Brute Force (XBOX 2003) - Models

[shakotay2]

@manbig343: thanks! But there's too many objects to find what you want so I just chose this one:
.

m01_b_0x1b0c54.png

----------------------------------------
For the object im my previous post the params from the xbm file are very clear:
...
167. 3a860, 168 at 0x35beec
168. 3aab0, 746 at 0x35ceac <<<<<
169. 3b400, 518 at 0x36149c

You just need to trick around with the face indices count.

Which you can get from a 000001000200 search in the ivd file:
ivd, search for face indices blocks:
...
658. 3a860 296 -> vCnt: 167
659. 3aab0 1192 -> vCnt: 745 <<<<<
660. 3b400 1000 -> vCnt: 517
661. 3bbd0 120 -> vCnt: 73
662. 3bcc0 136 -> vCnt: 70
...

(This couldn't be found automatically because the vertex count was 1 to small, sadly.)

[manbig343]

Okay, that's one of the other weapons in the game. I believe I actually have the texture for that weapon saved.

Last night I went in deep on trying to see if I could simply use renderdoc's EXPORT as CSV function as a way to get these character models from the game with much trial and error, and finding scripts that would allow me to import them into Blender as a 3d object. What happened was this -

I essentially got the entire models of Tex and Brutus in their original T-Pose/A-Pose doing this.

This is probably not the most optimized way to try and get these models but I did figure this out. Each model you see in the screenshots has separate parts to it. The body, head, legs, and armor on the characters are all separate. I hope that somewhat helps at all knowing that.

Now for some reason Tex's legs imported perfectly into blender through the .csv file I was able to make. I don't know why. Every other .csv model in the importer script once put in blender or anything else, probably due to the .csv importing script I used is missing many triangles.

If somewhere there exists a script to import .csv data into some program like blender or 3dsmax with these .csv exports having all of their faces intact instead of them being missing, or perhaps some sort of fix to the script I used which I could link if anyone wanted to see it, the models can then just be rigged afterward and can be used like that. But again this is most likely not that good of an idea.

The picture you see of Brutus to the right is from the Ultimate Halo Companion DVD which shows the exact same T-Pose!

[shakotay2]

I essentially got the entire models of Tex and Brutus in their original T-Pose/A-Pose doing this.

This is probably not the most optimized way to try and get these models but I did figure this out.

Yeah, cool! There's many ways to Rome - what counts is that you reach it (without feeling sad or so ).
.

m01_b-2293ac.png

[shakotay2]

rest of the beast (hands not tri stripped for some unknown reason):
.

Brutus.png

[manbig343]

Wait, how did you just get that complete model of brutus? Using hex2obj? Did you somehow use the numbers in my screenshot from the thing I sent in anyway for that?
Also can you link an .obj or fbx file for him? I'd like to try and get it rigged, and since those UV maps are correct I'm assuming I'll be able to use the brutus textures I got from the game, which I'll send here.

This winrar file contains all 3 of Brutus needed textures for his model they may need to be rotated.
I don't think the skeletons from the game are able to be imported whatsoever.

[manbig343]

That model in your screenshot actually might be a feral colonist or one of the outcasts, looking closer.

[shakotay2]

Wait, how did you just get that complete model of brutus? Using hex2obj?

The parameters are in the xmb files, see here (click up arrow):

Then I put them into hex2obj, yes.

Did you somehow use the numbers in my screenshot from the thing I sent in anyway for that?

No. (Counts, if any, are unreadable for me)

Also can you link an .obj or fbx file for him?

Here's the H2O files (not sure whether it's for Brutus or a "feral colonist", though):
.

BrutusOrNotBrutus.zip

Load objects-m01_b.ivd into hex2obj, then a H2O file. Press 'mesh' button.
(Use File\SaveAs mesh for exporting as wavefront obj file.)

btw: it would help if you had the vertex counts for "the real" Brutus' body parts, then I could search for them among a hundreds of parts.

(The tool for getting the params from xmb files was tested with two of them only, I might release it after further tests.)

[manbig343]

I'll have to find out Brutus' vertex things later then.

The model ended up actually being the Feral Shaman model.
I will have to get the textures for him through renderdoc, so when I do i'll post them in here. There is no proper working texture extractor from the game except for one which only works with 1 file, and isn't too functional.

[manbig343]

It appears something is up with this model when used with the proper textures it's asking for ingame.
The UV Map is also different than the actual texture of Brutus himself.
Perhaps the texture for this character is not accessible in renderdoc, but I will keep looking for it.

[manbig343]

Oh, fixed it. Just couldn't find the right textures. This model here is (actually) the Yellow feral outcast shaman.
Here i've modified the UV map so it would redirect to the green parts of the texture itself to match though. It's hard to get the exact textures in renderdoc that you want.

[manbig343]

Here are the textures for the model extracted of the feral outcast shaman.

[manbig343]

Here are the textures for the model extracted of the feral outcast shaman.

[manbig343]

The model I've put together with a rig from mixamo.com, the rig can be deleted afterward if not necessary for use.
It goes with the 2 textures I've posted.

[shakotay2]

well, I see, you've got some motivation!

Me, I don't have the time to finish the extraction tool so I'll leave here some H2O files and a short crash course on "How to use them":

BruteForce, use of H2O files explained (more or less)
-----------------------------------------------------
Understand how to use it and don't blame me!
(Please give a feedback on errors, if any, of course!)

Finishing the tool costs me too much time so I'll upload here
some intermediate H2O and log files for objects-m01_b.ivd
which should do the trick, too.

So you'll need objects-m01_b.ivd for this!
------------------------------------------
(Doesn't work for other .ivd files. This should be obvious, but some guys, well...)

The 631 H2O files in the zip are derived from the xbm file, that's why xbm_ is part of their names.

Looking into one of them you'll find a line like this:
0xdb0 3

So, yeah, each of these (first) lines will contain a face indices count of 3.
That's NOT a bug, it's, well, a workaround.

The good news: I'll show how to find the face index counts.
---------------------------------
example (.txt files are in the zip file)
-------

Code: Select all

- load objects-m01_b.ivd into hex2obj

- open "m01b xmb, params-log.txt" in notepad or any other text editor

- lines contain this;
  <no of H2O file>, startAddrFIs, vertex count (at) vertex startaddr

- as an example I start with this entry:
  504. 1d840, # 207 at 0x15ac74

- load the H2O file "objects-m01_b.xmb_504.h2o" into hex2obj
  check for addr 15ac74 and count 207 in the editboxes of step 3

- now for the face indices
  1) use 1d840 (address for step 1) for search in m01_b FIs.txt
     (which you need to open in notepad, too)

     The result is
     426. 1d840 480 -> vCnt: 206

    (Ignore the 426. It's just for numbering the lines here.)

  2) enter the count 480 in hex2obj, step 1 editbox for count
     (buttons at the right show 'Strip' and 'noPtC')

  3) press Go1 button, scroll down in the lower left list box,
     last line says: max face index: 207

  4) perfect! Press mesh button - wow, a chair.
     Enter 16 for UVpos in step 2.

     Use File/SaveAs mesh to export as a wavefront obj file.

(It's point 1) to 4) here, don't confuse it with step 1/ step 3
 in hex2obj, btw.)

------------------------------
yeah, now for a weird example, as there are many of them.
(Use the above as a guide in case you lost the overview somehow.)

I'll choose 459. 176c0, # 430 at 0x1144c4 from the ..params-log.txt file.

To speed up the procedure you could switch to 'PtCld' or 'Fake',
then press 'mesh' for a pre check of the object. This one looks uninteresting to me - well, next try:

486. 2d9c8, # 739 at 0x12f6b4

(486. in the ..params-log.txt file is the H2O number, so load that file into hex2obj)

Use 2d9c8 from step 1 (in hex2obj) for the search in m01_b FIs.txt (see point 1) above).

Result is:

Code: Select all

Notepad
---------------------------
Cannot find "2d9c8"

ok, manual search to be done, scroll down in that file 'till
-----------------------------
615. 2d9d4 1706 -> vCnt: 6140

This should be what we're searching for.
It's very important to use both params here, the address 2d9d4 AND the count 1706

(That's because the address is "off" some bytes in objects-m01_b.xmb for some unkn reason.)

Switch back to 'noPtC' in case it isn't active.
Change 2d9c8 to 2d9d4, replace 3 by 1706.

Point 3), see above gives us: max face index: 24573, what?

Reduce 1706 by 10, press Go1 (still wrong max face index)
and again (so 1686 now)
max face index: 739, perfect

Press 'mesh', interesting "wires"
----------------------------------

Attention:
- you may need to swap from 'Strip' to 'no Str' incase the meshviewer shows an ugly object.
- even if all parameters appear to be correct some sub meshes are spoiled for some unknown reason.
.

objects-m01_b.xmb_H2O.zip

[shakotay2]

This is the result when you use
571. 327b8, # 1308 at 0x189f94
from .. params-log.txt file in the zip in above post:
.

whoever.png

But I didn't find the head so I decided to let other people do it.