Driveclub (PS4)

[micro777]

I started looking for values manually in Model Researcher and have no idea what information the BODY_COL'BODY'BODY_LOD1|BODY_LOD1Shape file carries.

What magic do you use?

https://www.mediafire.com/file/oshqmra1gxn2vw9/Screen+camaro.zip/file

 

Edited June 18 by micro777

[shak-otay]

You could use the parameters from the first picture here (H2O file for hex2obj).

0x2F3A72 8589
Vb1
44 16
0x120B84 20792
020000
0x0 255

 

Or use the bms script from h3x3r which produces 3 Index and 72 vertex buffers from the Camaro.

The real magic is in finding out which indices fit for which vertex blocks. I.e. finding the starting addresses of sub index blocks.

Camaro chassis

(Will need some fine tuning and scaling but basically it's the chassis.)

Camaro chassis scaling

Edited June 19 by shak-otay

[shak-otay]

camaro_face_index_blocks_stopp_start

camaro_interior

Still dividing face index blocks manually...

 

edit: could anyone make sense of the IDs? Example:

|BODY_COL|BODY|BOOST_L2|BOOST_L2_EXTERIOR_ONLY

is addressed/linked to in two ways here

# 0xf946: 10000 449b 1 0 be8 #58 28 35 b6 4e f4 1e 0 58 0x1336e

by the ID #58 28 35 b6 4e f4 and by the offset 0x1336e.

There are other IDs (or signatures) which seem to mark  a (lod?) group, like 010057d12351.

Edited June 24 by shak-otay

[shak-otay]

Couldn't find a sense of logic here: both face index blocks are far, far away from each other:

Camaro_checking_FI_blocks

 

A good trick is to find/reduce a vertex block to one module, such as the cabin here.  Then find the suiting face index block dependent on the vertex count of that module:

camaro cabin (and point cloud)

---------------

Chevi_Camaro-parts-missing

Edited June 26 by shak-otay

[h3x3r]

My turn... LaFerrari FXXK but it missing 3 things.

Tires. I was able get only uv :(

Back glass. Same, only uv 😞

And lastly scale factor. Not sure where to get. I found one but it wasn't even close.

[shak-otay]

I'm using an estimated scale factor of 2.2

btw, the connection between mesh names and vertex blocks - did you get them by using the ids?

If so, how?

Edited July 15 by shak-otay

[h3x3r]

As I said before it's stored in *.Mesh file.

For example file > ferrari_laferrari_fxxk\intermediates\vehicles\art\vehicles\ferrari_laferrari_fxxk\maya\ferrari_laferrari_fxxk_exportbackup.mb\_BODY_COL_BODY_BODYShape.Mesh

I have a template which prints info about vertex block and index block.

     Model Index: 0
Vertex Buffer Id: 4060042755
 Index Buffer Id: 3886391541
    Vertex Count: 12939
    Index Offset: 28
     Index Count: 120

    Vertex Count: 12895
    Index Offset: 268
     Index Count: 216

    Vertex Count: 12817
    Index Offset: 700
     Index Count: 1242

    Vertex Count: 12499
    Index Offset: 3184
     Index Count: 10584

    Vertex Count: 10210
    Index Offset: 24352
     Index Count: 18288

    Vertex Count: 6477
    Index Offset: 60928
     Index Count: 3168

    Vertex Count: 5328
    Index Offset: 67264
     Index Count: 1752

    Vertex Count: 4494
    Index Offset: 70768
     Index Count: 12822

    Vertex Count: 567
    Index Offset: 96412
     Index Count: 2016

So i combine 4060042755.VertexBuffer and 3886391541.IndexBuffer and print *.obj from given values. Index offset is header size 28 + given offset. Vertex offset/count doesn't matter since it's controlled by indices. I just print whole *.VertexBuffer.

Most messy thing there is stride order. For example stride 44 has 7 types.

 

EDiT: I just realized that there is also collision vertex buffer. So here we go... tire

Also realized that collision has correct scale factor and it has a same amount of vertices/indices. But no uv or normals.

Edited July 15 by h3x3r

[shak-otay]

1 hour ago, h3x3r said:

As I said before it's stored in *.Mesh file.

Well, see. Must have overread that. Had/used the rpk file only.

What a waste of time and life...

Thanks anyways!

[h3x3r]

Well it start to be tricky. I noticed that *.HierArchy is constructor for collision meshes. It has matrix4x4. Also collision meshes uses only 3 floats all the time. So there is no confusion in order type. But it lack uv's a normals somtimes.

So my best bet is output all collision meshes then extract uv's and merge it together. Which will be pain in ass without automation.

EDiT: So far so good... Collision mesh. Some parts must be corrected with matrix.

Assembled...

Edited July 16 by h3x3r

[juanmasilsca]

On 5/9/2024 at 8:23 AM, MichaelBFS said:

...Here is an example of an extracted car from Driveclub. Screenshot and in FBX format and textures in the archive link below.

example extracted car from Driveclub in FBX and textures:

https://www.mediafire.com/file/cv8wfb0yco1j2ih/Driveclub_ferrari_f40.zip/file

Where did you get this extracted model? Any idea who did it?

[juanmasilsca]

I used the bms script to unpack the vehicle files. Textures are there as .tif. However, no editor I have tried was able to open them. Anyone else could?

[bicit]

I'd like to share my extracted driveclub raw car files so everyone who come across here doesn't have to do the tedious part. Hope it helps!
Driveclub Car Files

[bohhu]

On 16.07.2024 at 12:34, h3x3r said:

Ну, это начинает быть сложным. Я заметил, что *.HierArchy - это конструктор для сеток столкновений. У него есть matrix4x4. Также сетки столкновений используют только 3 float все время. Так что нет путаницы в типе порядка. Но иногда ему не хватает uv и нормалей.

Так что мой лучший выбор - вывести все сетки столкновений, затем извлечь UV и объединить их вместе. Что будет головной болью без автоматизации.

EDiT: Пока все хорошо... Сетка столкновений. Некоторые части нужно исправить с помощью матрицы.

Собрано...

Hi, I'm new to this topic and didn't quite understand what you did with the rpk file to convert it to a 3d model, I would be very grateful if you could tell me in more detail for beginners what can be done with the rpk to convert it to a 3d model. How working templates for 010 editor?