Page MenuHome

Unwrapping doesn't take into account mirror modifier geometry
Closed, InvalidPublic

Description

System Information
Win8 64 bit, 560Ti

Blender Version
Broken: 2.77

Short description of error
Unwrapping doesn't take into account mirror modifier geometry. What this means is that unwrapping only takes into account the "real" geometry, so for example if you unwrap a sphere with a mirror modifier you will unwrap one half of the sphere which gives severe UV distortion at the mirror axis. This of course kind of makes sense because the geometry isn't real (and you can only select half of the mesh), but it would be nice if Blender could act as if the mirrored geometry was there so you could get a better result. After all with the mirror modifier you're probably looking for an unwrap that also has mirrored UVs, so the less distortion the better.
Here's an example, mirrored to the left:

Exact steps for others to reproduce the error
Create two UV spheres.
Delete half of one UV sphere and add a mirror modifier.
Add a seam parallel to the X axis on both spheres.
Unwrap. They will look different which is unwanted behavior.

Details

Type
Bug

Event Timeline

Henrik Berglund (cyaoeu) updated the task description. (Show Details)
Henrik Berglund (cyaoeu) raised the priority of this task from to Needs Triage by Developer.
Henrik Berglund (cyaoeu) set Type to Bug.

Mirror modifier have U and V option to create UVs for mirrored geometry according center of UVspace.
So, you can use an non symmetrical texture on a mesh using mirror modifier.

But you have to set your UVIslands in UVspace according to this logic.

You are requesting another way to mirror.
Instead a global setting, you are asking same functionnality per island.

It would imply definition of edges as aligned to symmetrical axe.

Anyways, it is not a bug but a feature request.

I'm aware that UVs can be mirrored along the U axis which is nice and the reason why I created this bug report. It's probably possible to make a complete model while using the mirror modifier and I'm looking for ways to make this workflow better. And yes, UV islands would have to be manually edited after unwrapping to get a good mirrored result. That's not the point though. It would be better if the mirrored geometry was taken into account when unwrapping because it would give you a closer result to what you really want. It's not really another way to mirror.

It's more obvious when unwrapping a character mesh. For the mirrored mesh you would get an unwrap that is distorted (curved) at x 0 because that is not the "middle" of the mesh since it's mirrored and the unwrap only sees half of the mesh. For the same mesh with the mirror modifier applied you get a perfectly straight line at x 0 because that is truly the middle and this is what you want when you later want to texture the mesh.

Whether this is a bug report or a feature request is debatable. Sure, the unwrapping itself works so it's not bugged in that sense. However the currect behavior is not what you want when unwrapping meshes that are mirrored so this is the reason I filed a bug report.

I just tried to clarify things for devs. Your report could be mistaken as a bug of U and V options of the modifier.

I don't disagree that it would be better than current situation.

By the way, this year, there is a GSOC dedicated to UVtools.
Did you try to contact its dev ?
http://blenderartists.org/forum/showthread.php?397599-GSOC-2016-UV-Tools

It is probably out of scope of its GSOC. Maybe he will look at it after.

Antony Riakiotakis (psy-fi) claimed this task.

I'm afraid I'll have to agree that this is indeed not a bug - it works as intended so I will be closing this.

Unwrapping generally does not take modified geometry into account. It's possible to make it do so - as in the case of the subsurf uv unwrapping feature for instance - but this is the whole point: This is a feature request that needs an exception for the mirror modifier.

I've experimented in the past with mirrored unwrapping. The basic idea is simple but aligning to an axis is more difficult and requires more custom code than "take the mirrored mesh into account".