New LTB to LTA -> Model_Unpacker

[y41161882]

I tried to, but for some reason it drops connection around 35mb every time

Maybe you can upload it to your website and I'll find it by myself.

[y41161882]

@supercarz1991
When I was just having dinner, I realized a important thing. the ST_M_Child.ltb, which is a child model. It only houses some animations for models. If you just import the child model into the game without a main model which has mesh and skeletons, the model will not show up in game.
I don't know if it's because of the conversion. You can try to use the original model in the same way to test it.
The Model_Unpacker not only supports CF but also CA, so I'd be happy to solve your problem.
All roads lead to Rome. I have already got the ST_M_Child.ltb from my friend, Tupac. But there are three models whih the same name. The file sizes are 6.39 MB, 17.2 MB, and 14.5 MB, respectively. Which one do you mean?

[supercarz1991]

@supercarz1991
When I was just having dinner, I realized a important thing. the ST_M_Child.ltb, which is a child model. It only houses some animations for models. If you just import the child model into the game without a main model which has mesh and skeletons, the model will not show up in game.
I don't know if it's because of the conversion. You can try to use the original model in the same way to test it.
The Model_Unpacker not only supports CF but also CA, so I'd be happy to solve your problem.
All roads lead to Rome. I have already got the ST_M_Child.ltb from my friend, Tupac. But there are three models whih the same name. The file sizes are 6.39 MB, 17.2 MB, and 14.5 MB, respectively. Which one do you mean?

The original ST_M_Child ltb file is 65mb, when converted to LTA with your tool, it comes out to 240mb. With the original LTB2LTA, there's 1543 animations with 53 bones each, but apparently Valofe broke something as the LTA comes out to only 1meg lol. With yours, the file size looks right at over 235megs for the LTA, but compiling back to LTB it drops to 25 megs instead of close to 65 like its original.

When we USED to mod this LTB file back in the day, it was for vJump, glitch key and jump key, which I'll explain below for those curious. When we would convert it to lta, mod it, then back to ltb, we didn't have to do anything extra. We'd just compile and soldiers in game still worked. Something somewhere either changed with the model or changed with the ltb to lta conversion

For those curious, vjump was done by modifying the Crouch animations Y Dimension to 150, which would raise the player camera up higher rather than "crouching down." From here, you could still shoot players. Great to combo this with no recoil and no spread hacks.

Glitch Key was another modification of crouching. It would make the XYZ Dimensions of the Crouch animations 1, which would let you move through parts of the map and get out of bounds

Jump Key was essentially fly mod. There's a time limit on the animation obviously. We'd modify the standing jump animations Y dimension to 10,000 and then start jumping. When the animation time would end, you'd be floating. Other players could see this. I have a video somewhere of me way up in the sky and getting shot from the other side of the map with bullet tracers, which proved you are essentially flying with mods

[y41161882]

When we USED to mod this LTB file back in the day, it was for vJump, glitch key and jump key, which I'll explain below for those curious. When we would convert it to lta, mod it, then back to ltb, we didn't have to do anything extra. We'd just compile and soldiers in game still worked. Something somewhere either changed with the model or changed with the ltb to lta conversion

For those curious, vjump was done by modifying the Crouch animations Y Dimension to 150, which would raise the player camera up higher rather than "crouching down." From here, you could still shoot players. Great to combo this with no recoil and no spread hacks.

Glitch Key was another modification of crouching. It would make the XYZ Dimensions of the Crouch animations 1, which would let you move through parts of the map and get out of bounds

Jump Key was essentially fly mod. There's a time limit on the animation obviously. We'd modify the standing jump animations Y dimension to 10,000 and then start jumping. When the animation time would end, you'd be floating. Other players could see this. I have a video somewhere of me way up in the sky and getting shot from the other side of the map with bullet tracers, which proved you are essentially flying with mods

It looks like my tool can work normally.

The original ST_M_Child ltb file is 65mb, when converted to LTA with your tool, it comes out to 240mb. With the original LTB2LTA, there's 1543 animations with 53 bones each, but apparently Valofe broke something as the LTA comes out to only 1meg lol. With yours, the file size looks right at over 235megs for the LTA, but compiling back to LTB it drops to 25 megs instead of close to 65 like its original.

It's normal. There's something wrong with algorithm of Model_Packer (LTA back to LTB). To save memory for the model, the tool detects the bones that don't move in the animations and saves them as still bones. These bones have only one frame of animation. This greatly reduces the file size.
How does the tool check? It does this by determining whether the values of position or rotation are equal. But there is something wrong with algorithm of the tool. It checked first and then compressed the values. Some of values are pretty close instead of equal, which I'll explain below for these values. This causes the bones to be judged as moving. After the values would be compressed by lossy algorithm, However, the values are actually equal! The data should be compressed, but it isn't.(I haven't studied the subjunctive, so forgive me for using the present simple tense)

For instance:
original position1: 0.000000 0.000000 0.000000
original position2: 0.000001 0.000001 0.000001 (not equal)

compressed position1: 0.000 0.000 0.000
compressed position2: 0.000 0.000 0.000 (equal)

The correct way is to compress and then compare. But by the wrong algorithm, although the value is the same, it is judged to be different.

Now you can see the problem. Here's the interesting part. lol
The Model_Unpacker will decode the compressed position 0.000 to 0.000000 as original position and save into the lta file. if we compile the lta back to ltb, the values will be juged as equal by the tool so that the tool can compress it even less in memory.

To sum up, "The LTB drops to 25 megs" is normal. Instead of loss of date, the tool can correctly unpack models. Don't worry it.

Do you have any questions else? Feel free to contact me.

By the way, I'd like to ask you a question. What does the following code mean in the LTA?
(anim-binding
(name "idle_0" )
(dims
(2.000000 2.000000 2.000000 )
)
(translation
(0.400000 -0.500000 0.100000 )
)
(interp-time 0 )
)

[supercarz1991]

It looks like my tool can work normally.


It's normal. There's something wrong with algorithm of Model_Packer (LTA back to LTB). To save memory for the model, the tool detects the bones that don't move in the animations and saves them as still bones. These bones have only one frame of animation. This greatly reduces the file size.
How does the tool check? It does this by determining whether the values of position or rotation are equal. But there is something wrong with algorithm of the tool. It checked first and then compressed the values. Some of values are pretty close instead of equal, which I'll explain below for these values. This causes the bones to be judged as moving. After the values would be compressed by lossy algorithm, However, the values are actually equal! The data should be compressed, but it isn't.(I haven't studied the subjunctive, so forgive me for using the present simple tense)

For instance:
original position1: 0.000000 0.000000 0.000000
original position2: 0.000001 0.000001 0.000001 (not equal)

compressed position1: 0.000 0.000 0.000
compressed position2: 0.000 0.000 0.000 (equal)

The correct way is to compress and then compare. But by the wrong algorithm, although the value is the same, it is judged to be different.

Now you can see the problem. Here's the interesting part. lol
The Model_Unpacker will decode the compressed position 0.000 to 0.000000 as original position and save into the lta file. if we compile the lta back to ltb, the values will be juged as equal by the tool so that the tool can compress it even less in memory.

To sum up, "The LTB drops to 25 megs" is normal. Instead of loss of date, the tool can correctly unpack models. Don't worry it.

Do you have any questions else? Feel free to contact me.

By the way, I'd like to ask you a question. What does the following code mean in the LTA?
(anim-binding
(name "idle_0" )
(dims
(2.000000 2.000000 2.000000 )
)
(translation
(0.400000 -0.500000 0.100000 )
)
(interp-time 0 )
)

So the issue with it still is though that compiling it back to LTB after using Model_Packer is that no player animation happens in game. No player models show up at all. You end up a 1x1x1 (XYZ) dimension model that's just a place holder for if no proper model is found. Its the default basically. That's where the issue actually lies I believe. I've tried again with LTB2LTA, and while the models come out in LTA format like garbage (that's partially why we stopped using it), compiling it back to ltb with ModelPacker after modification, player models in game work.

As for that code from the LTA, I'm not sure honestly. From what I can tell... its an animation named idle_0 with a dimension of 2x2x2 (XYZ) but it rotates 4x-.5x.1 (XYZ). Not sure the interp-time 0 code though as I've never seen it before.

[y41161882]

As for that code from the LTA, I'm not sure honestly. From what I can tell... its an animation named idle_0 with a dimension of 2x2x2 (XYZ) but it rotates 4x-.5x.1 (XYZ). Not sure the interp-time 0 code though as I've never seen it before.

First of all, thank you for your answer.

So the issue with it still is though that compiling it back to LTB after using Model_Packer is that no player animation happens in game. No player models show up at all. You end up a 1x1x1 (XYZ) dimension model that's just a place holder for if no proper model is found. Its the default basically. That's where the issue actually lies I believe. I've tried again with LTB2LTA, and while the models come out in LTA format like garbage (that's partially why we stopped using it), compiling it back to ltb with ModelPacker after modification, player models in game work.

As I soon learn, although the models which are come out in LTA format with LTB2LTA like garbage, Them can work.
On the contrary, although the models which are come out with my tool look perfect, them can't show up in the game.
It seems that there may be something wrong with my tool. I really need your models to test it.

[supercarz1991]

First of all, thank you for your answer.


As I soon learn, although the models which are come out in LTA format with LTB2LTA like garbage, Them can work.
On the contrary, although the models which are come out with my tool look perfect, them can't show up in the game.
It seems that there may be something wrong with my tool. I really need your models to test it.

Here's just ST_M_CHILD models. the -modded is modified with your tool, the -original is obviously straight from a rez file and untouched

https://www.virustotal.com/gui/file-...Nw==/detection
https://virusscan.jotti.org/en-US/fi...job/msksgkddm5

@Ninja could you please approve? <3

[Janitor]

MODERATOR NOTICE

Approved

[y41161882]

Here's just ST_M_CHILD models. the -modded is modified with your tool, the -original is obviously straight from a rez file and untouched

I can't convert your model with LTB2LTA. My operation may be wrong. All that comes out is mesh and not contain any animation.

I have finished the third version of Model Unpacker (Beta 2.1). I solved the problem of a incorrect compression parameter. You can have a try. See if the problem is solved.

Rome was not built in a day, so I'll improve my tool step by step.

Virus scan:
https://www.virustotal.com/gui/file/...27c7/detection

Could you please approve it? @Ninja

[Janitor]

MODERATOR NOTICE

Approved

[supercarz1991]

I can't convert your model with LTB2LTA. My operation may be wrong. All that comes out is mesh and not contain any animation.

I have finished the third version of Model Unpacker (Beta 2.1). I solved the problem of a incorrect compression parameter. You can have a try. See if the problem is solved.

Rome was not built in a day, so I'll improve my tool step by step.

Virus scan:
https://www.virustotal.com/gui/file/...27c7/detection

Could you please approve it? @Ninja

Its awesome you've even taken this on. We few that still mod are always grateful for new things

As for the LTB2LTA, I'll have to find an older ST_M_Child then. It worked for me, but computers are weird...

[y41161882]

Its awesome you've even taken this on. We few that still mod are always grateful for new things

As for the LTB2LTA, I'll have to find an older ST_M_Child then. It worked for me, but computers are weird...

I say goodnight to you when I'm in the morning.

[y41161882]

@Flengo
Could you please answer me a French question?
Pardon，je veux vous poser une question. Avez-vous entendu le mot "autonomiser"ou“s’autonomiser”?

[[MPGH]Flengo]

@Flengo
Could you please answer me a French question?
Pardon，je veux vous poser une question. Avez-vous entendu le mot "autonomiser"ou“s’autonomiser”?

I don't speak French and please use English only on the forums.