[Bf-committers] Vertex Pinning

Mon Mar 24 03:53:30 CET 2014

Inline:

> Date: Mon, 24 Mar 2014 12:42:30 +1100
> From: ideasman42 at gmail.com
> To: bf-committers at blender.org
> Subject: Re: [Bf-committers] Vertex Pinning
> 
> > Hiding works fine in some cases, but the disadvantage here is that you lose the overview of those vertices (+ adjacent edges/ faces) and thus the final result. A quick example that comes to mind could be a low/ mid-poly skirt. You want to deform the skirt to have some pleads and creases using proportional editing, but want to leave the band at the top as is (let's assume it's fit to a character). Hiding the vertices works, but you won't be able to see the full skirt as you're pushing and pulling around the model, making it harder to estimate what looks good.
> 
> I've linked to patch D424, if all you want to do is see geometry for
> hidden faces this could be a simpler alternative to pinning.

I will try the patch tomorrow, but initially I feel like this is rather a clunky solution; especially with the recent focus on ease-of-use and clear user interaction.

Hiding and pinning are semantically very different i.m.h.o.

I think it would be analogous to an Object mode situation where you would have to hide an object and run a "show all objects" operator, rather than just locking the position.

I feel hiding serves a fundamentally different purpose, and it's just currently being (ab)used as a substitute for pinning. Hiding elements currently also excludes them from transforms, but that is not its main function. Calling "hide", followed by "show hidden" just feels a bit makeshift I think.

In addition to that, the ability to still select vertices may be useful with regards to both cursor positioning, as well as for use in determining the operation pivot (in case of median point pivot and such).

I could also imagine a scenario where you want to hide part of a mesh (i.e.: body) to focus in on a specific portion of it (i.e.: face), but still want the ability to pin certain vertices within that portion. "Show all" would reveal back the entire body, and hiding those specific vertices may not be ideal.

Lastly, I could also envision a situation where you would keep vertices pinned for a long period of time, but still want to keep hiding and revealing parts of the mesh while working on it.

> > Another example could be a face for instance, where you're making shapekeys for the eyelids, but it's important the brows never move.
> >
> > I've personally actively missed this feature several times, so it's not really a spur-of-the-moment thing for me. =)
> >
> >> One thing thats still not clear to me is when should tools support this?
> >>
> >> - Should edges between pinned vertices be subdivided?
> >>
> >> - Should assigning vertices to a group skip pinned vertices?
> >>
> >> - Should the rip tool allow pinned vertices to be ripped?
> >>
> >> I think we should be clear about where pinning applies to avoid adding
> >> some feature then getting a steady stream of bug reports when users
> >> notice some tool doesn't respect pinned vertices.
> >>
> >> If we limit pinning to transform + deformation tools then this could
> >> work (grab, rotate... etc + smooth, snap, warp, blend-from-shape)
> >> ...  but not attempt to support this for every tool (don't make this a
> >> second 'hide' option).
> >
> > I agree this should be clear. To me, the real value lies in excluding certain vertices from transformation tools like the ones you mentioned (including blend-from-shape which I haven't touched yet). So a 'no' on Vertex Group assignment skipping pinned vertices from my end. As for subdiving edges, I quite like the way it's handled right now; just move what's not pinned, and leave pinned ones alone. It doesn't feel weird or unintuitive I think.
> >
> > Rip Tool for me sounds like a gray area. On one hand, pinned = pinned = pinned, meaning it can't move. On the other hand though, usability-wise it would be a drag to have to unpin, rip, drag, repin. I'm not feeling too strongly either way, though I think I'm leaning towards not respecting pinning in this case.
> 
> Right - its not clear which way to go here. you also get into some
> problem areas could show up since tools transform immediately after
> executing...
> If you extrude pinned vertices what happens?
> (... ok, making it so duplicated geometry is never pinned will work,
> but go over _every_ tool that does this and consider how it should
> work).

I agree it would cause some corner cases like this, but I think the benefits would far outweigh the hassle.

Coming back to my Object mode analogy btw, this same problem exists with position-locked objects. Duplicating the object will render it unable to move. If this makes it in it might be good to create some unison there?

> > One other case I considered was during sculpting, but for the sake of scope and due to the fact that there already is a Mask brush I would probably leave it at Edit mode mesh editing only.
> 
> Agree, this is for editmode only.
> 
> > Same for deformation modifiers and the like; these can already be controlled better by vertex groups and would just lead to clutter and (user) confusion i.m.h.o.
> 
> Modifiers are definitely out of the question - pinning is like
> selection, editing feature only, not mesh data to go through the
> modifier stack.

Agreed and agreed. Same page FTW!

> >> One more thing...
> >> The possability to have laplacian deform in editmode (to work like
> >> pinned UV's), would be a really nice feature.
> >
> > I definitely thought about functionality this, but decided it would probably be a rarely used feature and hard to code (I was originally imagining this with regards to rotate, scale, to sphere and such). I agree Laplacian Smooth might be a good (isolated) candidate for this though. On the other hand, where would we draw the line? Should regular Smooth respect pinning as well? According to my earlier definition, yes. But how this would look/ act in practice I'm not too sure about.
> 
> If we define only transform tools respect pin (and similar tools which
> just move vertices) - this seems manageable.

Alright.

> >> ----
> >>
> >> Regarding internals...
> >>
> >> I'm more concerned long term with tools having to check both
> >> `BM_ELEM_HIDDEN` and some `BM_ELEM_PIN` flag.
> >>
> >> Though some inline function could work...
> >>
> >> `BM_elem_flag_test(ele, BM_ELEM_HIDDEN)`
> >>
> >> ... could be replaced with a macro/fincion...
> >>
> >> ` BM_elem_is_editable(ele)`
> >> ---
> >> inline bool BM_elem_is_editable(BMElem *ele)
> >> {
> >>     char testflag = (BM_ELEM_HIDDEN | BM_ELEM_PIN);
> >>     return (BM_elem_flag_test(ele, testflag) == testflag);
> >> }
> >
> > Sounds good to me! Definitely a lot cleaner and more clear. =)
> >
> >
> > Regardless of if this will make it into Blender, I would really like this feature working for myself. The main problems I'm facing now that I can't quite solve though are the undo/ redo situation I described and the BM_ELEM_PINNED flag getting cleared after toggling out and back into Edit mode. Any chance you could give me some pointers on these issues? I for the life of me can't seem to find where the flag is being cleared.
> 
> We're run out of flags here, but Im not so worried about getting the
> patch working just now, first figure out if we even want to have this
> in blender.

I already set the flags back on my local copy. Regardless, like I said I would really like this feature working on my local Blender. =) I've made some adjustments in the past that just made my modeling life a bit easier, so "worst-case scenario" I would just have a version of pinning working inside my own Blender builds only, and only to the degree that I need it. If I can get that working I would be one very happy camper already!

-Patrick

> > Cheers and thanks again,
> > Patrick
> >
> >> On Mon, Mar 24, 2014 at 3:28 AM, patrick boelens <p_boelens at msn.com> wrote:
> >> > Hi Antony,
> >> >
> >> > I'm not at all familiar with Blender's codebase, so could you give me a fix proposal? 1 << 6 is unused; is it safe to push BM_ELEM_DRAW and BL_ELEM_INTERNAL_TAG back one, or could that cause problems?
> >> >
> >> > Cheers,
> >> > Patrick
> >> >
> >> >> Date: Sun, 23 Mar 2014 10:13:13 +0200
> >> >> From: kalast at gmail.com
> >> >> To: bf-committers at blender.org
> >> >> Subject: Re: [Bf-committers] Vertex Pinning
> >> >>
> >> >> One basic flaw:
> >> >>
> >> >> hflag is a character so you are causing an overflow with
> >> >>
> >> >> BM_ELEM_INTERNAL_TAG or it may simply be ignored.
> >> >> _______________________________________________
> >> >> Bf-committers mailing list
> >> >> Bf-committers at blender.org
> >> >> http://lists.blender.org/mailman/listinfo/bf-committers
> >> >
> >> > _______________________________________________
> >> > Bf-committers mailing list
> >> > Bf-committers at blender.org
> >> > http://lists.blender.org/mailman/listinfo/bf-committers
> >>
> >>
> >>
> >> --
> >> - Campbell
> >> _______________________________________________
> >> Bf-committers mailing list
> >> Bf-committers at blender.org
> >> http://lists.blender.org/mailman/listinfo/bf-committers
> >
> > _______________________________________________
> > Bf-committers mailing list
> > Bf-committers at blender.org
> > http://lists.blender.org/mailman/listinfo/bf-committers
> 
> 
> 
> -- 
> - Campbell
> _______________________________________________
> Bf-committers mailing list
> Bf-committers at blender.org
> http://lists.blender.org/mailman/listinfo/bf-committers