I don't think you can load .g1a files separately, do you?

Allegedly you'll need to load a g1m file before. (In Noesis you might also have a lock at Tools\Project G1M.)

There's a floats block with 82 entries, but they don't show up as a skeleton:
.

Well, half count looks better but it's not complete. The second blocks gives 49 sub meshes only (block 1: 133):
.

Well, maybe I should try use half of the counts? (couldn't believe it)
There's a second block (marked blue) which also gives part of the body and more dislocated vertices:
.

VIF-tags again - we had that fun already:

shakotay2 wrote: ↑Sat May 01, 2021 10:31 pm

This time (captain.ghg) I met a strange occurrence that some vertices of each sub mesh don't "join" the body:
.

Thank you! Seems, during my tests, MONOLITH_L_break_break_1.geom.edge0.obj had already been created and I've overlooked it merely.
.

Thanks!
How does one extract the vertices and indices of one single geom.edge, say MONOLITH_L_break_break_1.geom.edge?

When I uncomment //#define DECODE_ONLY
the code doesn't compile.

BlackStarEOP wrote: ↑Wed Aug 09, 2023 12:36 pm I will update the code soon (tm)

That would be extremely cool. (Since I don't know any source code except the original from Sony.)

BlackStarEOP wrote: ↑Tue Aug 08, 2023 10:18 pm, I am now looking at some source code for decompressing the PS Edge geometry!

Sony's edge compression? Was mentioned here (click up arrow):

shakotay2 wrote: ↑Tue Jun 25, 2019 1:56 pm

Hi, welcome back to the forum after a year and 7 months!

I compiled your source; after adjusting the files.push_back() path it created an mtl file.

allWithout.txt is an input file? How should its contents look like?

This mesh format is rather simple. It would be very easy for you to check it yourself using hex2obj or similar programs.
.

Tosyk wrote: ↑Sun Aug 06, 2023 10:44 pm Is that a bird?

Looks like so. Is from the characters folder of human_male_scout, though.

Tosyk wrote: ↑Sun Aug 06, 2023 10:44 pmI would really like to have them with skeletons and weights

yeah, that's what people usually request these days. Sadly my time stock is limited to about 15 minutes per model, usually.

On a quick glance - first submesh only, some superfluous faces:
.

Complex one is also simple:
. H2O file:

0x76 455160
Vb1
24 16
0x1BC856 107797
040000
0x0 255

I didn't found any 48 number except self.inFile.seek(0x48, NOESEEK_ABS) That line is not related. It's a matter of formulas. You'll need to be a coder to get this. The difference between 16 and 8 is 8. Since the original script outputs 48 where it should be 40 the workaround was to subtract 16 inst...