diff --git a/example.ez80 b/example.ez80
index 78091ce..5352bb0 100644
--- a/example.ez80
+++ b/example.ez80
@@ -26,9 +26,13 @@
 	ld	de, $D30000-2
 	call	vxZX7Uncompress		; uncompress the texture generated by convpng
 
+; about vertex coordinate :
+; the format inputed in glib is pure integer 16 bits coordinates, ]-32768,32768[
+; A 1.0 coordinate in blender is equivalent to 256 in glib
+
 ; setup global variable for rendering, euler angle and the translation of WorldMatrix
 
-; about translation : the format inputed in glib is pure integer 16 bits coordinates. On blender the conversion is made like this : 1.0=256
+; about translation :
 ; translation are added post transformation. So they carry the multiplication with the matrice, which aren't divided by the fixed point value of 1.0 (the matrix equivalent or 64). To have a translation of 1.0 in worldview matrix, the translation need to be 256*64 = 16384.
 ; However, there is two type of matrix in glib : the model matrix, which is in MODELSPACE and the world type matrix in WORLDSPACE. Modelspace matrix have only a 16 bits translation and doesn't have the 64 factor described previously.