[Plugin] Manifold v2.2

driven

this should upload.jpg
something screwy going on with uploads, sorry it makes thing disjointed

TIG

I get no problems... but can you PM me that model asap, so that I have exactly the same starting point...

The 'no problems found' report is on any geometry that hasn't already be relegated to 'red'... looks like everything is removed BEFORE we do the health check - it's not a group inside a group is it ?

driven

TIG,

they were groups inside a group, and got a pass, (they are all manifold) it only zaps them if you say YES,
the same thing happened after I BoolTool'd all 3 groups into 1 group, initial pass (as it should), but zapped if you choose YES, I only got the RED on that one drawing, I'll re run the first set-up and post the if no red is shown again.

I have just set a Bool'd group run that Passed initial and didn't blitz the drawing when I hit YES

It's on a pair of torus + geodasic sphere bool group, and it is running.

original set ungrouped, but not booled

and others work so smoothly

TIG

Here's v1.9 http://forums.sketchucation.com/viewtopic.php?p=218691#p218691

It now traps for 'No Geometry' inside the Group...

driven

NO Zapping with 1.9... is it possible for the health check to report 'NO geometry to Check because the selection contains sub-groups'?

I'll try some other breakers I've been working on...

cheers john

driven

TIG,
I've got a couple of test on the go, I'll report on them when they're cooked.

some additional ?'s

which should be faster for manifold, simply continuing after fail or a manual repair than re-test continue

would it be possible to choose bothquad's and tri's as two separate groups, if you have two different post destinations for the object?

or would EraseCoplanar on a copy of a Tri'd group give identical mesh results?

driven

TIG- found the problem with the Bool'd groups -BT can create a group inside a group when bool-ing more than two groups, which I should have spotted in Outliner.

When manifold highlighted RED and said 'NO' I simply exploded the extra group and answered my earlier ?, it was much faster on the bool'd group and I assume would be the similar on a manually repaired group.

Note for users: The manual option also gives more control over the mesh shape at the torus/box intersection.

Basically TIG you've done a fantastic job on this, thank you

I'm convinced you, Chuck, and I won't be the only end-user's for this, it has soo many uses.

for now I'll keep putting up images that help or hinder, and get back to some bent text testing....

do you think it's ready for a wider audience yet, I'm planing on trying to do a tutorial when your happy?
un-intersected single group initial test picked up the buried group and then 'presto'

ADD: it's what I thought re: tri/quads, just thought I'd ask

TIG

@driven said:

TIG,
I've got a couple of test on the go, I'll report on them when they're cooked.
some additional ?'s
which should be faster for manifold, simply continuing after fail or a manual repair than re-test continue
would it be possible to choose bothquad's and tri's as two separate groups, if you have two different post destinations for the object?
or would EraseCoplanar on a copy of a Tri'd group give identical mesh results?

Manual fix versus Auto-fix depends on what needs fixing - making it right in the first place is the best bet
If you choose NOT to triangulate but then want a triangulated version then copy it and simply select all faces and run my [+tt's] tool triangulate_faces and you'll have the same result - uses the same code...

driven

Even Likes Curves, but they will slow it down.

seems a lot quicker than before on this sort of thing

TIG

Here's v2.0 http://forums.sketchucation.com/viewtopic.php?p=218691#p218691

Glitch in health-check if very small edges in faces fixed...

thomthom

v2 ? is this a major or minor update?

TIG

@thomthom said:

v2 ? is this a major or minor update?

The tweaks were relatively minor - BUT as I am unlikely to make many more changes in the short-term I thought I'd mark it with '2.0'...

pilou

I have missed this new true gem!

pilou

When I use Manifold I obtain that ok
But When I kill or move the group red repair
I have the original copy with killed internal faces Yes ok , but always segments on the faces!
Does it normal? A simple box with 6 faces must be result ? No clean selection?

TIG

@unknownuser said:

When I use Manifold I obtain that ok
But When I kill or move the group red repair
I have the original copy with killed internal faces Yes ok , but always segments on the faces!
Does it normal? A simple box with 6 faces must be result ? No clean selection?
[attachment=0:2jmg80n7]bug.png[/attachment:2jmg80n7]

This is correct.
The internal faces are removed by any edges they has forming part of the external envelope are kept.
If you want rid of then use one of the tools that removes coplanar edges...
It could seek to remove coplanar edges too ?

TIG

Here's v2.1
http://forums.sketchucation.com/viewtopic.php?p=218691#p218691

It now has the option to remove coplanar edges before triangulation, this will minimize the number of faces -useful if for example internal faces have been removed and their old edges are not needed to keep the external 'manifold' skin intact...

pilou

Cool now we have possibility or not to "clean" the model in the same time

pilou

2 volumes
copy move

for each selected (2 volumes) Intersect Selected
Group

for each "group" apply Manifold
with same answers : Yes for the first question (of course)
and No for the others questions

Curious result is not the same!
Is that normal?

test.skp

driven

@Pilou,

tested your .skp on mac and got two different results? first time I opened SU from your download file, 2nd time I opened it from inside SU, but that was the only thing I changed, something odd??
double check manifold.jpeg

TIG

Manifold has to start somewhere in its checking - the first entity it encounters might not always be the same so you get differing results if there's a big mess...

Manifold was written to do specific job - check that a group is 'manifold' and if not try to fix it - it is so that when you export your model - e.g. to 3D-printer software - it will work. As it says at the start 'give us break' and try and at least offer it a set of geometry you hope might be manifold - it is NOT a boolean tool to merge volumes - it tries to decide which is the face to keep but sometimes removing or healing something changes the form - hence the colored 'report' group.

Before using 'manifold' try manually to get rid of at least a few known internal faces etc [x-ray mode + hide is good for this] - then the processing is quicker and more consistent...