Dead or Alive series formats and tools

[Rosalin]

[Lilstormcloud]

omg.. dat calculation lol

[Rosalin]

When editing vertex structure, especially in lisa04, I realized that I don't need to extend index buffer size.
because I can use previous mesh's index buffer, which is shrunk to 0,0,0 and unused.

also previous vertex buffer can be brought to add more vertices.

This will not going to require much correction in TMC file.
and also I can bring most buffers of previous mesh for for further editing.
They can remain unused.

This doesn't corrupt vertex count vertex ID and vertex structures remained, so Stickitback and Morph-target still work for those edited.

[tcl36478655]

Only adding 4 vertex cost a whole night...(╯‵□′)╯︵┻━┻

[tcl36478655]

did anyone met this following:

RACHEL_DLCU_001.TMCL
info offset in 'blp' = 0xf10 <==== must be wrong,should be 0xd94
ftable order in 'blp' = 2257
size in 'blp' = 29024 <blp says that this file is unpacked>

same the SARAH_DLCU_002.TMC

[Protocol X27]

@tcl, nice edit, Did you add just the missing faces?

Yes, I did see some discrepancies in some of the arch calculations. I mostly just grab the key if I need it. I've still been doing my offsets and injection manually.

@lsc, Yeah, I definitely see where some complexity comes into play. I guess I'll stick to the basic questions first. Which mesh object are you using for the replacement? One of the undies?

Once you decide on which mesh, (probably repeating myself or forgot) ... Are selecting each vertex in a face and following that process for every single one?

[tcl36478655]

@Protocol X27
I did like these below:
1.add the vertexs into body(may lend some from other obj) <TMCL>
2.fix vertexs' opsition\UV\Normal\bone_id<TMCL>
3.fix faces<TMCL>
4.fix body's index<TMC>

RACHEL_DLCU_001.TMCL
info offset in 'blp' = 0xf10
ftable order in 'blp' = 2257<==== must be wrong,should be 2022
size in 'blp' = 29024 <blp says that this file is unpacked>

Seems that 2022 is not in ftlinkdict.
but if i don't make this words work:

if filetableindex in ftlinkdict :

everything is great??? why??? OTL...

[Lilstormcloud]

"(╯‵□′)╯︵┻━┻"

Lol that's awesome.

*

Pro,
replacing saves time but beware, you still need to write in vertices manually.

you know how when you close gaps between two parts that are supposed to be attached and move in sync?
well if the two parts have a mismatch in number of vertices on the edge, a visible gap will open when the part moves.
that's why you still need to write in more, or remove extras.

[Rosalin]

@tcl36478655
That's wonderful work.

***

I was able to convert vt(UV) also.
it's based on IEEE 754 and half size of single floating.

http://en.wikipedia.org/wiki/Half-preci ... int_format


***

When adding extra vertices.. it seems that setting weight is most annoying part.
now normals can be calculated. UV can calculated.
but weight is not only about calculation you know.
well.. I don't want to think about that.

[b0ny]

any advance with DOA5 motions??

nope

When adding extra vertices.. it seems that setting weight is most annoying part

how many vertices are we talking here? you can test the weights in blender for example, by setting the weight values and bending the joint to see how it goes.

tcl36478655
this script is a remake of doa5 vanila which is a remake of ninjagaiden3 unpacker. it needs some heavy refactoring, but i'm not that type of person who does this sort of things

[Darko]

any advance with DOA5 motions??

nope

When adding extra vertices.. it seems that setting weight is most annoying part

how many vertices are we talking here? you can test the weights in blender for example, by setting the weight values and bending the joint to see how it goes.

tcl36478655
this script is a remake of doa5 vanila which is a remake of ninjagaiden3 unpacker. it needs some heavy refactoring, but i'm not that type of person who does this sort of things

Hehehehe don't worry, thanks for the answer.

[tcl36478655]

@Lilstormcloud

@Rosalin
THX
@b0ny
You have already gave much more help,thank you
*
I did a little edit so this script solve my problem.ONLY the problem I care.
so it may lead to other problem.

[Rosalin]

I actually made a script for catching multiple vn from OBJ file and injecting back to TMCL.

It looked it's working. but it didn't worked right for some parts.
especially in some narrow creases. and I couldn't figure out why.

I'm just guessing that they are actually getting normals from some more detailed high polygon models or something like that.

If it worked, I was about to do some hardcore mesh editing based on the script.
But well.. there's something I can do with it still though..

@b0ny
never mind. that was just pointless thinking.
I can actually do for some few in 3dsmax.

[Rosalin]

Now I have a serious question.
Can anyone teach me about how index buffer organizes faces?

I was analyzing some parts in body.
but I couldn't understand orders in some parts.

index buffer :
7733 7674 7734 7673 7735 7671 7736 7670 7737

Look around 7735 plz.
they're totally jumping around vertices without following the line.
So if I change 7735 into FFFF (I was about to delete it), then there happens concatenative loss of faces.
What lil told me worked for me. but it seems there're some another rules behind.

Could it be explained in simple ways?

[tcl36478655]

Now I have a serious question.
Can anyone teach me about how index buffer organizes faces?

I was analyzing some parts in body.
but I couldn't understand orders in some parts.



index buffer :
7733 7674 7734 7673 7735 7671 7736 7670 7737

Look around 7735 plz.
they're totally jumping around vertices without following the line.
So if I change 7735 into FFFF (I was about to delete it), then there happens concatenative loss of faces.
What lil told me worked for me. but it seems there're some another rules behind.

Could it be explained in simple ways?

strange the order which index buffer is don't match that picture.
the index buffer should means that:
face1:7773 7674 7734
face2: 7674 7734 7673
face3: 7734 7673 7735
...
facen-1: 7671 7736 7670
facen: 7736 7670 7737

I also have a question maybe you already know.
Did DOA5U use another zlib for compress?
Because I try to use Quickbms for compressing but error.