Coordinate System to support OBJ models

[ThomasOrtner]

There is a conversation thread with Christian Tate link. Objs downloadable from sketchfab on Christians account do have the wrong orientation.

Further, xyz axis do align with north east and up and should be used for dip and strike and other measurements.

yellow expected behavior

best what we can do currently in PRo3D ... yx horizontal plane is vertical

add new coordinate system to support roverframe northed data with y as Up and z as North

[haraldsteinlechner]

isn't this just basis change for objects (pretransform?)

[cdt59]

Hi @haraldsteinlechner. A pre-transform would work if PRo3D could recognize or be told that it is importing an OBJ in this particular reference frame. Let me know if there is any information I can provide. Thanks!

[ThomasOrtner]

So I tried out a pretransformation which can be activated by a checkbox similar to 'flipZ' called 'sketchFab'. I had a look at some other exports and tried out Sol 195 and Sol 204

Should we build sth. more general, where you can pick the up vector and then apply the transformations?

[ThomasOrtner]

I did find another variant, Sol51. So I do think we need a UI and transformations that are capable of the following things:

• specifying any signed axis as up vector
• flipping of any axis
• switching left and right handed

[haraldsteinlechner]

basically the arguments to FromCoordinateFrame - question is how well this needs to be prepared for users - could this be just a choice of presets?

[ThomasOrtner]

maybe we can agree on one or two presets, A full blown import dialog or UI would be quite costly to implement and difficult to use.

[ThomasOrtner]

current implemention of sketchFab flag switches y and z axis und rotates then around z by -90.0 degrees

[cdt59]

I did find another variant, Sol51. So I do think we need a UI and transformations that are capable of the following things:

• specifying any signed axis as up vector
• flipping of any axis
• switching left and right handed

Hi @ThomasOrtner, note that the Sol51 model and many of the early models are in a different transformation. Only the last few models are in the X-Y-Z=East-North-Up orientation that I think makes the most sense going forward. Sorry for the confusion, but the scalebar is there to specify North, and the text on the scalebar is normal to the Up vector.

If a strategy works for the Sol 204 model, then that is sufficient.

[cdt59]

In light of the above, I don't think we should call this special transformation "sketchfab" or "cdt". How about "ENU" to describe East-North-Up for X-Y-Z?

[cdt59]

current implemention of sketchFab flag switches y and z axis und rotates then around z by -90.0 degrees

Hi @ThomasOrtner and @haraldsteinlechner,

There is something odd happening with the left-handed XYZ bars that PRo3D displays. The X-axis crossed with the Y-axis should give Z-axis (per the right-hand rule), but these bars give X x Y = -Z. This really confuses me! I could believe it if the "flip Z" button turns Z into -Z and just forgets to actually turn around the blue vector.

Ok, so, the blue vector might or might not be accurate. What is important is the following:

• X=East
• Y=North
• the model is not mirrored (what the "flip Z" button fixes)

If the above is satisfied, then the Z-axis can be up or down and it shouldn't matter.

Once we get this settled, then I can start measuring Strike&Dip and confirming that we get the values expected.

I appreciate your work on this, PRo3D will be an incredible tool once it works for these models. Thanks a lot!

[ThomasOrtner]

The coordinate system and the data finally appear correct. Please verify with this . To active the changes one must be in the XZY coordinate system and apply the sketchFab trafo to the surface by ticking the respective box as shown in the screenshot above.

[cdt59]

Hi Thomas, Thank you. I checked the strike and dip tool and its angles look right. I imported a cube that was rotated by 30-degrees along each of the three Euler axes. I expected to get slopes between 30 and 60 degrees and bearings approximately120 degrees apart. That is what I got! I was confused by the difference between Bearing and Slope in the Measurements table versus Dipping Angle and Orientation in the Dip&Strike table. Do you know what bearing and slope are in this context? CT [image: image.png]

[cdt59]

Nevermind, I read the manual about the bearing and slope.

[ThomasOrtner]

I hope you values make sense now, these values and namings are often discussed and not 100% clear. I am open to suggestions, Steve and Rob also have some comments on that I suppose.

[haraldsteinlechner]

what is the status of this one @ThomasOrtner