I am creating a sequence of steps that will import an .eps file (since no .svg import exists) , extrude it into one plane, then inset to create an edge around the extruded objects.

The problem is when removing holes - the bridge gets inset as well (as seen on picture).

I tried moving the hole removal later into the chain but then the final result is incorrect.

Any recommendation for a workaround?

Was thinking of selecting the bridge somehow and ignoring that group when doing the inset, but open to all suggestions.

would you mind uploading your scene with the eps ? might have a workaround for you.

Entagma shows here .svg export, but it's a hard way i think
https://vimeo.com/282137418 [vimeo.com]

later “import not export *beating head upon a wall*”

you're probably just missing a coupla simple nodes…and it doesn't have to be caused by Hole, you could easily get an eps with the fault as is. I've tested with a sample eps file and as you can see on the left…it is the fault that you are experiencing while on the right is the fix

it's quite simple…just a coupla nodes…but I'd like to test the workflow on your actual file with your eps.

so upload it.

vusta
would you mind uploading your scene with the eps ? might have a workaround for you.

I've updated my main topic with both the .eps and .hipnc files.

here you are…now in my test, as said all I needed was a coupla nodes, Fuse Facet and maybe a Clean…but boy oh boy…yours refused to play nicely…so had to add in a remesh. Way more finicky than I thought.

you don't need all those groups just use output Front/Back groups in the polyextrude so you
can easily refer to them further down the line.

Can you work with the remesh version ? Or is it a definite no no?

btw, I blew up your geo to 100x cos working on tiny scale is not good…imagine you deal with numbers like 0.00001…just work with decent numbers…then scale it down to whatever once you're done.

Hi,

to avoid remesh you can also remove the points, which have two neighbours and at least two vertices.

Aizatulin
Hi,

to avoid remesh you can also remove the points, which have two neighbours and at least two vertices.

yeah I used facet to remove inline points for this purpose…but for whatever reason, this ankh refused to play nice.

nice but vusta i'm lost a bit - could you tell why merge node is called remesh1 does it gain some functionality somehow :?
seems everything is alright if to uncheck Fix Overlaps in clean :?

I knew there has to be a neater way as I found in my initial test, here's one without the remesh
much nicer geo.

what I found was…the Fuse…do not trust it, it does not always fuse even if you crank up the threshold…yet the Clean/Consolidate points would do the job properly.

(if you go to the clean just after convert, toggle Consolidate Points on off, you'll see the ‘offending line’ splits into more points…that should have been fused had Fuse done its job)

ryuku
nice but vusta i'm lost a bit - could you tell why merge node is called remesh1 does it gain some functionality somehow :?
seems everything is alright if to uncheck Fix Overlaps in clean :?

not sure I follow you, remesh is just as is…it wasn't renamed from anything.

not sure I follow you, remesh is just as is…it wasn't renamed from anything.

ohh i've meant this, but don't bother

ryuke are you using the most recent version of houdini?

Because for me remesh1 is a remesh node and not a merge node (18.0.287).

Hey, Aizatulin, ty for pointing you're right
i was under .346 just updated the build
funny why my Houdini saw remesh as merge now it's alright
tbh i was interested to see your workflow how you would go about?
you've meant using smth like this and to fuse after or how :?

here is a modification of the previous examples. I'll check, if the neighbour count is 2 (point is a border bound) and if it has two vertices or more, which means, that the point has more than one prim or it has more than one (vertex)reference on the same prim (which is the case here).

It is like vusta said (do not trust the fuse sop).

If you look into my example, there is switch node. The results seems to be the same, but if you check the point number, they are different.

hmmm…I take it all back, sorry about that Fuse…

If I use a Fuse instead of the Clean after the Convert, it works too.

Aizatulin


here is a modification of the previous examples. I'll check, if the neighbour count is 2 (point is a border bound) and if it has two vertices or more, which means, that the point has more than one prim or it has more than one (vertex)reference on the same prim (which is the case here).

It is like vusta said (do not trust the fuse sop).

If you look into my example, there is switch node. The results seems to be the same, but if you check the point number, they are different.

possibly because Clean works on a different threshold.
If I change your Fuse to 0.001 then it's the same with Clean…ahh yes…there is a Fuse inside the Clean which uses 0.001

Aizatulin


here is a modification of the previous examples. I'll check, if the neighbour count is 2 (point is a border bound) and if it has two vertices or more, which means, that the point has more than one prim or it has more than one (vertex)reference on the same prim (which is the case here).

It is like vusta said (do not trust the fuse sop).

If you look into my example, there is switch node. The results seems to be the same, but if you check the point number, they are different.

ty so much for the nice explanation and example file and wrangle of course just <3
really interesting that with clean and fuse there is different point count if not to increase fuse to say 0.001 and with fuse=0.0001 there is one broken edge at the left side… clean worked cleaner
ty

ryuku

Aizatulin


here is a modification of the previous examples. I'll check, if the neighbour count is 2 (point is a border bound) and if it has two vertices or more, which means, that the point has more than one prim or it has more than one (vertex)reference on the same prim (which is the case here).

It is like vusta said (do not trust the fuse sop).

If you look into my example, there is switch node. The results seems to be the same, but if you check the point number, they are different.

ty so much for the nice explanation and example file and wrangle of course just <3
really interesting that with clean and fuse there is different point count if not to increase fuse to say 0.001 and with fuse=0.0001 there is one broken edge at the left side… clean worked cleaner
ty

well no….as explained in my prev post…inside the clean…it's using a fuse anyway (at the top), just that the default fuse is at 0.001…while the Manual Fuse was at 0.0001 hence less fusing at smaller threshold.

ryuku

Aizatulin


here is a modification of the previous examples. I'll check, if the neighbour count is 2 (point is a border bound) and if it has two vertices or more, which means, that the point has more than one prim or it has more than one (vertex)reference on the same prim (which is the case here).

It is like vusta said (do not trust the fuse sop).

If you look into my example, there is switch node. The results seems to be the same, but if you check the point number, they are different.

ty so much for the nice explanation and example file and wrangle of course just <3
really interesting that with clean and fuse there is different point count if not to increase fuse to say 0.001 and with fuse=0.0001 there is one broken edge at the left side… clean worked cleaner
ty

sure

vusta

Aizatulin


here is a modification of the previous examples. I'll check, if the neighbour count is 2 (point is a border bound) and if it has two vertices or more, which means, that the point has more than one prim or it has more than one (vertex)reference on the same prim (which is the case here).

It is like vusta said (do not trust the fuse sop).

If you look into my example, there is switch node. The results seems to be the same, but if you check the point number, they are different.

possibly because Clean works on a different threshold.
If I change your Fuse to 0.001 then it's the same with Clean…ahh yes…there is a Fuse inside the Clean which uses 0.001

It looks like, that the Fuse Node in the Clean Node is the old (deprecated) version. Probably they changed the default value for the new one. I wasn't even aware of it, that the Clean Node is editable.

P.S. but anyway the different result was due to the threshold and not caused by the different versions.

P.P.S. No I was wrong, they haven't changed the default value… .