Page 113 of 138
Re: Forza Motorsport Resource Extraction (.carbin)
Posted: Sun Jan 29, 2017 3:04 pm
by shakotay2
well, long story short: quicksort seems to be the solution. (now, it wasn't! see version
j2 and ignore this post)
I thought of quicksorting the faceindices from the very beginning but for some unknown reason
I feared there'd be a reason for their scrambling and Qsort would spoil erverything.
gosh, it didn't.
Only problem is the multitexture for the
body_a4 submesh.
Still no clue how to separate them other than manually placing "g forced_submesh_a4_x" lines.
(Also I've the strange feeling that we need to figure out the difference between material submeshes
and lod submeshes.)
Be informed that the quicksort version can handle ONE modelbin file only so far.
BMW_i8_body-Quicksort_result.JPG
You wouldn't believe about the simplicity of the code used for creating the groups, would you?
Code: Select all
for (i=0;i<FIcnt/3;i++) indArr[i]= i ;
quickSort((int*)FImedian, (int*)indArr, 0, FIcnt/3) ;
cntEx= 0 ;
for (i=0;i<FIcnt/3;i++) {
a= FIarr[0][indArr[i]] ; b= FIarr[1][indArr[i]] ; c= FIarr[2][indArr[i]] ;
if (((a+b+c)/3 > dwUVsum[cntEx])&&(cntEx<cntMax)) { // next Material-Submesh
_itoa(cntEx, szNo, 10) ; //
strcpy(szFileName, szTmp) ; strcat(szFileName, szNo) ;
if (bUV_28[cntEx+1]) strcat(szFileName, "_UVB28") ; // +1, warum auch immer
fprintf( stream, "g %s\n", szFileName) ;
cntEx++ ;
}
fprintf( stream, "f %d/%d %d/%d %d/%d\n", a,a, b,b, c,c) ;
}
(
http://www.algolist.net/Algorithms/Sorting/Quicksort)
Re: Forza Motorsport Resource Extraction (.carbin)
Posted: Sun Jan 29, 2017 3:33 pm
by Andrakann
Output for BMW_i8_15_body_a is very messy for me, looks like all lods for meshes put in one mesh..
Upd: I move them a little vertically for this pic, only lowest LOD seems to be separated:
Re: Forza Motorsport Resource Extraction (.carbin)
Posted: Sun Jan 29, 2017 3:42 pm
by shakotay2
you mean body_a_4?
If so, my fears come true: qsort merges the lods?
I'm sorry to say, that I don't have a clue how to fix that other than using previous versions of the exe (-gc, -j).
Reason is that the different algos are incompatible, i.e. I can't "mix" them.
Re: Forza Motorsport Resource Extraction (.carbin)
Posted: Sun Jan 29, 2017 4:05 pm
by Andrakann
shakotay2 wrote:you mean body_a_4?
If so, my fears come true: qsort merges the lods?
Yes, it's merges them, but a little strange:
(another strange thing is i copied this part list from search results for "g" in NP++ and there's missing first mesh in list with weird (autogenerated?) name "object #0" - copied it from DE)
object #0 - lods0-4 for black material
body_a_0 - lod5 for black material
body_a_UVB28_1 - lods0-5 for black material undergrille mesh
body_a_2 - lods0-4 for black/undercarriage material
body_a_3 - lod5 for black/undercarriage material
body_a_4 - lods0-4 for dark carpaint material
body_a_5 - lod5 for dark carpaint material
body_a_6 - lods0-4 for light carpaint material
body_a_7 -lod5 for light carpaint material
Re: Forza Motorsport Resource Extraction (.carbin)
Posted: Sun Jan 29, 2017 4:27 pm
by Georgie55
Hey guys, made some progress on extracting the horizon edition huracan. Thanks for the tool!
Re: Forza Motorsport Resource Extraction (.carbin)
Posted: Sun Jan 29, 2017 4:34 pm
by dimon4ik6565
shakotay2 wrote:you mean body_a_4?
If so, my fears come true: qsort merges the lods?
I'm sorry to say, that I don't have a clue how to fix that other than using previous versions of the exe (-gc, -j).
Reason is that the different algos are incompatible, i.e. I can't "mix" them.
I don't know how hard to do it, but is your tool can see mesh is lod, or no? maybe you can just skip lods? just for test of course, and then split meshes by materials. j version merge lods in this lamborghini like: glass (3 lods), body_a (one lod) and in some meshes like bumpers, hood, trunk, lods contains parts of lod0 meshes.
Anyway, thanks for this big job that you do.
1.jpg
Re: Forza Motorsport Resource Extraction (.carbin)
Posted: Sun Jan 29, 2017 4:56 pm
by shakotay2
dimon4ik6565 wrote:, but is your tool can see mesh is lod, or no?
obviously: "no"
j version merge lods in this lamborghini like: glass (3 lods), body_a (one lod) and in some meshes like bumpers, hood, trunk, lods contains parts of lod0 meshes.
-j uses a brute force algorithm, as already mentioned several times. It creates 3 (to 10) times more groups than the qsort algo (which mostly produces exact borders) but in most cases -j doesn't hit the border between groups exactly.
There's a single line in the -j algo that "switches" between 3 and 10 times. But the later produces a bunch of superfluous groups
(guess, most people wouldn't like that).
Best results I got was gathering some "g" lines from a "-j x3" created obj file and inserting them manually into a -q created obj file.
Andrakann wrote:there's missing first mesh in list with weird (autogenerated?) name "object #0"
yep, thx, stupid me forgot to create the first group's name.
(updated the link in my sig)
Re: Forza Motorsport Resource Extraction (.carbin)
Posted: Sun Jan 29, 2017 8:28 pm
by Frazzlee
Re: Forza Motorsport Resource Extraction (.carbin)
Posted: Sun Jan 29, 2017 8:49 pm
by Lord Neophyte
think fake like always YCA made strange announcement *fooling*
so why nobody else is having all parts and textures complete ?!
sounds all stange imho
Re: Forza Motorsport Resource Extraction (.carbin)
Posted: Sun Jan 29, 2017 9:12 pm
by TheAdmiester
Lord Neophyte wrote:
think fake like always YCA made strange announcement *fooling*
so why nobody else is having all parts and textures complete ?!
sounds all stange imho
Isn't the Forza model an Acura and LHD? Theirs is an RHD Honda. I doubt it's from FH3.
Re: Forza Motorsport Resource Extraction (.carbin)
Posted: Sun Jan 29, 2017 9:30 pm
by baba0rum
TheAdmiester wrote:
Lord Neophyte wrote:
think fake like always YCA made strange announcement *fooling*
so why nobody else is having all parts and textures complete ?!
sounds all stange imho
Isn't the Forza model an Acura and LHD? Theirs is an RHD Honda. I doubt it's from FH3.
Acura NSX '16 from gamemodels.ru
and
YCA NSX are 100% same (3d models)
Cant check the texture as i still havent got it to work but would be nice to compare. Quick pic taken:
http://img15.hostingpics.net/pics/642315471.jpg
http://img15.hostingpics.net/pics/675029312.jpg
Re: Forza Motorsport Resource Extraction (.carbin)
Posted: Sun Jan 29, 2017 11:07 pm
by shakotay2
I think the uvmaps are ok (BMW_i8_15, exterior), more or less, it's a matter of manually inserting grouping lines so far:
BMW_i8_15_uvs.jpg
(updated sig to -q1, which inserts (modulo 1000) grouping lines)
Re: Forza Motorsport Resource Extraction (.carbin)
Posted: Sun Jan 29, 2017 11:20 pm
by baba0rum
Very nice progress!
Re: Forza Motorsport Resource Extraction (.carbin)
Posted: Mon Jan 30, 2017 5:03 am
by Da7K
I'm not sure if this issue is already noticed or not, but on some cars that i have tried exporting, the dashstructure and dashpad are usually misscaled, seems like the dashpad is following the steeringwheel scale and dashstructure is just plain wrong, here is some pics and the problematic files (dashpad and dash structure from 86 Audi Quattro).
http://www.mediafire.com/file/oyeie315v ... oblems.rar
Re: Forza Motorsport Resource Extraction (.carbin)
Posted: Mon Jan 30, 2017 3:08 pm
by shakotay2
Da7K wrote:the dashstructure and dashpad are usually misscaled, seems like the dashpad is following the steeringwheel scale
wrong position/scaling is a known issue for moveable parts.
Didn't know about the dash, but I consider it as a minor issue (as for the moveable parts I can't change it atm).
and dashstructure is just plain wrong
I'm no car expert, how should it look?
There's some major problem with separating uv maps when joining a_1000_ex and a_2000_ex to a_2000_ex
some unwanted map parts to appear ("ex" means autocreated modulo 1000):
AUD_Quattro_dash_n_steeringwheel.JPG
The map in the bottom pic looks better but for some strange reason only a quarter of the mesh is selected.
(Fixed a bug with -j, crashing with AUD_Quattro
dashpad_a.modelbin)
