Project Gotham Racing 4 models (.pak_hrd)

[shakotay2]

Strange mix of tris and quads and no idea, where to find the uv data.
.

GothamRacing_whatever_2.jpg

[gpktm]

It looks like the correct model. What you have there is the bonnet of the car. The topology seems messed up but I can fix that. I can also see some larger triangles which I presume is the lowest LOD model.

[shakotay2]

You may trick around with this interior's point cloud:
.

test-Inter_pct_cloud.zip

(Sadly the belonging face indices don't fit, dunno why.)

This is the last bigger block I've extracted. (Too tedious to care for all the smaller ones.)

[gpktm]

Thanks, mate. That point cloud will keep me busy, for a while.

In the meantime, can you tell me how is the process that you just did is called, please? So I can lookup for any relevant tutorials on the net and see if I can keep going myself?

[shakotay2]

Thanks, mate. That point cloud will keep me busy, for a while.

In the meantime, can you tell me how is the process that you just did is called, please?

It's "unzipping" zipped data without using header data (so there's headers but I don't use them).
Finding those headers depends on a signature. But I'd need another sample to tell more.

[gpktm]

Here is your sample. This is another file uploaded on the same forum where I found the GT90 file. I hope it helps.
https://sharemods.com/ryu5pg5dymfu/Indi ... k_hrd.html

[shakotay2]

Some results:
.

GT90_and_Indigo.jpg

[UB833]

ahh, looks like you guys are making some progress on those. keep it up

[shakotay2]

Since there's no feedback any more I thought I should release what I had done so far (don't expect a "nice" tutorial, sorry):

It's about handling 2 files from the GT90 block0, 00000005.dat (vertices) and 00029f6a.dat (face indices).
hex2obj can't handle separated data. Adding such a feature is a shitload of coding work.

It's much simpler to binary add those files. But this requires to know which face index.dat
belongs to which vertex.dat. Seems not all files in the block folders fit together. That's the problem.
(tool for block folders see next post)

So let's give the actual scenario a last chance, won't we (GT90)?
btw, FVFize is the offs from one vertex to the next one
For floats (4 bytes) it's often 32=8x4 bytes (vvvnnnuu), vertex, normals, uvs.

- In hex2obj
toggle upper button to bigE (big endian)
toggle the 'noPtC' button to 'PtCld'.
select ShortAll (step 3 in hex2obj)

vertices, big endian, FVFsize: 20
size of 00000005.dat: 0x4B103
div 20 -> 15373 = vertex count

- load that .dat file,
press 'mesh' button (view point cloud) -> test.obj being created,

rename test.obj in block_0 to test_0.obj
----------------------------------------

face indices, Strip, terminator FFFF
00029f6a.dat, size 0x125b9
div 2 -> 37596 = face indices count

- load that .dat file,
toggle the 'PtCld' button to 'noPtC'.
press 'mesh' button (ignore the ugly mesh; know why?)

creates test.obj,
- copy face indices from it to test_0.obj:
---------------------------------------
g submesh_0
f 1/1 2/2 3/3
f 2/2 4/4 3/3
f 5/5 6/6 7/7
f 6/6 8/8 7/7
f 9/9 10/10 11/11
f 10/10 12/12 11/11
...
.

GT90_face indices.jpg

[shakotay2]

You need to be familiar with offzip (see my previous posts) to get .dat files.
(Maybe first unzip the appended zip before going through this wall of text.)

You urgently need to understand that you'll need empty block folders for each model file.
You HAVE BEEN WARNED!

Here's explanation and the tool to ease things (a little bit only).

Credits go to Luigi Auriemma for offzip.exe
-------------------------------------------

Extracting Gotham Racing pak_hrd files
--------------------------------------

FAQ:
Q: there's 97 block directories filled with uncompressed data.
Do you really expected us to use hex2obj on them?
A: well, I feel your pain. But I'm a little bit exhausted now.
Time may come that someone writes a script, who knows...

Q: should I use separate working dirs for each pak_hrd file?
A: that's absolutely required. Create a copy of the unzipped
folder for each pak_hrd you want to uncompress.

**************************************************************
Be sure to start in the correct directoy where all
block_xx sub directories are EMPTY!

Otherwise you'll be flooded with hundreds of "overwrite?"
requests.

You can stop this only by continuously pressing ctrl-c
or killing the app process in the taskmanager.

**************************************************************

Use Make_H2O_GR.exe on your own risk!
-------------------------------------
No responsibility is taken over from my side for any loss,
data, nerves, whatever.
-----------------------

How to...

• Extract the files in the zip. (There's 128 empty "block_xx" dirs contained.)
• copy the .pak_hrd into the app dir. I use GT90_Concept.pak_hrd here.
• Start Make_H2O_GR.exe
• File\open to load a .pak_hrd file

BE SURE to have it in the same directory as Make_H2O_GR.exe.
YOU HAVE BEEN WARNED.
------------------------------------------------------------

- after "the Logging done." message popped up close the app.
------------------------------------------------------------

There's a lot of [absoluteAddr]_block_xx.bin files. You can ignore them for now.
The decompressed data should be found in the block_xx directories.

---------------------------
>>> Let's look into block_0.

I give an expample how to create an obj file (without uvs, so far!)
from the .dat files in the Block_0 directory:
--------------------------------------------
Same example as in the previous post, but more detailed now.
- In hex2obj toggle the 'noPtC' button to 'PtCld'.

vertices, big endian, FVFsize: 20
size of 00000005.dat: 0x4B103
div 20 -> 15373 = vertex count

- load that .dat file,
press 'mesh' button (view point cloud) -> test.obj being created,

rename test.obj in block_0 to test_0.obj
----------------------------------------
face indices, Strip, terminator FFFF
00029f6a.dat, size 0x125b9
div 2 -> 37596 = face indices count

- load that .dat file,
toggle the 'PtCld' button to 'noPtC'.
press 'mesh' button (ignore the ugly mesh; know why?)

creates test.obj,
- copy face indices from it to test_0.obj:
---------------------------------------
g submesh_0
f 1/1 2/2 3/3
f 2/2 4/4 3/3
f 5/5 6/6 7/7
f 6/6 8/8 7/7
f 9/9 10/10 11/11
f 10/10 12/12 11/11
...

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

Another bigger sized candidate, block_42:
(yes, "42 is the answer";-)

Other than the above Block_0 example, the FVFsize is 28 here, surprise!
So size of 00000005.dat: 0x66440 div 28 -> 14960 vertices

(REMEMBER the 'noPtC' button handling, see above.)

- press 'mesh' -> test.obj being created,
rename it to test_42.obj in Block_42
-----------------------------------
face indices
0003fc2f.dat, size 0x22C8E div 2 -> 71239
div 2 -> 37596 = face indices count

- press 'mesh' button

creates test.obj,
- copy face indices from it to test_42.obj

-----------------------------------
Loading this interior into blender sadly shows up so much garbage that I'm not sure
whether 0003fc2f.dat is the suiting face indices file?

---------------------------------
Now for those "intermediate" blocks. They were spread all over the pak_hrd file,
but I collected them at the end.

Anyways, block_91
-----------------

I'd start with 0000c0e3.dat, size 0x48D8C, div 20 -> 14919
(well, looks like interior)

wheres the .dat file containing the suiting face indices?
Frankly, my dear, I've got no idea.
.

Gotham Racing.zip

[shakotay2]

How awful. I really hate it. (and I guess, you will, too... )

Anyways, here's a merging .dat files tool.

Merging two .dat files ( 1) vertex and 2) face_indices)
There's dozens of .dat files and the riddle to be solved is:
which face_index.dat belongs to which vertex.dat,
BEFORE merging them!
-------------------------------------------------------

Make_H2O_GR_b.exe, choose Extra\Merge dat

Using the exe requires two dll files from previous zip!
(I'm pretty sure, there will be dozens of people who don't get this.
I really should have linked those dlls statically; next time... )
-------------------------------------------------------

Name of the merged dat is like so:
nameOfVertexDat(vertex_count)_HexAddressOfFaceIndices_nameOfFaceIndexDat.dat

See picture in zip how to use parameters.

Face index count to be estimated (2.5 x vertexcount?)
until pressing go1 leads to a fitting vertex count
in the lower left list box.

GothamRacing_merging_dat_files.zip

edit: for suiting vertex-/faceindex .dat pairs see list here:

Sadly there's not too many fitting pairs. Plus the models are not very detailed and the interest is low.
------------------------------------------------------
Thus I consider this project being finished.
So - not to do: uvs
------------------------------------------------------

last not least here's a quote from gpktm:

It seems that the DAT files contain two LODs. One high detailed and the other one is low. Once the coordinates of the high LODs finish, it starts creating the low LOD. That's why you see those big polygons overlapping the detailed model.

And a 2nd one:

Another interesting finding is in the interior in Block_42. The data of the interior up to a point is matching the geometry and then it starts doing some strange leaps from one point to another, then again it starts creating the mesh again properly.

1st Part = Correct Coordinates
2nd Part = Wrong Coordinates
3rd Part = Correct Coordinates
4th Part = Wrong Coordinates

[UB833]

To shakotay, recently Nenkai made a documentation of the file format:

viewtopic.php?p=192218#p192218

[shakotay2]

The format is not the problem (except for the uvs, maybe) as you can see from my results:

If it were someone would have shown a model, using the format description, wouldn't he?

The problem is to find suiting vertices/faces .dat file pairs.

So what the project is missing is a list of counts for all .dat files. (I have no idea why I didn't create it...)
"Face index count to be estimated". (Ok, need to check this. Should be doable...)

I checked W12.pak_hrd and seems there's some problems with offzip.

So I strongly recommend to start with GT90_Concept.pak_hrd as explained in the ...HowTo text file.

edit: made that counts list. .dat files containing face indices have a suffix "_FIs" (surprise )
The other .dat file lines contain 3 counts, for the FVFsizes of 28, 20 and 14.

Where (11) in "00000004 (11).dat" for example means that I was too lazy to do the files' check recursively. Thus those with the same names/addresses were numbered.

(Files < 3kB excluded)

Sadly there's less matches than expected. Exact matches/pairs for the counts 15373, 14960 and 3475 only, afaics.

GT90_Concept.pak_hrd:

Code: Select all

 00000004 (10).dat 645, 903, 1290
 00000004 (11).dat 645, 903, 1290
 00000004 (12).dat 566, 793, 1132
 00000004 (13).dat 566, 793, 1132
 00000004 (14).dat 566, 793, 1132
 00000004 (15).dat 566, 793, 1132
 00000004 (16).dat 462, 648, 925
 00000004 (17).dat 462, 648, 925
 00000004 (18).dat 462, 648, 925
 00000004 (19).dat 462, 648, 925
 00000004 (2).dat 1909, 2672, 3818
 00000004 (20).dat 363, 509, 727
 00000004 (21).dat 360, 504, 720
 00000004 (22).dat 347, 486, 694
 00000004 (23).dat 344, 482, 688
 00000004 (24).dat 344, 482, 688
 00000004 (25).dat 185, 260, 371
 00000004 (26).dat 185, 260, 371
 00000004 (27).dat 185, 260, 371
 00000004 (28).dat 185, 260, 371
 00000004 (29).dat 170, 239, 341
 00000004 (3).dat 1907, 2669, 3814
 00000004 (30).dat 147, 207, 295
 00000004 (31).dat 147, 207, 295
 00000004 (32).dat 116, 163, 232
 00000004 (33).dat 115, 162, 231
 00000004 (34).dat 115, 161, 230
 00000004 (35).dat 115, 161, 230
 00000004 (36).dat 115, 161, 230
 00000004 (37).dat 115, 161, 230
 00000004 (38).dat 111, 156, 222
 00000004 (39).dat 111, 156, 222
 00000004 (4).dat 1905, 2667, 3810
 00000004 (5).dat 750, 1051, 1501
 00000004 (6).dat 727, 1018, 1454
 00000004 (7).dat 652, 914, 1305
 00000004 (8).dat 645, 903, 1290
 00000004 (9).dat 645, 903, 1290
 00000004.dat 1910, 2674, 3820
 00000005 (2).dat 10980, 15373, 21961
 00000005 (3).dat 10656, 14919, 21312
 00000005 (4).dat 3475, 4865, 6950
 00000005.dat 14960, 20944, 29920
 000001b7.dat 111, 156, 222
 000001cf.dat 750, 1051, 1501
 000001e1.dat 185, 260, 371
 000001e6 (2).dat 185, 260, 371
 000001e6.dat 185, 260, 371
 000001eb.dat 185, 260, 371
 00000205.dat 652, 914, 1305
 00000213.dat 727, 1018, 1454
 000004c8.dat_FIs 3475
 00000743.dat 363, 509, 727
 00000d45.dat 347, 486, 694
 00000e04.dat 566, 793, 1132
 00000e13.dat 645, 903, 1290
 00000e34.dat 645, 903, 1290
 00000e44.dat 645, 903, 1290
 000017a7.dat_FIs 1909
 000017de.dat_FIs 1910
 00001e60.dat_FIs 645
 0000251e.dat 645, 903, 1290
 00002779.dat_FIs 914
 0000297a.dat_FIs 914
 00002b21.dat_FIs 1018
 00002bf9.dat 360, 504, 720
 00002d30.dat_FIs 1018
 000030b4.dat_FIs 1051
 000030ee.dat 1910, 2674, 3820
 0000327f.dat_FIs 1051
 000034c7.dat 566, 793, 1132
 00003515.dat 566, 793, 1132
 000039f1.dat 116, 163, 232
 0000437a.dat_FIs 645
 00004bb7.dat 566, 793, 1132
 00005815.dat 462, 648, 925
 00005945.dat 1909, 2672, 3818
 00006fb5.dat 462, 648, 925
 00007006.dat 1905, 2667, 3810
 0000718e.dat_FIs 1907
 000071e1.dat_FIs 1905
 00007745.dat 170, 239, 341
 00008582.dat 344, 482, 688
 00008750.dat 1907, 2669, 3814
 0000a0fd.dat 344, 482, 688
 0000a368.dat 462, 648, 925
 0000bb42.dat_FIs 1910
 0000c0e3.dat 10656, 14919, 21312
 0000cb62.dat 462, 648, 925
 0000db98.dat 115, 161, 230
 0000e1e3.dat_FIs 1905
 0000e305.dat_FIs 1909
 0000f8da.dat_FIs 1907
 00010319.dat 115, 161, 230
 00010387.dat 115, 161, 230
 00011a45.dat 115, 161, 230
 00029f6a.dat_FIs 15373

 000316ca.dat 14960, 20944, 29920
 0003fc2f.dat_FIs 14960
 000712f4.dat_FIs 14960
 00087a43.dat 3475, 4865, 6950
 0009adf0.dat_FIs 3475
 0009e1d3.dat 111, 156, 222

Then I merged (old list without FVFsize of 28)
000316ca.dat 20944 (29920) and 0003fc2f.dat_FIs 14960
( 000712f4.dat_FIs 14960, alternatively?)

to 000316ca(20944)_0x66440_0003fc2f.dat and loaded them into hex2obj.
Surprise, FVFsize is 28, thus the vertex count to be corrected to 14960!
.

000316ca(20944)_0x66440_0003fc2f.png

Sadly there's many weird faces, so my motivation is low, again...

[shakotay2]

Here's the list for Indigo and a result (00000005(27832)_0x87e60_0000e8ef.dat):

Code: Select all

 00000004 (10).dat 394, 551, 788
 00000004 (11).dat 386, 541, 772
 00000004 (12).dat 386, 541, 772
 00000004 (13).dat 386, 541, 772
 00000004 (14).dat 386, 541, 772
 00000004 (15).dat 377, 527, 754
 00000004 (16).dat 377, 527, 754
 00000004 (17).dat 374, 523, 748
 00000004 (18).dat 237, 332, 474
 00000004 (19).dat 237, 332, 474
 00000004 (2).dat 3336, 4670, 6672
 00000004 (20).dat 237, 332, 474
 00000004 (21).dat 237, 332, 474
 00000004 (22).dat 172, 242, 345
 00000004 (23).dat 172, 242, 345
 00000004 (24).dat 171, 240, 342
 00000004 (25).dat 171, 240, 342
 00000004 (26).dat 171, 240, 342
 00000004 (27).dat 171, 240, 342
 00000004 (28).dat 162, 228, 325
 00000004 (29).dat 162, 228, 325
 00000004 (3).dat 3336, 4670, 6672
 00000004 (30).dat 162, 228, 325
 00000004 (31).dat 162, 228, 325
 00000004 (32).dat 159, 222, 318
 00000004 (33).dat 159, 222, 318
 00000004 (34).dat 159, 222, 318
 00000004 (35).dat 159, 222, 318
 00000004 (36).dat 147, 207, 295
 00000004 (37).dat 140, 197, 281
 00000004 (38).dat 140, 197, 281
 00000004 (39).dat 140, 196, 280
 00000004 (4).dat 3336, 4670, 6672
 00000004 (40).dat 116, 163, 232
 00000004 (41).dat 115, 162, 231
 00000004 (5).dat 3336, 4670, 6672
 00000004 (6).dat 771, 1079, 1542
 00000004 (7).dat 769, 1076, 1538
 00000004 (8).dat 615, 861, 1230
 00000004 (9).dat 614, 859, 1228
 00000004.dat 4113, 5759, 8227
 00000005 (2).dat 14960, 20944, 29920
 00000005 (3).dat 10656, 14919, 21312
 00000005 (4).dat 5227, 7319, 10455
 00000005 (5).dat 3874, 5423, 7748
 00000005 (6).dat 3475, 4865, 6950

 00000005.dat 19880, 27832, 39760
 00000030 (2).dat 377, 527, 754
 00000030 (3).dat 377, 527, 754
 00000030 (4).dat 374, 523, 748
 00000030.dat 394, 551, 788
 000001cf.dat 750, 1051, 1501
 000001e1.dat 185, 260, 371
 000001e6.dat 185, 260, 371
 000001eb.dat 185, 260, 371
 00000206.dat 172, 242, 345
 00000207.dat 172, 242, 345
 00000211.dat 171, 240, 342
 0000021b.dat 140, 197, 281
 00000279.dat 5227, 7319, 10455
 00000286.dat 771, 1079, 1542
 00000743.dat 363, 509, 727
 00000e04.dat 566, 793, 1132
 00000e2d.dat 159, 222, 318
 00000e34.dat 645, 903, 1290
 00000e44.dat 645, 903, 1290
 000011cd.dat 386, 541, 772
 000011ce.dat 769, 1076, 1538
 00001f73.dat 615, 861, 1230
 0000251e.dat 645, 903, 1290
 0000267f.dat 237, 332, 474
 000027c0.dat 237, 332, 474
 00002e65.dat_FIs 615 (1827)
 00002e79.dat_FIs 614 (1778)
 000030ee.dat 1910, 2674, 3820
 0000327f.dat_FIs 1051 (2891)
 00003515.dat 566, 793, 1132
 00003a01.dat 237, 332, 474
 00003b6e.dat_FIs 769 (2187)
 00003be0.dat_FIs 771 (2203)
 00003cec.dat 237, 332, 474
 00003e62.dat_FIs 771 (2203)
 00003f0d.dat 3336, 4670, 6672
 0000437a.dat_FIs 645 (1474)
 000049af.dat 171, 240, 342
 00004bb7.dat 566, 793, 1132
 00004d38.dat_FIs 769 (2187)
 00004dd4.dat_FIs 615 (1827)
 000055ce.dat 159, 222, 318
 00005719.dat 140, 197, 281
 0000588d.dat 171, 240, 342
 00005945.dat 1909, 2672, 3818
 000064ba.dat 159, 222, 318
 00006724.dat 386, 541, 772
 00006f94.dat_FIs 3336 (6862)
 00006fb5.dat 462, 648, 925
 00006fc9.dat_FIs 3336 (6792)
 00006fef.dat_FIs 3336 (6860)
 00007014.dat_FIs 3336 (6788)
 0000761d.dat 386, 541, 772
 00007a14.dat 614, 859, 1228
 00008750.dat 1907, 2669, 3814
 00009419.dat 3336, 4670, 6672
 0000a154.dat_FIs 2369 (10410)
 0000a32f.dat 3336, 4670, 6672
 0000a368.dat 462, 648, 925
 0000a889.dat_FIs 614 (1778)
 0000ae9d.dat_FIs 3336 (6862)
 0000b1b7.dat 171, 240, 342
 0000bb42.dat_FIs 1910 (4740)
 0000bdc2.dat 159, 222, 318
 0000cb62.dat 462, 648, 925
 0000cef4.dat 386, 541, 772
 0000db98.dat 115, 161, 230
 0000e0e8.dat 162, 228, 325
 0000e305.dat_FIs 1909 (4799)

 0000e8ef.dat_FIs 19880 (61076)
 0000f8da.dat_FIs 1907 (4917)
 0000fc09.dat 3336, 4670, 6672
 00010387.dat 115, 161, 230
 000103de.dat_FIs 3336 (6792)
 0001133f.dat_FIs 3336 (6788)
 00011a45.dat 115, 161, 230
 00013584.dat 162, 228, 325
 00014516.dat 162, 228, 325
 00014cbb.dat 3874, 5423, 7748
 00014fe8.dat_FIs 7319 (28443)
 00016bf4.dat_FIs 3336 (6860)
 00019e13.dat 162, 228, 325
 00029c9e.dat_FIs 7319 (28443)
 0003fc2f.dat_FIs 14960 (66681)

(The Merger emits the FVF20 vertex count (27832) while it's 19880 for FVFsize= 28.)
.

00000005(27832)_0x87e60_0000e8ef-dat.png

Faces inside this interior model are weird. Maybe it's several sub meshes?

[daemon1]

I'm making a tool, it works with all cars from PGR 3 & 4.
Will be posted soon after some testing.