XeNTaX

Renegaderobbie wrote:How do you convert .model files / .rended files?

With a tool that hasn't been released yet. Stop asking, you already caused enough drama on FacePunch.

Alexis wrote:

RosaWeyland wrote:

dibe91 wrote:I do not know if this is the place to ask, but I'd be interested in a Halo 2 anniversaly model that would be this:

I just wondered if it was possible using the Halo 5 instruments. unfortunately I do not know how to get the game where there is the model in question :/

Doubtful. We don't have the raw files for H2A (I believe), much less any way to get anything meaningful from them.

Going to check the collision and physics models when I get back, I have a strong feeling that the collision box will have the bounding box data in it.

If it's similar to previous Halo games, the bounding info has typically been found in the .render_model tag and/or the .model tags as it's generated when 343/Bungie process the model with their tools - As in within the tag data itself, not the resource chunks.

The current .module extractor is pretty crude, so it just dumps everything, destroying the structure of the tags that store pre-processed assets (Bitmaps, models).

TL;DR
Check the .render_model tag associated with the resource chunks of the model you're trying to extract.

Here's the actual tag structures from H5. Both tags have specific references to bounding offsets. I'm not sure how helpful it'll be... but here they are anyway.

So super quick question, what program can the tag structures be used in? I've tried using that to convert the .model file to an obj file with little success

Renegaderobbie wrote:So super quick question, what program can the tag structures be used in? I've tried using that to convert the .model file to an obj file with little success

None. The only real way to get data out is by using the hex data as shown in this thread already - at least, until a better tool to read the module formats is made. You'd know this if you did some research and stopped simply asking for handouts.

Oh, and by the way - if you're asking such basic questions as this, I'd be interested to know why you're telling me over on Facepunch that you have access to the UV and bounding box data.

RosaWeyland wrote:

Renegaderobbie wrote:So super quick question, what program can the tag structures be used in? I've tried using that to convert the .model file to an obj file with little success

None. The only real way to get data out is by using the hex data as shown in this thread already - at least, until a better tool to read the module formats is made. You'd know this if you did some research and stopped simply asking for handouts.

Oh, and by the way - if you're asking such basic questions as this, I'd be interested to know why you're telling me over on Facepunch that you have access to the UV and bounding box data.

I wasn't asking for hand outs just a way to do that. And I private messaged you on facepunch a while ago. I have the hex2obj program. Just having trouble reading the data

Renegaderobbie wrote:I wasn't asking for hand outs just a way to do that. And I private messaged you on facepunch a while ago. I have the hex2obj program. Just having trouble reading the data

Constantly asking for tools I made quite clear you wouldn't be getting is asking for a handout. And, if you want to figure out how to read the data would suggest reading the hex2obj thread a little more closely. I've read your PMs and your posts, and the answer is pretty simple:

No. I'm not giving you the tools I have, because they're not ready to be published. At all. You obviously don't have any idea of what you're doing, you don't have the UV and bounding box information, and the only way I'd even halfway believe you is if you sent over that supposed Warden Eternal file for me to look at. Since I'm expecting you not to be able to do that, I think it's pretty clear what the answer in return is going to be.

Stop being impatient, wait for the tools to be finished, and for the love of god please stop telling people that you're working with me or helping me.

And what is there a problem with the UV? It normally read as a 2 value type short. Or you're talking about the scale?

erik945 wrote:And what is there a problem with the UV? It normally read as a 2 value type short. Or you're talking about the scale?

Both, honestly. Haven't found the data for either, though if I at least could integrate UV data into my tool then I could publish the assets I have managed to extract with the texture data intact for others to use. Bounding box info could come after, probably.

fragment of my maxscript, read vertex data. You need values u_f and w_f
...and full script. Hasn't skin, scale and names but work for most weapons and vehicle models

erik945 wrote:fragment of my maxscript, read vertex data. You need values u_f and w_f

Code: Select all

x = readshort f__mesh #unsigned
y = readshort f__mesh #unsigned
z = readshort f__mesh #unsigned		
fseek f__mesh 0x2 #seek_cur		
u = readshort f__mesh #unsigned
w = readshort f__mesh #unsigned	
u_f = (1.0 * u)/65536
w_f = (1.0 * w)/65536

...and full script. Hasn't skin, scale and names but work for most weapons and vehicle models

Nice work.

One thing I've noticed is that your UVs seem to be flipped vertically. Is that something you can correct in your script?

Ok, no problem
If you get a model which does not work - reported.

erik945 wrote:Ok, no problem
If you get a model which does not work - reported.

Awesome, thanks for correcting the UVs!

So, I am getting a couple errors here and there with specific models. For example, spartan_vale's model gives me an error when creating "Mesh 60" and "Mesh 61". Judging by the face count, "Mesh 61 appears to be the highest LOD for Vales head model.

It also appears that the UVs for this model are also scaled incorrectly. They are scaled too small, and centered to the left of the UV space.

Here's Max's listener output for Vale.

Having issues with the Hydra as well - had the same issues with the model when I was pulling the file using my own tools as well, so I suspect it's something to do with a difference in the file. Seems I can only get the lowest LoD and one of the sub-HD LoDs, and a few decal patches.

Script uploaded below.

I think I've figured out a pattern for scaling the character models more or less, looks like they need to be scaled down to 75 on the Y-axis, and 30 on the X-axis. I've tried it out on a few characters and it seems to work fine.



Now weapons are a whole different thing, I checked out the Assault Rifle and the Magnum last night and they seem to both be scaled differently.

SpookyMajora wrote:I think I've figured out a pattern for scaling the character models more or less, looks like they need to be scaled down to 75 on the Y-axis, and 30 on the X-axis. I've tried it out on a few characters and it seems to work fine.



Now weapons are a whole different thing, I checked out the Assault Rifle and the Magnum last night and they seem to both be scaled differently.

According to my math, assuming the box at the bottom is supposed to be a perfect cube, it should be 75.368331922% on the Y-Axis and 25.208156329 on the X-Axis.

So, you're not too far off. (Unless my math is off!)

Alexis wrote:According to my math, assuming the box at the bottom is supposed to be a perfect cube, it should be 75.368331922% on the Y-Axis and 25.208156329 on the X-Axis.

So, you're not too far off. (Unless my math is off!)

I think it might be. I used your values on the Helljumper helmet when compared against the Halo Waypoint model rip, and there's a large amount of compression along the Y axis, as well as a difference in the X axis. I used the Z scale (height) to compare and contrast.

What I've ended up doing is exporting the model from Max with that weird cube on the bottom, and then scaling it against a spawned cube mesh in Blender. It's not ideal, but until bounding box data is implemented I can make do with manual scaling. I'd show my maths but... I'm going to be the first one to admit that working out ratios and whatnot is not my area of expertise. Basic maths I can do, anything more is a whole bunch of 'nope'. I can post my numbers tomorrow and let you guys work it out if you like?