[Bf-modeling] inset options

Howard Trickey howard.trickey at gmail.com
Thu Aug 27 15:49:38 CEST 2015 

Offline chat with Campbell has convinced me that the current options have
utility. To record here:

- offset_even can produce long internal 'spikes' from reflex vertices, so
for, say, organic models the annoyance of those spikes is greater than the
annoyance of not having an even thickness inset; hence the desire to be
able to turn off offset_even sometimes
- with large regions or polygons that have fine details (bumps, serifs,
etc.) along their outline, using offset_relative can sometimes avoid help
self intersections in those fine detail areas while allowing larger insets
along the long edges; so offset_relative can be useful in more than just
the case of insetting many independent faces of differing scales
- as many have said, the boundary option has a number of uses

There's still some argument that the factor multiplied by for
offset_relative should be normalized somehow, so that the largest thickness
has factor 1; but nobody is complaining about this right now.

So I will leave the options as is.

But it remains the case that to implement the functionality I want  --
handling self collisions, a la straight skeletons (see
http://web.engr.illinois.edu/~jeffe/pubs/pdf/cycles.pdf) -- is a lot harder
if not offset_even and/or if offset_relative is on, so I will, at least for
now, not implement this functionality in those cases.

- Howard


On Thu, Aug 27, 2015 at 9:39 AM Ray Mairlot <madog_theone at hotmail.co.uk>
wrote:

> Just thought I would throw my two cents in and say I tend to use the
> boundary option when insetting faces on a model with a mirror modifier. The
> boundary option allows both the real face and the mirrored face to act as
> one, instead of creating inset faces along the edge of the mirror seam.
> Can’t say I know what other options do, I haven’t found the need to
> investigate them.
>
> Ray.
>
> *From:* Jonathan Williamson <jonathan at cgcookie.com>
> *Sent:* Thursday, August 27, 2015 2:29 PM
> *To:* bf-modeling at blender.org
> *Subject:* Re: [Bf-modeling] inset options
>
>
>> I don't usually touch the Offset controls but quite regularly toggle the
>> boundary option. It can be very useful to quickly create equidistant edge
>> loops for example.
>
> Same here. I toggle Boundary all the time, but very seldom adjust the
> Offset Even or Offset Relative.
>
> Jonathan Williamson
> http://cgcookie.com
>
> On Thu, Aug 27, 2015 at 6:29 AM, Howard Trickey <howard.trickey at gmail.com>
> wrote:
>
>> Well I've answered my own questions with a bit more searching and trying
>> things out.
>>
>> It seems that 'offset relative' is useful if one is doing a lot of
>> individual insets all at once, and the polygons you are insetting vary in
>> size from large to small. You might like the thickness of the inset to
>> scale down as the polygon scales down. I saw a discussion about making
>> shells where this was true, for instance.
>>
>> And I do see that Boundary can make a useful difference when insetting a
>> region, so now I see the point of that one.
>>
>> I do question the method used for 'offset relative', however. What it
>> does now is multiply the thickness at each corner by the average length of
>> the two adjacent sides. For rectangles this gives an even inset, but for
>> polygons with differing lengths and angles at each corner, the result is
>> that the inset shape does not match the shape of the original (that is,
>> there is not a constant thickness between the original edges and the inset
>> ones).  Is this desired and useful? I would have thought that most people
>> would want and expect the 'even thickness' property. Maybe a better way to
>> calculate the factor for offset relative would be to multiply the thickness
>> by the average edge length of the whole polgyon? That would lead to an even
>> offset. Or, even better, multiply by the ratio of the average edge length
>> of the whole polygon to the max average edge length over all polygons in
>> the selection? That way, the user-specified 'thickness' value would have an
>> intuitive meaning (the inset thickness for the biggest polygon) whereas
>> right now you have to specify a really small thickness if your polygon
>> edges happen to have a large scale -- only the relative change in thickness
>> makes any sense.
>>
>> Also, the current ability to specify 'offset relative' and 'offset even'
>> independently of each other seems a bit strange. Don't people always want
>> 'offset even'? (And thus, why have the option). With 'offset relative'
>> being an added feature on top of that to get the 'polygon scale' scaling of
>> the thickness, but remaining even?
>>
>> Marc: your question about UV interpolation for Depth. I think you mean
>> this case: if you give a non-zero depth but a zero thickness, so that the
>> effect of the inset is the same as extrude region, then the walls of the
>> extrusion get mapped into zero-area rectangles (they appear to be lines) in
>> the original UV map. And you'd like some non-zero area there to play with.
>> Extrude has the same problem.  The issue here is that it is not clear what
>> math to use to get this, because then the top face(s) will not get the
>> ideal UV interpolation compared to the original UV map -- some kind of
>> artificially made-up shrink factor has to be applied to those faces in the
>> UV map in order to make room for the walls to have non-zero area.  What
>> shrink factor should be used?
>>
>>
>>
>> On Thu, Aug 27, 2015 at 5:20 AM Marc Dion <marcdion1974 at gmail.com> wrote:
>>
>>> Correction, I meant "Depth" when I said "Offset" in the previous post.
>>>
>>> On Thu, Aug 27, 2015 at 2:19 AM, Marc Dion <marcdion1974 at gmail.com>
>>> wrote:
>>>
>>>> To me, those changes seem good.  I feel like they all produce the same
>>>> results with only the slider value being at a different place depending on
>>>> the choice made.  Those extra options seem like redundant clutter.
>>>>
>>>> ----
>>>> In addition to the changes you mentioned, would you be willing to put
>>>> some thought into adding UV interpolation for the Offset option?
>>>>
>>>> When using UV's/Inset, it's almost certain a person would like to see
>>>> some area of the UV's assigned to any new geometry since not having UV
>>>> space assigned to new faces does tend to cause baked textures to fail at
>>>> what they do.  Same for texture painting.
>>>>
>>>>
>>>>
>>>> On Wed, Aug 26, 2015 at 6:57 PM, metalliandy <
>>>> metalliandy666 at googlemail.com> wrote:
>>>>
>>>>> Hey Howard,
>>>>>
>>>>> I don't usually touch the Offset controls but quite regularly toggle
>>>>> the boundary option. It can be very useful to quickly create equidistant
>>>>> edge loops for example.
>>>>>
>>>>> As a side note I think Edge Rail would benefit from being on by
>>>>> default as having it off can cause some some subd smoothing issues and
>>>>> somewhat messier geometry.
>>>>>
>>>>> Cheers,
>>>>>
>>>>> -Andy
>>>>>
>>>>>
>>>>>
>>>>> On 27/08/2015 02:07, Howard Trickey wrote:
>>>>>
>>>>> I'm working on a possible change to inset (optional corrections for
>>>>> self intersections/crossings) and was wondering about several options in
>>>>> the existing tool.
>>>>>
>>>>> Boundary - does anyone ever turn this off?
>>>>> Offset Even - does anyone ever turn this off?
>>>>> Offset Relative - does anyone ever turn this on?
>>>>>
>>>>> In particular, 'offset relative' makes the geometry of what happens
>>>>> after self intersections much more complicated, so I was wondering if
>>>>> anyone would miss these options if they were gone and if also acted like
>>>>> just 'Offset Even' is on.  (And Boundary; I don't really care about
>>>>> Boundary, but it does seem unlikely to me that people would turn that one
>>>>> off.)
>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Bf-modeling mailing listBf-modeling at blender.orghttp://lists.blender.org/mailman/listinfo/bf-modeling
>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Bf-modeling mailing list
>>>>> Bf-modeling at blender.org
>>>>> http://lists.blender.org/mailman/listinfo/bf-modeling
>>>>>
>>>>>
>>>>
>>>
>>> _______________________________________________
>>> Bf-modeling mailing list
>>> Bf-modeling at blender.org
>>> http://lists.blender.org/mailman/listinfo/bf-modeling
>>>
>>
>> _______________________________________________
>> Bf-modeling mailing list
>> Bf-modeling at blender.org
>> http://lists.blender.org/mailman/listinfo/bf-modeling
>>
>>
>
> ------------------------------
> _______________________________________________
> Bf-modeling mailing list
> Bf-modeling at blender.org
> http://lists.blender.org/mailman/listinfo/bf-modeling
> _______________________________________________
> Bf-modeling mailing list
> Bf-modeling at blender.org
> http://lists.blender.org/mailman/listinfo/bf-modeling
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.blender.org/pipermail/bf-modeling/attachments/20150827/fc7ab72e/attachment.html>

More information about the Bf-modeling mailing list