Noesis g1m/g1t/g1a importer

[icoee]

Thanks for your script. But when I try to export the PC00B g1m file from Atelier Ryza, get an error: WARNING: Weight contains an out of range bone index. Discarding model. Please help!!!

Here is the file: https://mega.nz/#!89wykQwB!vi0v9NTKKmHv ... u6cNX-Kf4Y

[Joschka]

...

...

Great news, thanks for testing ! I'll try to make the optimization I talked about before to make the script run faster this week, the few Nioh samples I tried took quite a bit of time to load.

Maybe on koei fandom but even there titles seems to be missing, i also didn't notice that Personas are from Atlus while DQ builders is Square Enix despite using the same engine of many Koei games. Maybe there're some more lost in other company xD

Yep basically KT do their own games but also work with other studios sometimes, for example Hyrule Warriors and Fire Emblem for Nintendo and the others you mentionned. I actually had no idea that they made DQB2 before you mentionned it for example, thought the game was made by Square internally.
I think the best thing to do would just be to edit that wikipedia list at that point, it has most of the titles already.

Thanks for your script. But when I try to export the PC00B g1m file from Atelier Ryza, get an error: WARNING: Weight contains an out of range bone index. Discarding model. Please help!!!

Here is the file: https://mega.nz/#!89wykQwB!vi0v9NTKKmHv ... u6cNX-Kf4Y

Your model had some cloth type 2 with the new physics bones I implemented for Dissidia in the last updates. But I didn't know these meshes could have bone indices to the main armature too, which was the case in your model. I pushed a fix for that, should now work as expected.

[icoee]

...

...

Great news, thanks for testing ! I'll try to make the optimization I talked about before to make the script run faster this week, the few Nioh samples I tried took quite a bit of time to load.

Maybe on koei fandom but even there titles seems to be missing, i also didn't notice that Personas are from Atlus while DQ builders is Square Enix despite using the same engine of many Koei games. Maybe there're some more lost in other company xD

Yep basically KT do their own games but also work with other studios sometimes, for example Hyrule Warriors and Fire Emblem for Nintendo and the others you mentionned. I actually had no idea that they made DQB2 before you mentionned it for example, thought the game was made by Square internally.
I think the best thing to do would just be to edit that wikipedia list at that point, it has most of the titles already.

Thanks for your script. But when I try to export the PC00B g1m file from Atelier Ryza, get an error: WARNING: Weight contains an out of range bone index. Discarding model. Please help!!!

Here is the file: https://mega.nz/#!89wykQwB!vi0v9NTKKmHv ... u6cNX-Kf4Y

Your model had some cloth type 2 with the new physics bones I implemented for Dissidia in the last updates. But I didn't know these meshes could have bone indices to the main armature too, which was the case in your model. I pushed a fix for that, should now work as expected.

It works!! Thanks very much!
But now I found a ploblem for the hair of other model after export.

Here is the file : https://mega.nz/#!hlYG2aba!oKr4O4ARw1fa ... T5-Fr3-h4c

[howfie]

Excellent work guys! Will be sure to try it out! I see you guys actually figured out what the vertex shader was doing rather than trying to just interpret the byte code like I did .

[Yretenai]

Excellent work guys! Will be sure to try it out! I see you guys actually figured out what the vertex shader was doing rather than trying to just interpret the byte code like I did .

There's still a bit to go, normals are re-generated in the vertex shader in newer versions. It's causing a bit of a headache, but we're back to interpreting bytecode

[fullinu]

edit:
there was a problem here, it's gone now

[Joschka]

It works!! Thanks very much!
But now I found a ploblem for the hair of other model after export.

Looks like your were unlucky enough to find the 3rd NUN mesh that doesn't work among the hundred others that work fine. I guess that my cloth computation must be slightly off somewhere, fixing that in non trivial unfortunately and I don't want to do some hack that could break other models so better wait for us to update the equations when we'll add proper normal support for the cloth meshes.

Excellent work guys! Will be sure to try it out!

The gas machine man himself ! Really glad to see you here Semory ! :D
As mentionned in the credits all of this wouldn't have been possible without your gas machine source code as a basis, thanks a ton for your hard work on these formats ! (as you can see they're still milking their engine with only a few changes...)

I see you guys actually figured out what the vertex shader was doing rather than trying to just interpret the byte code like I did :mrgreen: .

Oh don't worry we still have almost no clue about what's actually happening, the shader code has just been converted to a more compact form hehe
I did find similar ideas with the control points and all in APEX clothing but I think they did their own solution. I'll probably look a bit more into it soon now that most bugs are solved, as you can see above there are still some rare cases where it gives some weird results.

What we have so far is that there's a computation of the center of mass of the 4 closest control points using the 4 components of the 4D vertices of the cloth meshes as weights for each one (done in what I've called processClothVertex). Then these centers of mass are themselves used in another weighted sum to get a final vector (so the center of mass of the center of masses) This is repeated 3 times to get all the vectors (d4,d5,d6)
Then these vectors are used in a scalar triple product and the resulting values are fed to the vector components of the position vector

[icoee]

Looks like your were unlucky enough to find the 3rd NUN mesh that doesn't work among the hundred others that work fine. I guess that my cloth computation must be slightly off somewhere, fixing that in non trivial unfortunately and I don't want to do some hack that could break other models so better wait for us to update the equations when we'll add proper normal support for the cloth meshes.

Ok, I get it now, thanks!

[Joschka]

Looks like your were unlucky enough to find the 3rd NUN mesh that doesn't work among the hundred others that work fine. I guess that my cloth computation must be slightly off somewhere, fixing that in non trivial unfortunately and I don't want to do some hack that could break other models so better wait for us to update the equations when we'll add proper normal support for the cloth meshes.

Ok, I get it now, thanks!

This custom script will fix your model, normals are not computed correctly though but this is easier to fix in a 3d software. Only use it for that model, it's just a quick and dirty hack until a proper fix.

[icoee]

This custom script will fix your model, normals are not computed correctly though but this is easier to fix in a 3d software. Only use it for that model, it's just a quick and dirty hack until a proper fix.

It's awesome!! Thank you so much!!!

[Joschka]

Pushed a quick fix for BE models. Fatal Frame models should now be fully supported, with cloth meshes fixed.
Still have no immediate plans for texture support since it's the only one using Wiiu swizzle, don't really want to implement that for just one game.

[Crazy31139]

Hi, i try open Noesis and see this error, maybe you can help me

[Joschka]

Hi, i try open Noesis and see this error, maybe you can help me

viewtopic.php?f=16&t=21666#p159400

[Joschka]

It's awesome!! Thank you so much!!!

I did more research and pushed a universal fix for that so you won't need the custom script anymore, just download the latest version and you should be good. Thank you for the bug report, made me aware of that error.
Just need to fix the normals properly and I think the script will be pretty much complete. After that I'll be able to focus on optimization.

All the models should now import correctly for all games, cloth/hair meshes included.

[shenglong]

Hello, thank you very much for sharing such an excellent tool.

I'm also analyzing G1M files, so I'd like to ask you some questions about the G1M file format:

1.What is the structure of the G1M file? (PC platform file).

2.The vertex data at the beginning of the 0x04000100 mark in the GM1G4400 block contains vertex structures described as 0x7C size. I guess these structures are cloth physical meshes, because I can't extract mesh data correctly, they always appear as flat meshes. How can I correctly analyze these data?

3.What is the structure of ONUN8200 and VNUN1100 blocks?

The above questions have perplexed me for a long time. I hope to get your answer. Thank you again!