As a relative newbie I'm still baffled by how the Transform coordinates work in Houdini. I get all kinds of different results depending if I transform the parent first and then the child, or vice versa. Or if I type in a value, or use the transform gizmo or match the transform from one object to the other etc.

For instance:

When I translate an object 5 units only in the X axis, that has a parent that is translated and rotated, I get parent coordinates in the input fields while moving. But when I let go of the gizmo, I get some kind of coordinates in the input fields that are not world and neither parent related it seems.
So even though the child has moved only 5 units in the X axis relative to its parent, I get a value in the Y and Z axis as well.

When I then reset the Y and Z by typing zero in the input fields, the position does not change because somehow parent coordinates are at play again ?

Also when I type in 5 units in the X axis I visually get the same result in the viewport as translating it 5 units, because parent coordinates are at play, but with a different result in the input fields then when I use the transform gizmo.

I could give more examples of strange results like this, but I hope this explains my confusion well enough.

I've been googling for an explanation, and trying to adjust all kinds of settings to get some kind of intuitive behavior, but I can't seem to wrap my tiny brain around the logic.
I've used many 3D programs over the years, and most work basically the same, but somehow Houdini seems use a completely different way to handle transforms.

Maybe someone can shed some light on this Houdini magic, or point me to some documentation ?


Thanks,

R.

Hey toonafish,
Could it be that this values you get on y and z are just super tiny? I think this is just a Houdini thing like it sometimes shows you are 0.9999999 for a 1 or a 0.499999 for a 0.5. this tiny values represent 0 and can be ignored. I would like them to change that too. It's just confusing and adds no benefit whatsoever.

Cheers
CYTE

Hey,

Here's a screencap that might explain it better.

The green cube is the parent that is rotated in XYZ, the red one the child, and the blue in the last image is an unparented object.

In the first image you see the coordinates in the X input field while I'm still moving the child. This is what I expect to get, translation coordinates relative to the parent.

But in the second image you see what I get when I let go of the transform gizmo. Plus I get values in the rotation as well, even though I did not rotate the object.

At first I thought these are world coordinates, but when I copy the translate values and paste them to a unparented cube I get a completely different position, as you can see in the third image with the blue cube.

And when I reset the Y and Z transform fields of the child, the object does not change position, and I get the same result as in the first image. The weird thing is, the same goes for the blue unparented cube, because it actually is only translated in the X axis, the other values are just there to make the whole thing confusing ?

R.

Like I said these values are super tiny and should be thought of as 0. Look at them its 5.0555(whatever)e-14. These are values with 14 zeros behind the dot.

Cheers
CYTE

Ah I see, thanks.

I did not know that's what the e-14 means. I thought it was the amount of numbers not visible after the dot.

Still strange I get rotation values, even though I did not rotate the child.

Is there a way to disable this ? It's just easier to read when these ridiculously tiny values are rounded off to a zero.

R.

You can always right-click on any channel and choose to lock it. Also, if you l-click on the label itself, you will toggle the rounded values and non-rounded.