Stage extraction script issues and to-do list

[KidWizardOfTheWeb]

https://github.com/KidWizardOfTheWeb/OllieKingRETools/blob/2cfba07e88c0765cf8a717b7670cd7c6292004f3/OKNoesisScripts/Experimental/Ollie%20King%20Stages.py#L51-L63

The loop is not entered due to entries not finding the correct model data to start at. Find the right starting point in order to proceed with the parsing loop.

[KidWizardOfTheWeb]

Reason found as to what's missing in order to get output. Player models had a specific point (0x8) in the header that detailed the value for table2, which is used to later locate the mesh starting point. Then there are a set amount of entries at 0x14 to control how many times the loop runs based on the model header. Since this script was based on the player script, the files are technically different in the composition by a small amount, and I overlooked the header until now, where the stage models do NOT have these in the header. Or at least, not in the same place. Note the differences, where the top image is the player model header (grinner.mdb) and the bottom image is a stage model header (s0_area01.mdb)

The next step is to locate where we can find this information. The header values in most stages are rather small and might require different math to get to the start of the mesh consistently. More research will be done soon.

[KidWizardOfTheWeb]

Need to adjust coordinates to XZY and then rotate the stage properly. Will implement in the future. We can also possibly mesh group start with 1000 0000 0000 0000. Now that I have GPA tools, it's significantly easier to read stage data since it exposes all of the submeshes and even textures loaded in emulator. I'll come back and update ASAP when I find a consistent pattern for it. Hopefully I might soon, my sights are on the emulation side of things at the time of this writing.

[KidWizardOfTheWeb]

Years later... As recent as commit c26b44d06dad0aea715b417a16b68e5ef6fafe63, automation of stage extraction is real.

To do at this point (ordered easiest to hardest, descending):

• [ ] Implement UVs.
• [ ] Implement normals.
• [ ] Cover meshes with <2 table entries (does not properly find the first mesh, should be an easy fix).
• [ ] Flip meshes to XZY orientation.
• [x] Separate submeshes by determining group ID (contained in first table, there's an identifier with each entry that looks like 06000040 or something similar. Each submesh is grouped in order by these identifiers from top to bottom of table, starting with the first identifier).
• [x] Reads from secondary table entries for face data parsing.
• [x] Read from all mesh group entries.
• [ ] Find those texture IDs after all this time.
• [ ] Put it all together for full extraction capability.

It's within my reach now. It's just gonna take some work. Hopefully by this midsummer.