RF Online Unknow rare model extracted

[Rimbros]

RF Online its a old game with good stuff thinks and fashion, in the past someone made the unpackers for this game models, and now the game have all need to unpack and convert textures and meshes to redeable formats, but the conversor to OBJ can get a bunch of shit... hope somebody also Finale (now with good skills in Noesis) can explain wath its wrong with this extraction system. here i put all the tools necesary do this, its very easy extract.

Only need drag .RFS files inside of RFSUnPack.exe and u get .RFT if its texture, u get .MSH if its Models.
To get dds textures only need drag RFT inside of RFT-DDS.exe and now you get the .DDS files.
To get the models only need open the MshtoObj.exe Aplication and convert .MSH to .Obj format.

But .Obj are wrong extracted in rare cuts and parts like this...

I think its wrong calcualted half meshes and stored in the center of axis or something, if somebody can made the plugin for Noesis this can be very good.

Here the files samples Texture and Mesh, .exes unpackers and library i found abouth the format... thanks.

http://www.4shared.com/rar/EWhjUKu7/PARA.html

[finale00]

Maybe it's not applying any transformations, cause it's just skipping it.

Looks like a bunch of matrices defined for each mesh (there are 3 of them for each mesh. Don't know those 3 might be). But simply applying a single one of those matrices didn't produce any desirable results.

Anyways it looks like a good candidate for immMode rendering. For some reason I can't get the normals right though.

[Rimbros]

Game have 4 folders for player models, Ani, Bone, Mesh, Tex.

Here I include the 4 Elements

http://www.4shared.com/rar/NtdpFqL3/Player.html

[finale00]

A single model is broken down into various meshes.
I'm guessing it makes it easier to animate?

I can't get the normals and UV's right and have no clue what the problem is.
Maybe someone can figure out what's wrong with my mesh building.

http://xtsukihime.webs.com/Noesis%20Plu ... ine_msh.py

All of the data is tossed into lists and each mesh constructed with with `build_mesh`
I think that's how the format works from the C# code.

There's no format-checking here, so if you have any other msh plugins might want to remove them from the folder.

[Rimbros]

Finale i found this, its a link of something boy nava he made lots of thinks for RF also a viewer
in the viewer its posible see something thinks, i keep a link hope this help.

http://honor-rf.blogspot.com/2008/11/final-update.html

Btw he posted here in Xentax

viewtopic.php?f=16&t=3031

[CriticalError]

Finale i found this, its a link of something boy nava he made lots of thinks for RF also a viewer
in the viewer its posible see something thinks, i keep a link hope this help.

http://honor-rf.blogspot.com/2008/11/final-update.html

Btw he posted here in Xentax

viewtopic.php?f=16&t=3031

well my friend I remember for convert MSH to OBJ need be select different options, you need try, sometimes you get errors, so need activate Skip Sector or something like that, don't remember exactly, but someone can take a look it, I remember 1 topic you post with format structure + source convertor, but nobody take it I think

[Rimbros]

viewtopic.php?f=16&t=3031[/quote]well my friend I remember for convert MSH to OBJ need be select different options, you need try, sometimes you get errors, so need activate Skip Sector or something like that, don't remember exactly, but someone can take a look it, I remember 1 topic you post with format structure + source convertor, but nobody take it I think [/quote]

Thanks Crit, but now i have more info, if u can read the post you can see can be extracted but all the armors items have cuts in half, and stored in the center of the axis in every program. a complete bunch of trash. I ask abouth posibilty of extract each body part complete in good position, btw i see maybe if i use 3D ripper with this mshview program i can, but the little program its missing in broken links.

Finale wath you think abouth the info on this post? can be usefully?.

http://www.rfpoa.com/forum/showthread.php?t=71574

Btw i found this programe its new since 2010


Here the link to download
http://public.djzmo.com/?p=projects/rft ... mwo1.0.zip

But its something rare this only extract half parts.


Another option its i hope the Blender Master Szkaradek123 can take a view of the mesh building finale made in .py format.

[finale00]

I don't know how the transformations are applied.

[finale00]

Alright the first matrix in the mesh is the transformation that should be applied to the vertex group.
Just update the existing tools to apply that matrix and it should be fine.



The only issue now is that the UV's are stored per-face.

There are two ways to deal with this that I can think of, but both have not been obvious or have been unsuccessful

1: immediate mode rendering. It just looks weird...like really. All the normals looked wrong and the model just looked like a bunch of blocks put together

2: per-face UV to per-vertex UV. I'll have to duplicate vertices in this case I guess.

If someone can update the script to correctly map UV's then you can at least get the meshes out.

http://db.tt/WRtUTlF1

[Rimbros]

Maybe zskaradek but i think he leave the forum, or mr. adults but its so hard contact with he. then at least all its lost.

[Rimbros]

Interesting Finale you already fix the problem of the meshes, te only think need to fix its the uvMaps, simpli amazing also the meshes are nor more half cuts and now have good position in axis.
Its posible maybe you found how nava keep out the uvmaps info in the source code posted?

Here i put 3 files, OBJ files, one its keep out with Noesis meshes are fine but not uvmap, other its keep out with msh/obj converter mesh its wrong but uvmap its fine, and other its original mesh file the objects its the same, and i put the texture also to test, hope this can help.

http://www.4shared.com/rar/myIef1Ma/4objects.html

[finale00]

Here's a quickbms unpacker:

Code: Select all

#RF Online .RFS unpacker
#written by Tsukihime
#Feb 13, 2012

get FILES long
for i = 0 < FILES do
	savepos CURR
	math CURR += 40 #just hacking around
	get NAME string
	goto CURR
	get unk long
	get null long
	get unk2 long
	get unk3 long
	get OFFSET long
	get SIZE long
	log NAME OFFSET SIZE
next i

I know how the UV's are stored.

Code: Select all

Face {
   #indices
   int32[3] v1, v2, v3

   #uv's
   float[3] u1, v1, w1
   float[3] u2, v2, w2
   float[3] u3, v3, w3
}

Or something like that.
I'm just not getting any good results when creating a uv buffer for binding, and immediate mode rendering with those matrix transforms isn't turning out as great as it seems.

I was looking at the archives and it looks like some of the meshes actually have an idstring "MESH08", and that format is almost completely different since it doesn't store the UV's with the faces.

Can nava's tool load these?

[Rimbros]

Maybe this explain why the nava tool have this two options.

[finale00]

I'm asking because I couldn't get any results. Maybe I'm not using it properly? I wouldn't know what "sector3" is without looking at the source

EDIT: you know it just looks like a format they just left in and forgot to take out.

All of the indices are ordered like

Code: Select all

0 1 2
3 4 5
6 7 8
9 10 11
...

n n+1 n+2

And each vertex is like a 64 byte struct.

Why it's completely different from the other format except the mesh header is completely beyond me.

[Rimbros]

cant load the .obj file, but btw i put here the .obj generated by nava tool.

http://www.4shared.com/rar/AkNIJqo4/NAVATOOL.html

well the player mesh folders have 70% of the objects in game.
its can be fixed almost on this but cant be fixed in npc its ok i think.