Just a correction: the step size for the array index is adjustable..file1.zip

Edited March 19Mar 19 by KuWuniss6

How about this circle? (using triangle strips)

The face index data has to be accessed like so:

 

Edited March 13Mar 13 by shak-otay

13 hours ago, shak-otay said:

How about this circle? (using triangle strips)
这个圆圈怎么样？（使用三角形条）

The face index data has to be accessed like so:
人脸索引数据的访问方式如下：

 

Sorry, I didn't quite catch your meaning. I did some small tests, and when this 3D model is built using strips for the faces, it seems very messy and inaccurate in MR.

I'm not sure how you managed to link the outer faces. Also, there seem to be many unused face indices left in the face index data area, which is strange.

3 hours ago, KuWuniss6 said:

 I did some small tests, and when this 3D model is built using strips for the faces, it seems very messy and inaccurate in MR.

 

You can't use MR for faces here without code (afair there's a version that supports scripting).

When you're a coder (C or C#) and check the 2nd picture in my previous post you can rather easily get the faces.

(If you uploaded another sample I could try that.)

Quote

Also, there seem to be many unused face indices left in the face index data area, which is strange.

I used this obj file and if you rotate the circle backwards you can see some odd shape:

test.zip

Edited March 14Mar 14 by shak-otay

quick research

fmt_cmd.py

On 3/14/2026 at 7:49 PM, Durik256 said:

quick research   快速研究

fmt_cmd.py

Sir, thank you very much for your analysis. Based on your structure, I have made a new discovery, and perhaps we need to revise our reading rules.

When testing some models, they work normally, but some models (such as this arrow model) fail to load correctly. This suggests there are still parts of the data structure we haven’t got right.

In addition, I have found 4 unknown bytes, which should play a very important role. I have attached a new case here.arrow.7z

Edited May 19May 19 by KuWuniss6

On 3/14/2026 at 7:49 PM, Durik256 said:

quick research   快速研究

fmt_cmd.py

 

Sir, after a long period of comparison and analysis, we have made a brand-new discovery. However, we have not yet been able to figure out the structure of the geometry and face indices.

Edited May 19May 19 by KuWuniss6

After another round of research, I’ve made a new discovery regarding the surface index section. It appears to be a variable structure rather than a fixed one, though it follows data alignment. The key factor influencing the surface index step size lies in the data’s leading information index area, and I’ve successfully obtained a complete model. While errors still occur with other model files, there remains a significant amount of unknown data—perhaps this is the data we’re looking for. I will continue my research and keep you updated.

 

Edited March 19Mar 19 by KuWuniss6

On 3/15/2026 at 4:00 PM, KuWuniss6 said:

.arrow.7z

It's impossible to make a plugin with just one sample.

Edited March 19Mar 19 by Durik256

You're right; I've figured out the face indexing pattern now. However, I still have no clue about the UVs and normals—their storage order seems to vary as well. I'm currently looking for other CMD files to compare them with, and I'll update this post once I find new clues.

52 minutes ago, KuWuniss6 said:

You're right

I mean you need to share a lot of samples... oh well, never mind.