Hi,

in jme2 the <matrix> element is interpreted without the scale component. When importing sketchup models scaled elements appear with wrong size. Using the instructions about matrix decomposition from http://ziggyware.com/readarticle.php?article_id=15 the following patch solves this issue:

===================================================================

@@ -60,6 +60,7 @@

import com.jme.light.PointLight;

import com.jme.light.SpotLight;

import com.jme.math.FastMath;

+import com.jme.math.Matrix3f;

import com.jme.math.Matrix4f;

import com.jme.math.Quaternion;

import com.jme.math.Vector2f;

@@ -3965,9 +4315,34 @@

}

tm.set(data, true); // collada matrices are in row order.

child.setLocalTranslation(tm.toTranslationVector());

-            Quaternion q = tm.toRotationQuat();

-            q.normalize();

+            // find scale

+            Vector3f vCol1 = new Vector3f(tm.m00, tm.m10, tm.m20);

+            Vector3f vCol2 = new Vector3f(tm.m01, tm.m11, tm.m21);

+            Vector3f vCol3 = new Vector3f(tm.m02, tm.m12, tm.m22);

+            float scaleX = vCol1.length();

+            float scaleY = vCol2.length();

+            float scaleZ = vCol3.length();

+            child.setLocalScale(new Vector3f(scaleX, scaleY, scaleZ));

+            Matrix3f rm = new Matrix3f();

+            rm.m00 = tm.m00 / scaleX;

+            rm.m10 = tm.m10 / scaleX;

+            rm.m20 = tm.m20 / scaleX;

+

+            rm.m01 = tm.m01 / scaleY;

+            rm.m11 = tm.m11 / scaleY;

+            rm.m21 = tm.m21 / scaleY;

+

+            rm.m02 = tm.m02 / scaleZ;

+            rm.m12 = tm.m12 / scaleZ;

+            rm.m22 = tm.m22 / scaleZ;

+            Quaternion q = new Quaternion().fromRotationMatrix(rm);

+            //Quaternion q = tm.toRotationQuat();

+            //float scale = FastMath.sqrt(q.norm());

+            //System.out.println(scale);

+            //q.normalize();

child.setLocalRotation(q);

}

if (xmlNode.hasscale()) {

Vector3f scale = new Vector3f();