Okay so like many other people I have asked this question into google and gotten back some VERY complicated answers on how to do this. All I want to do is have the XYZ axis move itself into position relative to the object and it's orientation when in local mode.

Easy right?

Seems not. Every answer relies on hooking, creating an empty .. or doing a bunch of other things I can't remember and every time I have this problem its crippling.

Even in Second Life .. YES SECOND LIFE .. you can just pop into local mode and your axis behaves itself properly. Why the hell is this so damn hard to do in blender.

Perhaps I'm missing something, perhaps I've asked the question wrong (in the 96 different ways I've tried to ask it). Youtube videos on this subject often don't solve the problem and the few that do usually are non verbal and don't explain what the hell they did to reorientate.

So my question is more ... why can't blender have this functionality where in Local mode the axis just orienatates to the geomtry you're editing or moving?

And I'd be eternal grateful if someone can give a simple, comprehensive explanation how to get my axis to align to my geometry and stop me from going insane?!?

As you can see from the pic below .. my axis is fine on the Z .. but X and Y are not being helpful at all. All I wanna know is how do I get this to work properly. Local mode REALLY should automatically reset the axis to follow the object .. don't you think?

Axis Problem

I appreciate any and all answers .. espeically those that sort out this problem. And if any Dev's for blender are reading ... seriously if Second Life can do this .. why not blender?

  • 1
    $\begingroup$ It does. Your confusion stems from your understanding of the orientation of the object. Currently the object has no rotation applied to it (it is in its default orientation), and therefore the axes are aligned with the world. $\endgroup$
    – Sazerac
    Commented May 2, 2019 at 3:25
  • $\begingroup$ How can you tell? $\endgroup$
    – Hetoreyn
    Commented May 2, 2019 at 3:43
  • $\begingroup$ I applied rotation and scale .. the axes are still wrong. $\endgroup$
    – Hetoreyn
    Commented May 2, 2019 at 3:46
  • $\begingroup$ Applying the rotation won't change the axes because there is no rotation to apply. I can see it in the panel to the right of your viewport $\endgroup$
    – Sazerac
    Commented May 2, 2019 at 4:57

2 Answers 2


Create a new transform-orientation based on sub-object (component) geometry, like edges. The Z-axis will point in the direction of the cross-product of any two edges selected.

If you are not familiar with that term, the cross product is the vector that is normal to 2 other vectors. Keep the following in mind as examples:

The Z-axis is the cross-product of the X and Y-axes;

The X-axis is the cross-product of the Y and Z-axes;

the Y-axis is the cross-product of the X and Z-axes;

(You can also look up "cross-product" or "Right Hand Rule" for a way to visualize general cross-product vectors. It is not necessary to know the math to visualize a cross-product)

The cross-product of 2 vectors is perpendicular (in 3d we say "orthogonal" or "normal") to the plane the 2 vectors are in:


Say I wanted to create this fan shape so that it was oriented to the curvature of the bookshelf. The shape is a quarter-circle in the plane that I could have made a few ways, but I did it by rotating duplicates of edge components about this corner (the cursor is placed there and transformations are oriented about the cursor). To do that, I first had to get my transform orientation right. I did that by selecting the 2 orthogonal corner edges shown here, then creating a new transform orientation from that, as shown below:

Create new transform orientation

This is the result:

enter image description here

Notice that the transform orientation is still usable even after returning to object-mode:

enter image description here

Now I was able to rotate edge duplicates about the corner, in a plane that is normal to the bookcase. Rotations are done about the new Z-axis in this orientation. Big-picture:

enter image description here

  • $\begingroup$ I see ... I tried this by selecting a face and trying to make a new transform ... it didn't work .. I take it faces won't work for this .. has to be edges of vertices? $\endgroup$
    – Hetoreyn
    Commented May 2, 2019 at 4:50
  • $\begingroup$ I'm gonna have to re-read your comments again (frankly a lot of it is making me crosseyed .. but I'm trying to keep up with it. So far nothing I've tried makes the tranform behave like yours does. I'd upload a new photo to illustrate but I can't seem to do that in this comment. $\endgroup$
    – Hetoreyn
    Commented May 2, 2019 at 4:54
  • $\begingroup$ I think it didn't work because face normals will provide only enough information for the new Z-orientation, but not the X and Y, which as Sazerac said is based on the object's world orientation when no other info is available. I should also clarify that the the X and Y vectors orient to the segments I selected [presumably] because they are off by exactly 90 degrees. That's what made sense to me before I tried it and it worked =/ $\endgroup$ Commented May 2, 2019 at 4:57
  • $\begingroup$ Hmmm .. so that would mean there's no way to set the axis to just follow the face. This does seem a rather serious handicap .. I mean if there's no way to just have the axis point exactly in the direction of where the objects facing. I'm kinda screwed from now on it seems. $\endgroup$
    – Hetoreyn
    Commented May 2, 2019 at 5:01
  • $\begingroup$ 2.8 does have serious handicaps (for example, extruding vertices and single-face objects are broken compared to 2.79; instancing creates copies in outliner but not the scene, etc). But this isn't Blender's fault, there just isn't enough information to orient X and Y if you just select a single face for a new orientation; that will only inform the Z-axis. Select 2 edges or perhaps 2 faces instead. I also updated my post to include a visual example of cross-product and a zoom-out of what I was doing on my bookcase. $\endgroup$ Commented May 2, 2019 at 5:07

Here we go .. as you can see after setting rotation and scale .. and doing what you suggested (I think) .. the X and Y are still set to their global postion. I'm just not sure what I'm doing wrong.

Axis still off


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .