = [w, -v] 共軛 Norm: N(q) = w2 + x2 + y2 + z2 模 Inverse: q-1 = q* / N(q) Unit Quaternion: q is a unit quaternion if N(q)= 1 and then q-1 = q* Identity: [1, (0, 0, 0)] (when involving multiplication) and [0, (0, 0, 0)] (when involving addition) [編輯]用四元組進行旋轉(zhuǎn)的公式重要!!!!：只有單元四元組才表示旋轉(zhuǎn)。為什么????－－已解決（shoemake有詳細證明）It is extremely important to note that only unit quaternions represent rotations, and you can assume that when I talk about quaternions, I'm talking about unit quaternions unless otherwise specified. Since you've just seen how other methods represent rotations, let's see how we can specify rotations using quaternions. It can be proven (and the proof isn't that hard) that the rotation of a vector v by a unit quaternion q can be represented as v′ = q v q-1 (where v = [0, v]) (Eq. 3) //////////////////////////////////////////////////////////// 四元組和旋轉(zhuǎn)參數(shù)的變換一個四元組由(x,y,z,w)四個變量表達,假設給定一個旋轉(zhuǎn)軸axis和角度angle,那么這個四元組的計算公式為: void Quaternion::fromAxisAngle(const Vector3 &axis, Real angle){ Vector3 u = axis; u.normalize(); Real s = Math::rSin(angle/2.f); x = s*u.x; y = s*u.y; z = s*u.z; w = Math::rCos(angle/2.f); }由四元組反算回旋轉(zhuǎn)軸axis和角度angle的公式為: /void Quaternion::toAxisAngle(Vector3 &axis, Real &angle) const{ angle = acos(w)*2 axis.x = x axis.y = y axis.z = z}英文版:Rotating Objects Using QuaternionsLast year may go down in history as The Year of the Hardware Acceleration. Much of the work rasterizing and texture-mapping polygons was off-loaded to dedicated hardware. As a result, we game developers now have a lot of CPU cycles to spare for physics simulation and other features. Some of those extra cycles can be applied to tasks such as smoothing rotations and animations, thanks to quaternions. Many game programmers have already discovered the wonderful world of quaternions and have started to use them extensively. Several third-person games, including both TOMB RAIDER titles, use quaternion rotations to animate all of their camera movements. Every third-person game has a virtual camera placed at some distance behind or to the side of the player's character. Because this camera goes through different motions (that is, through arcs of a different lengths) than the character, camera motion can appear unnatural and too "jerky" for the player to follow the action. This is one area where quaternions come to rescue. Another common use for quaternions is in military and commercial flight simulators. Instead of manipulating a plane's orientation using three angles (roll, pitch, and yaw) representing rotations about the x, y, and z axes, respectively, it is much simpler to use a single quaternion. Many games coming out this year will also feature real-world physics, allowing amazing game play and immersion. If you store orientations as quaternions, it is computationally less expensive to add angular velocities to quaternions than to matrices. Both Tomb Raider titles use quaternion rotations to animate camera movements. There are many ways to represent the orientation of an object. Most programmers use 3x3 rotation matrices or three Euler angles to store this information. Each of these solutions works fine until you try to smoothly interpolate between two orientations of an object. Imagine an object that is not user controlled, but which simply rotates freely in space (for example, a revolving door). If you chose to store the door's orientations as rotation matrices or Euler angles, you'd find that smoothly interpolating between the rotation matrices' values would be computationally costly and certainly wouldn't appear as smooth to a player's eye as quaternion interpolation. Trying to correct this problem using matrices or Euler angles, an animator might simply increase the number of predefined (keyed) orientations. However, one can never be sure how many such orientations are enough, since the games run at different frame rates on different computers, thereby affecting the smoothness of the rotation. This is a good time to use quaternions, a method that requires only two or three orientations to represent a simple rotation of an object, such as our revolving door. You can also dynamically adjust the number of interpolated positions to correspond to a particular frame rate. Before we begin with quaternion theory and applications, let's look at how rotations can be represented. I'll touch upon methods such as rotation matrices, Euler angles, and axis and angle representations and explain their shortcomings and their relationships to quaternions. If you are not familiar with some of these techniques, I recommend picking up a graphics book and studying them. To date, I haven't seen a single 3D graphics book that doesn't talk about rotations using 4x4 or 3x3 matrices. Therefore, I will assume that most game programmers are very familiar with this technique and I'll just comment on its shortcomings. I also highly recommend that you re-read Chris Hecker's article in the June 1997 issue of the Game Developer ("Physics, Part 4: The Third Dimension," pp.15-26), since it tackles the problem of orienting 3D objects. Rotations involve only three degrees of freedom (DOF), around the x, y, and z coordinate axes. However, nine DOF (assuming 3x3 matrices) are required to constrain the rotation - clearly more than we need. Additionally, matrices are prone to "drifting," a situation that arises when one of the six constraints is violated and the matrix introduces rotations around an arbitrary axis. Combatting this problem requires keeping a matrix orthonormalized - making sure that it obeys constraints. However, doing so is not computationally cheap. A common way of solving matrix drift relies on the Gram-Schmidt algorithm for conversion of an arbitrary basis into an orthogonal basis. Using the Gram-Schmidt algorithm or calculating a correction matrix to solve matrix drifting can take a lot of CPU cycles, and it has to be done very often, even when using floating point math. Another shortcoming of rotation matrices is that they are extremely hard to use for interpolating rotations between two orientations. The resulting interpolations are also visually very jerky, which simply is not acceptable in games any more. You can also use angles to represent rotations around three coordinate axes. You can write this as (q, c, f); simply stated, "Transform a point by rotating it counterclockwise about the z axis by q degrees, followed by a rotation about the y axis by c degrees, followed by a rotation about the x axis by f degrees." There are 12 different conventions that you can use to represent rotations using Euler angles, since you can use any combination of axes to represent rotations (XYZ, XYX, XYY…). We will assume the first convention (XYZ) for all of the presented examples. I will assume that all of the positive rotations are counterclockwise (Figure 1). Euler angle representation is very efficient because it uses only three variables to represent three DOF. Euler angles also don't have to obey any constraints, so they're not prone to drifting and don't have to be readjusted. However, there is no easy way to represent a single rotation with Euler angles that corresponds to a series of concatenated rotations. Furthermore, the smooth interpolation between two orientations involves numerical integration, which can be computationally expensive. Euler angles also introduce the problem of "Gimbal lock" or a loss of one degree of rotational freedom. Gimbal lock happens when a series of rotations at 90 degrees is performed; suddenly, the rotation doesn't occur due to the alignment of the axes. Figure 1: Euler angle representation. For example, imagine that a series of rotations to be performed by a flight simulator. You specify the first rotation to be Q1 around the x axis, the second rotation to be 90 degrees around the y axis, and Q3 to be the rotation around the z axis. If you perform specified rotations in succession, you will discover that Q3 rotation around the z axis has the same effect as the rotation around the initial x axis. The y axis rotation has caused the x and z axes to get aligned, and you have just lost a DOF because rotation around one axis is equivalent to opposite rotation around the other axis. I highly recommend Advanced Animation and Rendering Techniques: Theory and Practice by Alan and Mark Watt (Addison Wesley, 1992) for a detailed discussion of the Gimbal lock problem. Using an axis and angle representation is another way of representing rotations. You specify an arbitrary axis and an angle (positive if in a counterclockwise direction), as illustrated in Figure 2. Even though this is an efficient way of representing a rotation, it suffers from the same problems that I described for Euler angle representation (with the exception of the Gimbal lock problem). In the eighteenth century, W. R. Hamilton devised quaternions as a four-dimensional extension to complex numbers. Soon after this, it was proven that quaternions could also represent rotations and orientations in three dimensions. There are several notations that we can use to represent quaternions. The two most popular notations are complex number notation (Eq. 1) and 4D vector notation (Eq. 2). w + xi + yj + zk (where i2 = j2 = k2 = -1 and ij = k = -ji with real w, x, y, z) (Eq. 1) [w, v] (where v = (x, y, z) is called a "vector" and w is called a "scalar") (Eq. 2) I will use the second notation throughout this article. Now that you know how quaternions are represented, let's start with some basic operations that use them. If q and q′ are two orientations represented as quaternions, you can define the operations in Table 1 on these quaternions. All other operations can be easily derived from these basic ones, and they are fully documented in the accompanying library, which you can find here. I will also only deal with unit quaternions. Each quaternion can be plotted in 4D space (since each quaternion is comprised of four parts), and this space is called quaternion space. Unit quaternions have the property that their magnitude is one and they form a subspace, S3, of the quaternion space. This subspace can be represented as a 4D sphere. (those that have a one-unit normal), since this reduces the number of necessary operations that you have to perform. It is extremely important to note that only unit quaternions represent rotations, and you can assume that when I talk about quaternions, I'm talking about unit quaternions unless otherwise specified. Since you've just seen how other methods represent rotations, let's see how we can specify rotations using quaternions. It can be proven (and the proof isn't that hard) that the rotation of a vector v by a unit quaternion q can be represented as v′ = q v q-1 (where v = [0, v]) (Eq. 3) The result, a rotated vector v′, will always have a 0 scalar value for w (recall Eq. 2 earlier), so you can omit it from your computations. Table 1. Basic operations using quaternions.Addition: q + q′ = [w + w′, v + v′] Multiplication: qq′ = [ww′ - v ? v′, v x v′ + wv′ +w′v] (? is vector dot product and x is vector cross product); Note: qq′ ? q′q Conjugate: q* = [w, -v] Norm: N(q) = w2 + x2 + y2 + z2 Inverse: q-1 = q* / N(q) Unit Quaternion: q is a unit quaternion if N(q)= 1 and then q-1 = q*Identity: [1, (0, 0, 0)] (when involving multiplication) and [0, (0, 0, 0)] (when involving addition) Today's most widely supported APIs, Direct3D immediate mode (retained mode does have a limited set of quaternion rotations) and OpenGL, do not support quaternions directly. As a result, you have to convert quaternion orientations in order to pass this information to your favorite API. Both OpenGL and Direct3D give you ways to specify rotations as matrices, so a quaternion-to-matrix conversion routine is useful. Also, if you want to import scene information from a graphics package that doesn't store its rotations as a series of quaternions (such as NewTek's LightWave), you need a way to convert to and from quaternion space. ANGLE AND AXIS. Converting from angle and axis notation to quaternion notation involves two trigonometric operations, as well as several multiplies and divisions. It can be represented as q = [cos(Q/2), sin(Q /2)v] (where Q is an angle and v is an axis) (Eq. 4) EULER ANGLES. Converting Euler angles into quaternions is a similar process - you just have to be careful that you perform the operations in the correct order. For example, let's say that a plane in a flight simulator first performs a yaw, then a pitch, and finally a roll. You can represent this combined quaternion rotation as q = qyaw qpitch qroll where: qroll = [cos (y/2), (sin(y/2), 0, 0)] qpitch = [cos (q/2), (0, sin(q/2), 0)] qyaw = [cos(f /2), (0, 0, sin(f /2)] (Eq. 5) The order in which you perform the multiplications is important. Quaternion multiplication is not commutative (due to the vector cross product that's involved). In other words, changing the order in which you rotate an object around various axes can produce different resulting orientations, and therefore, the order is important. ROTATION MATRIX. Converting from a rotation matrix to a quaternion representation is a bit more involved, and its implementation can be seen in Listing 1.Conversion between a unit quaternion and a rotation matrix can be specified as | 1 - 2y2 - 2z2 2yz + 2wx 2xz - 2wy | Rm = | 2xy - 2wz 1 - 2x2 - 2z2 2yz - 2wx | | 2xz + 2wy 2yz - 2wx 1 - 2x2 - 2y2| (Eq. 6) It's very difficult to specify a rotation directly using quaternions. It's best to store your character's or object's orientation as a Euler angle and convert it to quaternions before you start interpolating. It's much easier to increment rotation around an angle, after getting the user's input, using Euler angles (that is, roll = roll + 1), than to directly recalculate a quaternion. Since converting between quaternions and rotation matrices and Euler angles is performed often, it's important to optimize the conversion process. Very fast conversion (involving only nine muls) between a unit quaternion and a matrix is presented in Listing 2. Please note that the code assumes that a matrix is in a right-hand coordinate system and that matrix rotation is represented in a column major format (for example, OpenGL compatible).

Listing 1: Matrix to quaternion code. MatToQuat(float m[4][4], QUAT * quat){ float tr, s, q[4]; int i, j, k; int nxt[3] = {1, 2, 0}; tr = m[0][0] + m[1][1] + m[2][2]; // check the diagonal if (tr > 0.0) { s = sqrt (tr + 1.0); quat->w = s / 2.0; s = 0.5 / s; quat->x = (m[1][2] - m[2][1]) * s; quat->y = (m[2][0] - m[0][2]) * s; quat->z = (m[0][1] - m[1][0]) * s;} else { // diagonal is negative i = 0; if (m[1][1] > m[0][0]) i = 1; if (m[2][2] > m[i][i]) i = 2; j = nxt[i]; k = nxt[j]; s = sqrt ((m[i][i] - (m[j][j] + m[k][k])) + 1.0); q[i] = s * 0.5; if (s != 0.0) s = 0.5 / s; q[3] = (m[j][k] - m[k][j]) * s; q[j] = (m[i][j] + m[j][i]) * s; q[k] = (m[i][k] + m[k][i]) * s; quat->x = q[0]; quat->y = q[1]; quat->z = q[2]; quat->w = q[3]; }}If you aren't dealing with unit quaternions, additional multiplications and a division are required. Euler angle to quaternion conversion can be coded as shown in Listing 3. One of the most useful aspects of quaternions that we game programmers are concerned with is the fact that it's easy to interpolate between two quaternion orientations and achieve smooth animation. To demonstrate why this is so, let's look at an example using spherical rotations. Spherical quaternion interpolations follow the shortest path (arc) on a four-dimensional, unit quaternion sphere. Since 4D spheres are difficult to imagine, I'll use a 3D sphere (Figure 3) to help you visualize quaternion rotations and interpolations. Let's assume that the initial orientation of a vector emanating from the center of the sphere can be represented by q1 and the final orientation of the vector is q3. The arc between q1 and q3 is the path that the interpolation would follow. Figure 3 also shows that if we have an intermediate position q2, the interpolation from q1 -> q2 -> q3 will not necessarily follow the same path as the q1 ->q3 interpolation. The initial and final orientations are the same, but the arcs are not. Quaternions simplify the calculations required when compositing rotations. For example, if you have two or more orientations represented as matrices, it is easy to combine them by multiplying two intermediate rotations. R = R2R1 (rotation R1 followed by a rotation R2) (Eq. 7) Listing 2: Quaternion-to-matrix conversion. ([07.30.02] Editor's Note: the following QuatToMatrix function originally was published with a bug -- it reversed the row/column ordering. This is the correct version. Thanks to John Ratcliff and Eric Haines for pointing this out.) QuatToMatrix(QUAT * quat, float m[4][4]){float wx, wy, wz, xx, yy, yz, xy, xz, zz, x2, y2, z2; // calculate coefficientsx2 = quat->x + quat->x; y2 = quat->y + quat->y; z2 = quat->z + quat->z;xx = quat->x * x2; xy = quat->x * y2; xz = quat->x * z2;yy = quat->y * y2; yz = quat->y * z2; zz = quat->z * z2;wx = quat->w * x2; wy = quat->w * y2; wz = quat->w * z2;m[0][0] = 1.0 - (yy + zz); m[1][0] = xy - wz;m[2][0] = xz + wy; m[3][0] = 0.0;m[0][1] = xy + wz; m[1][1] = 1.0 - (xx + zz);m[2][1] = yz - wx; m[3][1] = 0.0;m[0][2] = xz - wy; m[1][2] = yz + wx;m[2][2] = 1.0 - (xx + yy); m[3][2] = 0.0;m[0][3] = 0; m[1][3] = 0;m[2][3] = 0; m[3][3] = 1;} This composition involves 27 multiplications and 18 additions, assuming 3x3 matrices. On the other hand, a quaternion composition can be represented as q = q2q1 (rotation q1 followed by a rotation q2) (Eq. 8) As you can see, the quaternion method is analogous to the matrix composition. However, the quaternion method requires only eight multiplications and four divides (Listing 4), so compositing quaternions is computationally cheap compared to matrix composition. Savings such as this are especially important when working with hierarchical object representations and inverse kinematics. Now that you have an efficient multiplication routine, see how can you interpolate between two quaternion rotations along the shortest arc. Spherical Linear intERPolation (SLERP) achieves this and can be written as (Eq. 9) where pq = cos(q) and parameter t goes from 0 to 1. The implementation of this equation is presented in Listing 5. If two orientations are too close, you can use linear interpolation to avoid any divisions by zero. Figure 3. Quaternion rotations. Listing 3: Euler-to-quaternion conversion. EulerToQuat(float roll, float pitch, float yaw, QUAT * quat){ float cr, cp, cy, sr, sp, sy, cpcy, spsy;// calculate trig identitiescr = cos(roll/2); cp = cos(pitch/2); cy = cos(yaw/2); sr = sin(roll/2); sp = sin(pitch/2); sy = sin(yaw/2); cpcy = cp * cy; spsy = sp * sy; quat->w = cr * cpcy + sr * spsy; quat->x = sr * cpcy - cr * spsy; quat->y = cr * sp * cy + sr * cp * sy; quat->z = cr * cp * sy - sr * sp * cy;}The basic SLERP rotation algorithm is shown in Listing 6. Note that you have to be careful that your quaternion represents an absolute and not a relative rotation. You can think of a relative rotation as a rotation from the previous (intermediate) orientation and an absolute rotation as the rotation from the initial orientation. This becomes clearer if you think of the q2 quaternion orientation in Figure 3 as a relative rotation, since it moved with respect to the q1 orientation. To get an absolute rotation of a given quaternion, you can just multiply the current relative orientation by a previous absolute one. The initial orientation of an object can be represented as a multiplication identity [1, (0, 0, 0)]. This means that the first orientation is always an absolute one, because q = qidentity q (Eq. 10) Listing 4: Efficient quaternion multiplication. QuatMul(QUAT *q1, QUAT *q2, QUAT *res){float A, B, C, D, E, F, G, H;A = (q1->w + q1->x)*(q2->w + q2->x);B = (q1->z - q1->y)*(q2->y - q2->z);C = (q1->w - q1->x)*(q2->y + q2->z); D = (q1->y + q1->z)*(q2->w - q2->x);E = (q1->x + q1->z)*(q2->x + q2->y);F = (q1->x - q1->z)*(q2->x - q2->y);G = (q1->w + q1->y)*(q2->w - q2->z);H = (q1->w - q1->y)*(q2->w + q2->z);res->w = B + (-E - F + G + H) /2;res->x = A - (E + F + G + H)/2; res->y = C + (E - F + G - H)/2; res->z = D + (E - F - G + H)/2;}As I stated earlier, a practical use for quaternions involves camera rotations in third-person-perspective games. Ever since I saw the camera implementation in TOMB RAIDER, I've wanted to implement something similar. So let's implement a third-person camera (Figure 4).To start off, let's create a camera that is always positioned above the head of our character and that points at a spot that is always slightly above the character's head. The camera is also positioned d units behind our main character. We can also implement it so that we can vary the roll (angle q in Figure 4) by rotating around the x axis. As soon as a player changes the orientation of the character, you rotate the character instantly and use SLERP to reorient the camera behind the character (Figure 5). This has the dual benefit of providing smooth camera rotations and making players feel as though the game responded instantly to their input. Figure 4. Third-person camera. Figure 5. Camera from top. You can set the camera's center of rotation (pivot point) as the center of the object it is tracking. This allows you to piggyback on the calculations that the game already makes when the character moves within the game world. Note that I do not recommend using quaternion interpolation for first-person action games since these games typically require instant response to player actions, and SLERP does take time. However, we can use it for some special scenes. For instance, assume that you're writing a tank simulation. Every tank has a scope or similar targeting mechanism, and you'd like to simulate it as realistically as possible. The scoping mechanism and the tank's barrel are controlled by a series of motors that players control. Depending on the zoom power of the scope and the distance to a target object, even a small movement of a motor could cause a large change in the viewing angle, resulting in a series of huge, seemingly disconnected jumps between individual frames. To eliminate this unwanted effect, you could interpolate the orientation according to the zoom and distance of object. This type of interpolation between two positions over several frames helps dampen the rapid movement and keeps players from becoming disoriented. Another useful application of quaternions is for prerecorded (but not prerendered) animations. Instead of recording camera movements by playing the game (as many games do today), you could prerecord camera movements and rotations using a commercial package such as Softimage 3D or 3D Studio MAX. Then, using an SDK, export all of the keyframed camera/object quaternion rotations. This would save both space and rendering time. Then you could just play the keyframed camera motions whenever the script calls for cinematic scenes. Listing 5: SLERP implementation. QuatSlerp(QUAT * from, QUAT * to, float t, QUAT * res) { float to1[4]; double omega, cosom, sinom, scale0, scale1; // calc cosine cosom = from->x * to->x + from->y * to->y + from->z * to->z + from->w * to->w; // adjust signs (if necessary) if ( cosom <0.0 ){ cosom = -cosom; to1[0] = - to->x; to1[1] = - to->y; to1[2] = - to->z; to1[3] = - to->w; } else { to1[0] = to->x; to1[1] = to->y; to1[2] = to->z; to1[3] = to->w; } // calculate coefficients if ( (1.0 - cosom) > DELTA ) { // standard case (slerp) omega = acos(cosom); sinom = sin(omega); scale0 = sin((1.0 - t) * omega) / sinom; scale1 = sin(t * omega) / sinom; } else { // "from" and "to" quaternions are very close // ... so we can do a linear interpolation scale0 = 1.0 - t; scale1 = t; } // calculate final values res->x = scale0 * from->x + scale1 * to1[0]; res->y = scale0 * from->y + scale1 * to1[1]; res->z = scale0 * from->z + scale1 * to1[2]; res->w = scale0 * from->w + scale1 * to1[3];}After reading Chris Hecker's columns on physics last year, I wanted to add angular velocity to a game engine on which I was working. Chris dealt mainly with matrix math, and because I wanted to eliminate quaternion-to-matrix and matrix-to-quaternion conversions (since our game engine is based on quaternion math), I did some research and found out that it is easy to add angular velocity (represented as a vector) to a quaternion orientation. The solution (Eq. 11) can be represented as a differential equation. (Eq. 11) where quat(angular) is a quaternion with a zero scalar part (that is, w = 0) and a vector part equal to the angular velocity vector. Q is our original quaternion orientation. To integrate the above equation (Q + dQ/dt), I recommend using the Runge-Kutta order four method. If you are using matrices, the Runge-Kutta order five method achieves better results within a game. (The Runge-Kutta method is a way of integrating differential equations. A complete description of the method can be found in any elementary numerical algorithm book, such as Numerical Recipes in C. It has a complete section devoted to numerical, differential integration.) For a complete derivation of angular velocity integration, consult Dave Baraff's SIGGRAPH tutorials. Quaternions can be a very efficient and extremely useful method of storing and performing rotations, and they offer many advantages over other methods. Unfortunately, they are also impossible to visualize and completely unintuitive. However, if you use quaternions to represent rotations internally, and use some other method (for example, angle-axis or Euler angles) as an immediate representation, you won't have to visualize them. Nick Bobick is a game developer at Caged Entertainment Inc. and he is currently working on a cool 3D game. He can be contacted at nb@netcom.ca. The author would like to thank Ken Shoemake for his research and publications. Without him, this article would not have been possible.

以上就是關(guān)于pos機用日文怎么說,為什么使用四元數(shù)的知識，后面我們會繼續(xù)為大家整理關(guān)于pos機用日文怎么說的知識，希望能夠幫助到大家！