diff --git a/fractional/src/geometry.rs b/fractional/src/geometry.rs
index ede5968..3319ca5 100644
--- a/fractional/src/geometry.rs
+++ b/fractional/src/geometry.rs
@@ -18,12 +18,12 @@
 // You should have received a copy of the GNU General Public License
 // along with this program.  If not, see <http://www.gnu.org/licenses/>.
 //
-use std::convert::From;
+use std::convert::{From, Into};
 use std::ops::{Add,Sub,Neg,Mul,Div};
 use std::fmt::Debug;
 
 use crate::easel::{Canvas,Coordinate,Coordinates,Polygon};
-use crate::transform::TMatrix;
+use crate::transform::{TMatrix, Transformable};
 use crate::trigonometry::Trig;
 use crate::vector::Vector;
 
@@ -34,10 +34,113 @@ where T: Add + Sub + Neg + Mul + Div + Copy + Trig {
     normal  :Option<Vector<T>>,
 }
 
+#[derive(Debug, PartialEq, Eq, Clone, Copy)]
+pub struct Point<T>(pub Vector<T>, T)
+    where T: Add + Sub + Neg + Mul + Div + PartialEq + Copy + Trig;
+
+impl<T> Point<T>
+where T: Add<Output = T> + Sub<Output = T> + Neg<Output = T>
+       + Mul<Output = T> + Div<Output = T>
+       + PartialEq + Trig + Copy + From<i32> {
+    pub fn new(x :T, y :T, z :T) -> Self {
+        Self(Vector(x, y, z), 1.into())
+    }
+}
+
+impl<T> Add for Point<T>
+where T: Add<Output = T> + Sub<Output = T> + Neg<Output = T>
+       + Mul<Output = T> + Div<Output = T>
+       + PartialEq + Trig + Copy {
+    type Output = Self;
+
+    fn add(self, other :Self) -> Self {
+        let Point(v1, w1) = self;
+        let Point(v2, w2) = other;
+        Self(v1 + v2, w1 + w2)
+    }
+}
+
+impl<T> Neg for Point<T>
+where T: Add<Output = T> + Sub<Output = T> + Neg<Output = T>
+       + Mul<Output = T> + Div<Output = T>
+       + PartialEq + Trig + Copy {
+    type Output = Self;
+
+    fn neg(self) -> Self {
+        let Point(v, w) = self;
+        Self(-v, -w)
+    }
+}
+
+impl<T> Sub for Point<T>
+where T: Add<Output = T> + Sub<Output = T> + Neg<Output = T>
+       + Mul<Output = T> + Div<Output = T>
+       + PartialEq + Trig + Copy {
+    type Output = Self;
+
+    fn sub(self, other :Self) -> Self {
+        self + -other
+    }
+}
+
+impl<T> Mul for Point<T>
+where T: Add<Output = T> + Sub<Output = T> + Neg<Output = T>
+       + Mul<Output = T> + Div<Output = T>
+       + PartialEq + Trig + Copy + From<i32> {
+    type Output = Self;
+
+    fn mul(self, other :Self) -> Self {
+        let a :Vector<T> = self.into();
+        let b :Vector<T> = other.into();
+
+        Point(a * b, 1.into())
+    }
+}
+
+impl<T> From<Vector<T>> for Point<T>
+where T: Add<Output = T> + Sub<Output = T> + Neg<Output = T>
+       + Mul<Output = T> + Div<Output = T>
+       + PartialEq + Trig + Copy + From<i32> {
+    fn from(v :Vector<T>) -> Self {
+        Point(v, 1.into())
+    }
+}
+
+impl<T> Into<Vector<T>> for Point<T>
+where T: Add<Output = T> + Sub<Output = T> + Neg<Output = T>
+       + Mul<Output = T> + Div<Output = T>
+       + PartialEq + Trig + Copy + From<i32> {
+    fn into(self) -> Vector<T> {
+        let Point(v, w) = self;
+
+        if w == 0.into() {
+            v
+        } else {
+            v.mul(&w.recip())
+        }
+    }
+}
+
+impl<T> Transformable<T> for Point<T>
+where T: Add<Output = T> + Sub<Output = T> + Neg<Output = T>
+       + Mul<Output = T> + Div<Output = T>
+       + PartialEq + Debug + Trig + Copy + From<i32> {
+    fn transform(&self, m :&TMatrix<T>) -> Self {
+        let Point(v, w) = *self;
+        let (v, w)      = m.apply(&v, w);
+
+        if w == 0.into() {
+            v.into()
+        } else {
+            v.mul(&w.recip()).into()
+        }
+    }
+}
+
 #[derive(Debug)]
 pub struct Polyeder<T>
-where T: Add + Sub + Neg + Mul + Div + Copy + Trig {
-    points :Vec<Vector<T>>,
+where T: Add + Sub + Neg + Mul + Div + PartialEq + Copy + Trig {
+    points :Vec<Point<T>>,
     faces  :Vec<Face<T>>,
 }
 
@@ -65,7 +168,7 @@ where T: Add + Sub + Neg + Mul + Div + Debug + Copy + Trig + From<i32> {
 impl<T> Camera<T>
 where T: Add<Output = T> + Sub<Output = T> + Neg<Output = T>
        + Mul<Output = T> + Div<Output = T>
-       + Debug + Copy + Trig + From<i32> {
+       + PartialEq + Debug + Copy + Trig + From<i32> {
     // This code assumes that the size of the viewport is always
     // equal to the size of the physical screen… e.g. window/canvas thus some
     // effects can't be done. See book for examples with different viewport
@@ -91,9 +194,9 @@ where T: Add<Output = T> + Sub<Output = T> + Neg<Output = T>
         self.project
     }
 
-    pub fn project(&self, v :Vector<T>) -> Coordinate {
-        let p = self.project.apply(&v);
-        Coordinate(T::round(&p.x()), T::round(&p.y()))
+    pub fn project(&self, p :Point<T>) -> Coordinate {
+        let Point(v, _) = p.transform(&self.project);
+        Coordinate(T::round(&v.x()), T::round(&v.y()))
     }
 }
 
@@ -113,16 +216,16 @@ where T: Add<Output = T> + Sub<Output = T> + Neg<Output = T>
 impl<T> Face<T>
 where T: Add<Output = T> + Sub<Output = T> + Neg<Output = T>
        + Mul<Output = T> + Div<Output = T>
-       + Debug + Copy + Trig + From<i32> {
-    fn new(corners :Vec<usize>, ps :&[Vector<T>]) -> Self {
+       + PartialEq + Debug + Copy + Trig + From<i32> {
+    fn new(corners :Vec<usize>, ps :&[Point<T>]) -> Self {
         let mut f = Face{ corners: corners, normal: None };
         f.update_normal(ps);
         f
     }
 
-    fn update_normal(&mut self, ps :&[Vector<T>]) {
-        let edge10 = ps[self.corners[1]] - ps[self.corners[0]];
-        let edge12 = ps[self.corners[1]] - ps[self.corners[2]];
+    fn update_normal(&mut self, ps :&[Point<T>]) {
+        let edge10 :Vector<T> = (ps[self.corners[1]] - ps[self.corners[0]]).into();
+        let edge12 :Vector<T> = (ps[self.corners[1]] - ps[self.corners[2]]).into();
         self.normal = Some(edge10 * edge12);
     }
 }
@@ -130,7 +233,7 @@ where T: Add<Output = T> + Sub<Output = T> + Neg<Output = T>
 impl<T> Polyeder<T>
 where T: Add<Output = T> + Sub<Output = T> + Neg<Output = T>
        + Mul<Output = T> + Div<Output = T>
-       + Debug + Copy + Trig + From<i32> {
+       + PartialEq + Debug + Copy + Trig + From<i32> {
     fn update_normals(&mut self) {
         for f in self.faces.iter_mut() {
             f.update_normal(&self.points);
@@ -156,10 +259,10 @@ where T: Add<Output = T> + Sub<Output = T> + Neg<Output = T>
         // half the deeps in z
         let _zh :T = T::sqrt(f3).unwrap() / f4 * a;
 
-        let ps = vec!( Vector( f0,  yc,  f0)
-                     , Vector(-ah, -yi, -zi)
-                     , Vector( ah, -yi, -zi)
-                     , Vector( f0, -yi,  zc) );
+        let ps = vec!( Point::new( f0,  yc,  f0)
+                     , Point::new(-ah, -yi, -zi)
+                     , Point::new( ah, -yi, -zi)
+                     , Point::new( f0, -yi,  zc) );
 
         let fs = vec!( Face::new(vec!(1, 2, 3), &ps)
                      , Face::new(vec!(1, 0, 2), &ps)
@@ -177,9 +280,9 @@ where T: Add<Output = T> + Sub<Output = T> + Neg<Output = T>
         let zc :T = T::sqrt(f3).unwrap() / f3 * a;
         let ah :T = a / 2.into();
 
-        let ps = vec!( Vector(-ah, f0, -zi)
-                     , Vector( f0, f0,  zc)
-                     , Vector( ah, f0, -zi) );
+        let ps = vec!( Point::new(-ah, f0, -zi)
+                     , Point::new( f0, f0,  zc)
+                     , Point::new( ah, f0, -zi) );
 
         let fs = vec!(Face::new(vec!(0, 1, 2), &ps));
 
@@ -191,14 +294,14 @@ where T: Add<Output = T> + Sub<Output = T> + Neg<Output = T>
     pub fn cube(a :T) -> Polyeder<T> {
         let ah :T = a / From::<i32>::from(2);
 
-        let ps = vec!( Vector(-ah,  ah, -ah)    // 0 => front 1
-                     , Vector(-ah, -ah, -ah)    // 1 => front 2
-                     , Vector( ah, -ah, -ah)    // 2 => front 3
-                     , Vector( ah,  ah, -ah)    // 3 => front 4
-                     , Vector(-ah,  ah,  ah)    // 4 => back 1
-                     , Vector(-ah, -ah,  ah)    // 5 => back 2
-                     , Vector( ah, -ah,  ah)    // 6 => back 3
-                     , Vector( ah,  ah,  ah) );  // 7 => back 4
+        let ps = vec!( Point::new(-ah,  ah, -ah)    // 0 => front 1
+                     , Point::new(-ah, -ah, -ah)    // 1 => front 2
+                     , Point::new( ah, -ah, -ah)    // 2 => front 3
+                     , Point::new( ah,  ah, -ah)    // 3 => front 4
+                     , Point::new(-ah,  ah,  ah)    // 4 => back 1
+                     , Point::new(-ah, -ah,  ah)    // 5 => back 2
+                     , Point::new( ah, -ah,  ah)    // 6 => back 3
+                     , Point::new( ah,  ah,  ah) );  // 7 => back 4
 
         let fs = vec!( Face::new(vec!(0, 1, 2, 3), &ps)    // front
                      , Face::new(vec!(7, 6, 5, 4), &ps)    // back
@@ -218,8 +321,10 @@ where T: Add<Output = T> + Sub<Output = T> + Neg<Output = T>
     fn transform(&self, m :&TMatrix<T>) -> Self {
         let Polyeder{ points: ps, faces: fs } = self;
 
-        let mut p = Polyeder{ points: ps.iter().map(|p| m.apply(p)).collect()
-                            , faces:  fs.to_vec() };
+        let mut p = Polyeder{
+              points: ps.iter().map(|p| p.transform(m)).collect()
+            , faces:  fs.to_vec()
+        };
 
         // TODO alternatively we could rotate the normals too, but this cannot
         // done with the original matrix… the question is, what is faster.
diff --git a/fractional/src/main.rs b/fractional/src/main.rs
index 139e69a..6f09ef1 100644
--- a/fractional/src/main.rs
+++ b/fractional/src/main.rs
@@ -34,7 +34,7 @@ use fractional::easel::{ Coordinate, Coordinates, Drawable, Line, Polyline
 use fractional::fractional::{Fractional, from_vector};
 use fractional::trigonometry::Trig;
 use fractional::vector::Vector;
-use fractional::transform::TMatrix;
+use fractional::transform::{TMatrix, Transformable};
 
 use fractional::xcb::XcbEasel;
 use fractional::easel::Canvas;
@@ -204,7 +204,8 @@ fn _transform<T>(v :Vector<T>, v1 :Vector<T>, v2 :Vector<T>, v3 :Vector<T>)
            + Debug + From<i32> + Copy + Display {
 
     println!("{:>14} : {}", "Vector v1", v1);
-    println!("{:>14} : {}", "translate v1", TMatrix::translate(v).apply(&v1));
+    println!( "{:>14} : {}", "translate v1"
+            , v.transform(&TMatrix::translate(v)));
     println!();
 
     fn _rot<T>( o :&str , n :&str , v :&Vector<T>
@@ -218,7 +219,7 @@ fn _transform<T>(v :Vector<T>, v1 :Vector<T>, v2 :Vector<T>, v3 :Vector<T>)
             let mi = fs.iter().map(|f| f(*d as i32));
             println!( "{:>14} : {}"
                     , format!("{} {} {}", o, d, n)
-                    , TMatrix::combine(mi).apply(v) );
+                    , v.transform(&TMatrix::combine(mi)) );
         }
     }
 
@@ -237,7 +238,7 @@ fn _transform<T>(v :Vector<T>, v1 :Vector<T>, v2 :Vector<T>, v3 :Vector<T>)
              , 210, 225, 240, 270, 300, 315, 330 ].iter() {
         println!( "{:>14} : {}"
                 , format!("rot_v {} v2", d)
-                , TMatrix::rotate_v(&v, *d as i32).apply(&v2));
+                , v2.transform(&TMatrix::rotate_v(&v, *d as i32)) );
     }
 }
 
@@ -273,9 +274,9 @@ fn _line() {
     let k = Vector(Fractional(-30,1), Fractional( 30,1), Fractional(0,1));
 
     let rot :TMatrix<Fractional> = TMatrix::rotate_z(20);
-    let Vector(ix, iy, _) = rot.apply(&i);
-    let Vector(jx, jy, _) = rot.apply(&j);
-    let Vector(kx, ky, _) = rot.apply(&k);
+    let Vector(ix, iy, _) = i.transform(&rot);
+    let Vector(jx, jy, _) = j.transform(&rot);
+    let Vector(kx, ky, _) = k.transform(&rot);
 
     fn to_i32(x :Fractional) -> i32 {
         let Fractional(n, d) = x;
@@ -294,9 +295,9 @@ fn _line() {
     let k = Vector(-30.0,  30.0, 0.0);
 
     let rot :TMatrix<f64> = TMatrix::rotate_z(20);
-    let Vector(ix, iy, _) = rot.apply(&i);
-    let Vector(jx, jy, _) = rot.apply(&j);
-    let Vector(kx, ky, _) = rot.apply(&k);
+    let Vector(ix, iy, _) = i.transform(&rot);
+    let Vector(jx, jy, _) = j.transform(&rot);
+    let Vector(kx, ky, _) = k.transform(&rot);
 
     fn to_i32_2(x :f64) -> i32 {
         x.round() as i32
@@ -334,24 +335,15 @@ fn _democanvas<T>( xcb         :&XcbEasel
         let     step  = Duration::from_millis(25);
         let mut last  = Instant::now();
 
-        let t :TMatrix<T> = TMatrix::translate(Vector( 0.into()
-                                                     , 0.into()
-                                                     , 150.into() ));
-        // We do not need this here… it is used within projection…
-        // let p :TMatrix<T> = camera.get_projection();
+        let t = TMatrix::translate(Vector(0.into() , 0.into() , 150.into()));
 
         loop {
             let deg = ((start.elapsed() / 25).as_millis() % 360) as i32;
 
-            let  rz :TMatrix<T> = TMatrix::rotate_z(deg);
-            let  rx :TMatrix<T> = TMatrix::rotate_x(-deg*2);
-            let  ry :TMatrix<T> = TMatrix::rotate_y(-deg*2);
+            let rz = TMatrix::rotate_z(deg);
+            let rx = TMatrix::rotate_x(-deg*2);
+            let ry = TMatrix::rotate_y(-deg*2);
 
-            // I can not apply the projection in one turn, as I generate the
-            // normals always… and this is no longer possible after the
-            // projection…
-            // let rot1 = TMatrix::combine(vec!(rz, rx, t, p));
-            // let rot2 = TMatrix::combine(vec!(rz, ry, t, p));
             let rot1 = TMatrix::combine(vec!(rz, rx, t));
             let rot2 = TMatrix::combine(vec!(rz, ry, t));
 
diff --git a/fractional/src/transform.rs b/fractional/src/transform.rs
index db67249..e7e3142 100644
--- a/fractional/src/transform.rs
+++ b/fractional/src/transform.rs
@@ -31,6 +31,11 @@ pub struct TMatrix<T>( (T, T, T, T)
                      , (T, T, T, T) )
     where T: Add + Sub + Neg + Mul + Div + Debug + Trig + From<i32> + Copy;
 
+pub trait Transformable<T>
+where T: Add + Sub + Neg + Mul + Div + Debug + Trig + From<i32> + Copy {
+    fn transform(&self, m :&TMatrix<T>) -> Self;
+}
+
 impl<T> TMatrix<T>
 where T: Add<Output = T> + Sub<Output = T> + Neg<Output = T>
        + Mul<Output = T> + Div<Output = T>
@@ -127,19 +132,20 @@ where T: Add<Output = T> + Sub<Output = T> + Neg<Output = T>
         mi.into_iter().fold(Self::unit(), |acc, x| x * acc)
     }
 
-    pub fn apply(&self, v :&Vector<T>) -> Vector<T> {
+    pub fn apply(&self, v :&Vector<T>, w :T) -> (Vector<T>, T) {
         let TMatrix( (a11, a12, a13, a14)
                    , (a21, a22, a23, a24)
                    , (a31, a32, a33, a34)
                    , (a41, a42, a43, a44) ) = *self;
         let Vector(x, y, z) = *v;
 
-        let v = Vector( a11 * x + a12 * y + a13 * z + a14
-                      , a21 * x + a22 * y + a23 * z + a24
-                      , a31 * x + a32 * y + a33 * z + a34 );
-        let w = a41 * x + a42 * y + a43 * z + a44;
+        let v = Vector( a11 * x + a12 * y + a13 * z + a14 * w
+                      , a21 * x + a22 * y + a23 * z + a24 * w
+                      , a31 * x + a32 * y + a33 * z + a34 * w );
+        let w = a41 * x + a42 * y + a43 * z + a44 * w;
 
-        v.mul(&w.recip())
+        //v.mul(&w.recip())
+        (v, w)
     }
 }
 
diff --git a/fractional/src/vector.rs b/fractional/src/vector.rs
index 30d126a..28ad18c 100644
--- a/fractional/src/vector.rs
+++ b/fractional/src/vector.rs
@@ -18,10 +18,11 @@
 // You should have received a copy of the GNU General Public License
 // along with this program.  If not, see <http://www.gnu.org/licenses/>.
 //
-use std::fmt::{Formatter, Display, Result};
+use std::fmt::{Debug, Display, Formatter, Result};
 use std::ops::{Add, Sub, Neg, Mul, Div};
 
 use crate::trigonometry::Trig;
+use crate::transform::{TMatrix, Transformable};
 
 #[derive(Debug, Eq, Clone, Copy)]
 pub struct Vector<T>(pub T, pub T, pub T)
@@ -126,3 +127,13 @@ where T: Add<Output = T> + Sub<Output = T> + Neg<Output = T>
               , ax * by - ay * bx )
     }
 }
+
+impl<T> Transformable<T> for Vector<T>
+where T: Add<Output = T> + Sub<Output = T> + Neg<Output = T>
+       + Mul<Output = T> + Div<Output = T>
+       + Trig + Copy + Debug + From<i32> {
+    fn transform(&self, m :&TMatrix<T>) -> Self {
+        let (v, _) = m.apply(self, 0.into());
+        v
+    }
+}