Seams in world geometry only detectable in normal Hammer

[ruby atlas]

I encountered this issue in goldsrc-style cliffs and for the most part I did some hacky vertex manipulation to seal them, but then it hit me that the Xen maps here have loads of irregular geometry and have loads of seams.
In H++'s preview (2nd image) this seam doesn't exist, but in normal Hammer (3rd image) this seam is displayed, there are plenty more seams in other Xen maps as well. Both unmodified and Slammin 1.6 SP's VBSP compiler (1st image) just compiles the seams in.
Is there a better way to fix it than having to use normal Hammer? Vertex float imprecision is something I don't want to deal with.

[dense plaza]

Um, yes, messy geometry will do that.

Is this a decompiled GoldSrc map? It may be better to clean it up in Trenchbroom first, if so, as decompiled GoldSrc and Quake I maps are really messy, and Trenchbroom doesn't destroy complex geometry on import like Hammer does (MAP and VMF are weird, and don't store vertices; RMF and H++ VMFs do, but they're only for the editor). Ctrl+J in Trenchbroom it your friend; it can turn a set of multiple brushes into one convex brush, regardless of if the source brushes are a convex shape.

The seams not showing up in H++ does seem to be a bug with how it handles generating vertex positions from the plane coordinates, so pinging @polar oar (but I won't be surprised it it ends up being a wontfix).

(also, man, I wish Valve would allow Gearbox to release the source RMFs for the expansions)

(judging by the screenshots, I presume it's a Blue Shift map)

[ruby atlas]

I manually cleaned up all brushes and made them to as close as a source file would probably look like as possible, some of the brushes I recreated from new brushes, though it wouldn't be out of the question for a few of them to still be odd from decompiling in some way.

(Yeah, it's a Blue Shift mappack port for HL:S)

[dense plaza]

Yeah, the reason Source maps decompile as well as they do is because they use the Quake II style of collision where the original convex brush shapes are stored in the compiled BSP; GoldSrc uses Quake I's more rudimentary method where "clipnodes" are generated for each "hull" instead, which is too technical to explain here, but means the brush shapes have to be recreated from scratch by the decompiler, using what it does know about the compiled map.

(So it's actually easier to port an already-compiled GoldSrc map to Source 2 than it is to port the map to Source 1)

[polar oar]

this is correct behavior

its a vbsp bug

slice the brush with clipping tool and the precise vertices will be removed (it may get fixed up actually)

[ruby atlas]

Oh... great it's back again but unfixable, I'm trying to fix this seam but clipping it and then vertex manipulating the problem vertex to the proper position just makes the seam happen again.
Back to destructive vertex manipulation.
thank you for being a real dick VBSP and Hammer

[ruby atlas]

Very Buggy Shit 'Piler...
I try to split it, morph it, anything, it just stays on a brush and the precise formation I want just seems to be cursed to never be achieved.
Is there any fix to VBSP that will help with this seam issue?

[timid wave]

select to the 2 brushes that are meant to sew hit Crtl+ H so everything else is hidden and check you don't have displacement faces touch each other, while you have to 2 brushes displayed you'll have a better view when you do vertex editing ,, also is this face red because you have selected it ? if not then its hammers way of telling you its an invalid face

[ruby atlas]

It's selected by me, it doesn't show as an invalid brush either

Here is the group of detail brushes I'm suffering from. Try using the clipping tool to reveal the seams as ficool said, then try to fix these seams with vertex edit and then use clipping tool again to check.
Replaced the textures with HL2 wood.

[timid wave]

hmm strange

[ruby atlas]

So you're getting the same results, then? Any ideas for this with as less of a compromise to the brushwork as possible?

[timid wave]

i tried with an 8 sided arch but i got visible seams then i think its the angle of the bottom brushes that could be causing it maybe making it from displacements would fix it

[ruby atlas]

Seems I can't make displacements out of the two other brushes making up the side, and even with whatever brushes I have turned into displacements they still seam in the same spots, god damn it
I checked the non-displacement brushes with the clip tool and they didn't display any of their vertices as off from it.

I'd upload images but Discord is not letting me do this here >:\

[dense plaza]

VBSP++ when