parsing new maps doesn't work properly

[xiv1337]

no i'm not talking about how you need to use vmdl_c, e.g train thinks its visible half way across the map (maybe some meshes are left out idk?)

[AtomicBool]

no i'm not talking about how you need to use vmdl_c, e.g train thinks its visible half way across the map (maybe some meshes are left out idk?)

Does it happens often? I remember some meshes are missing mesh data like boxes in mirage A

[xiv1337]

yes, also the vphys parser doesnt like the other maps (crashes and doesnt generate data)

[AtomicBool]

yes, also the vphys parser doesnt like the other maps (crashes and doesnt generate data)

So it crashes for train? I haven't play it for months but maybe it's happening to new maps?
Or it crashes for most maps for your laptop?

[xiv1337]

it doesn't crash on vphys_c but when parsing vmdl it does and no im not on a laptop and my pc is good

[AtomicBool]

it doesn't crash on vphys_c but when parsing vmdl it does and no im not on a laptop and my pc is good

This repo doesn't support VMDL format. There was a post about how to change source2viewer's code to export vphys from vmdl, simple copy-paste also works.
However, I'll take a look at the issue with Train, updating the game rn

[xiv1337]

yeh i meant vphys from vmdl

[B1txor]

So if i understand it correctly to get the .vphys file now, i have to open the world_physics.vmdl_c file in source2viewer, then just copy the contents of the PHYS tab, and that would be the correct data for the .vphys file?

Also, does anyone have a correctly parsed .tri file of dust2? I have done the above which i think is correct, but my exported .tri file seems to either be in a wrong format, and/or not contain the triangles. I'm not sure..

[exp1007]

Hulls work slightly different in vmdl files, compare them and you will spot the difference easily. I don't remember exactly the crashing reasons but I think I fixed those using a filter for vertices.
Here I mentioned how to export vmdls easier using source2viewer
#4 (comment)

[Xsintashi]

Ok so basically: All community maps and Train don't work properly or at all.
Snippet how some spots are read on train: https://youtu.be/ynzlnCqBDeY

[exp1007]

Ok so basically: All community maps and Train don't work properly or at all. Snippet how some spots are read on train: https://youtu.be/ynzlnCqBDeY

Mesh CollisionAttributeIndex = 1

[MissedShot]

looks like new vmdl phys using m_VertexPositions instead of m_Vertices for hulls pos

Reference

simply replace
vector<float> vertex_processed = bytes_to_vec<float>(parser.get_value("m_parts[0].m_rnShape.m_hulls[" + index_str + "].m_Hull.m_Vertices"));
with
vector<float> vertex_processed = bytes_to_vec<float>(parser.get_value("m_parts[0].m_rnShape.m_hulls[" + index_str + "].m_Hull.m_VertexPositions"));
to parse vmdl PHYS file

de_mirage vmdl PHYS parsed with m_Vertices hulls

de_mirage vmdl PHYS parsed with m_VertexPositions hulls

[JanEricNitschke]

Ok so basically: All community maps and Train don't work properly or at all. Snippet how some spots are read on train: https://youtu.be/ynzlnCqBDeY

Mesh CollisionAttributeIndex = 1

What do you mean by that? If i use the parser on the PHYS from vmdl_c i get this for the triangles from train and then obviously basically everything is visible from everywhere.

[JanEricNitschke]

Found the issue. We cant just assume that we need to use m_nCollisionAttributeIndex = 0. You need to first check m_collisionAttributes[{idx}].m_CollisionGroupString to find the idxs where this is Default or default and then use those indices.

[qhb1023]

@JanEricNitschke Can u pull the code for a look

[JanEricNitschke]

@JanEricNitschke Can u pull the code for a look

@qhb1023
What do you mean?