RE outbreak textures

[BeeswaX]

i extracted some of the textures from outbreak they were in .sld format which was just a renamed version of a tim2
however there is some damn compression plz someone help me with this

[Rheini]

Well compression is pretty hard to detect.
Can't recognize any technique known to me. Might be a special texture compression.
Best way is to disassemble the exe. Maybe you can upload it and the necessary dlls.

EDIT: Of course only if this is a PC game.

[BeeswaX]

Well compression is pretty hard to detect.
Can't recognize any technique known to me. Might be a special texture compression.
Best way is to disassemble the exe. Maybe you can upload it and the necessary dlls.

EDIT: Of course only if this is a PC game.

nope its a ps2 game

[Rheini]

Hmm indeed ps2 disassemblers exist. But I don't know how powerful they are. Can you provide the main executable file?

[BeeswaX]

Hmm indeed ps2 disassemblers exist. But I don't know how powerful they are. Can you provide the main executable file?

would that be the .elf ?

[Rheini]

Yes I guess.

[BeeswaX]

i dunno which of these is the usefull one so i packed all files up that could be it
anyway thnx for trying to help me
http://tjbp.net/upload/obstuff.zip

EDIT:
i talked with alera, he thinks its a simple RLE compression

[Rheini]

I don't think so. RLE would prevent long runs like EB EB EB EB or FA FA FA that are present in the file.

[BeeswaX]

I don't think so. RLE would prevent long runs like EB EB EB EB or FA FA FA that are present in the file.

hmm any other ideas then ?

[Rheini]

Maybe it's some kind of LZ algorithm (look at that TIM2 being uncompressed), but doesn't seem to be LZSS.

[alera]

Although the file may not be stricty RLE it does indeed use the technique.

during compression it identifies runs larger than 4 bytes(?) and compreses them into 2 or 4 byte instructions that basicaly fetch the data from a sliding window. I still don't understand it completly but I am positive it works this way, the instructions is what I don't understand. I cna barely fit it in my head and eventualy sometimes it fails when I try to uncompress small blocks manualy to see if I can understand it.

Every 32 bytes is the end of a block and inside this blocks is the instructions encoded as something similar to literal index(first byte) and runlength(second byte) there seems to be more data encoded in the 2 byte intstruction.

instructions look similar and can be put togheter, I can't figure out with precision how does the decompresor tels what is an instructions and what is data.

the instructions had a maximum runlength(I don't remember how much) so it is normal to see things like fetch[2]run[8]fetch[2]run[8] that explains the runs you mentioned. the minumum literal was like 4 bytes and it then copied those 4 bytes for the run length.

Attached is a picture that if I remember correctly is the compresseed and uncompressed version, one in sld and the other one in dds format so only the data is the same and not the header(forgot the size of the dds header).

BTW the compression achieved is negletable, hence why most files are similar in size to the could be uncompressed version(say always near 65k 16, 32k in size)

I managed to manualy decompress some blocks with my used to be knowledge :P but aparently I dind't writ eit down.. omg I cna't type

BeesWax: Ignore the file I sent you, data there is incorrect.

[Rheini]

So that 41h at the beginning somehow denotes a section with uncompressed data, in this case the FourCC "TIM2".

[alera]

Yes it does, all files start with that or 44h I believe :) found some notes:

***************************************
Compressed picture
4104 5449 4D32 0400 0100 0000 0118 3004
UKID ITVL
0100 0004 0510 0100 3000 0001 0001 0305
ITVL

UK = Unknown
ID = counter?
LTR? = Literal? char to copy?(2 bytes)
IT = Instruction
VL = Value(num times)

Value seems to be 8 bits(shift 3 bits)
11111000
1 = repeat count
0 = unk (flags?)

The IT(Instruction):
The it has the number of spaces to go back in the first 5 bits
count 2 bytes before that position to get the literal to copy for the run.
flags(last 3 bits) are still unknown.

if there is no count in the literal or the value then delete and do nothing.
maybe the literal must also exceed the 32 bytes(or 12) or something.

- Spacing is 32 bytes(maximum length of compressed output.)

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

ROLF and don't ask me what it mean! I can't understand what I wrote and is probably incorrect as I write what I discover as I go. the file I recovered from a dead file system and could not retrieve the time stamp so I don't know if it is the latest >.<

Just remembered: that data is wrong, back then I though there where 2 types of data one that was UKID and the ITVL but they are actualy both the same ITVL - instruction and value(2 bytes) is just I didn;t understand the part with the header.

Edit:

At 0x54 in yokotex1.sld and 0xCC in the uncompressed file

Compressed:
0208 080F 0600 1318 0101 0118 0303 0103 0210

ITVL = 1318, 0118, 0210

1318: don't get it but has a run of 3

0118: go back 1(2 byte pairs, first entry in index?) repeat 3 times

01 index number

18 = 00011000 shifting to the right 3 times gives us 00000011(3)

0210: select 4 bytes as the literal... ? copy 2 times

the uncompressed file shows this:

0208 080F 0600 0000 0000 0000 0101 0101
0101 0101 0303 0103 0303 0103

it is pretty close :)

IT must be 4 bits long since it uses pairs of 2, if it was 5 bits the IT will be able to go before the begining of the block that is 32 bytes.

instruction 1318 goes beyond the start of the block... to get 0s? I really don't get it.

[BeeswaX]

so any news ?

[Jamesuminator]

Yes it does, all files start with that or 44h I believe :) found some notes:

***************************************
Compressed picture
4104 5449 4D32 0400 0100 0000 0118 3004
UKID ITVL
0100 0004 0510 0100 3000 0001 0001 0305
ITVL

UK = Unknown
ID = counter?
LTR? = Literal? char to copy?(2 bytes)
IT = Instruction
VL = Value(num times)

Value seems to be 8 bits(shift 3 bits)
11111000
1 = repeat count
0 = unk (flags?)

The IT(Instruction):
The it has the number of spaces to go back in the first 5 bits
count 2 bytes before that position to get the literal to copy for the run.
flags(last 3 bits) are still unknown.

if there is no count in the literal or the value then delete and do nothing.
maybe the literal must also exceed the 32 bytes(or 12) or something.

- Spacing is 32 bytes(maximum length of compressed output.)

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

ROLF and don't ask me what it mean! I can't understand what I wrote and is probably incorrect as I write what I discover as I go. the file I recovered from a dead file system and could not retrieve the time stamp so I don't know if it is the latest >.<

Just remembered: that data is wrong, back then I though there where 2 types of data one that was UKID and the ITVL but they are actualy both the same ITVL - instruction and value(2 bytes) is just I didn;t understand the part with the header.

Edit:

At 0x54 in yokotex1.sld and 0xCC in the uncompressed file

Compressed:
0208 080F 0600 1318 0101 0118 0303 0103 0210

ITVL = 1318, 0118, 0210

1318: don't get it but has a run of 3

0118: go back 1(2 byte pairs, first entry in index?) repeat 3 times

01 index number

18 = 00011000 shifting to the right 3 times gives us 00000011(3)

0210: select 4 bytes as the literal... ? copy 2 times

the uncompressed file shows this:

0208 080F 0600 0000 0000 0000 0101 0101
0101 0101 0303 0103 0303 0103

it is pretty close :)

IT must be 4 bits long since it uses pairs of 2, if it was 5 bits the IT will be able to go before the begining of the block that is 32 bytes.

instruction 1318 goes beyond the start of the block... to get 0s? I really don't get it.

Hmm, i'm sort of way over my head, but i like math :P.
I'm thinking the instruction is actually 4 bit values and not a byte.
let me explain

Compressed:

Code: Select all

0208 080F 0600 1318 0101 0118 0303 0103 0210

Code: Select all

ITVL = 1318
what if it's actually: 

true is 0 cushion, false is repetition = 1 /*from what I've seen, consistent*/
how many pairs of bytes = 3    /*consistent*/
loop how many times = 1  /*consistent but fishy, again i'm over my head*/
how many bits?* = 8   /*see below*/

*I'm not sure about this one, by this i mean, how many bits it's reading, if we have two bytes that are this
0000 1111 0000 1011, the first 4 bits of each byte is unused, maybe it cancels it out?
unlikely seeing as the 0101 is kept the same, but still plausible

Uncompressed:

Code: Select all

0208 080F 0600 0000 0000 0000 0101 0101
0101 0101 0303 0103 0303 0103

Everything i said is consistent for what i've been supplied with :\.
Feel free to correct me if i'm wrong :)

Also, sorry if i confuse anyone, i have a poor way of explaining myself.