VAM Vam packages -repacks, cleanups, optimization, speed up. Need your opinion!

[mcwiki]

I do redux packages which speed up loading time of looks / clothing.. that part is going well.

However, it does not prevent my VAM from crashing ... probably due to 2000+ var in my AddonPackages. and long opening - hanging time!

Can anyone share some thoughts with me about VAM performance.
What is the best practice with plugins / morphs / assets in general

I'm planning to do some performance checks with HuaQ clothing set,
try to combine all (168) of the packages into 1, reduce textures.. compression. and I Will share my results here,
but please share some light.do you think any of these steps make sense?

All I was able to find is "having too many var files slows down or cause crashes." but nobody had a solid solution for it. (just reduce the amount)

 

[lbgtime]

Hello, I’ve seen your work, and it’s like a dream I’ve always imagined. After looking at your work, I realized that image compression is a very tedious task, but the workload is enormous. Do you do this manually, or do you use some kind of automation?

 

[mcwiki]

Hey, thank you!
I build a Python program that UNPACK -> recompress images -> REPACK .. so I can run it automatic.
The script need impovements(hence I'm not sharing YET!) ,
I'm learning file structure, file types.

for example I removed textures that were not in use... BUT 3rd party content when refer the package
it also refer specific texture!

So In the essence.. its save to unpack, rescale images and pack. BUT DONT REMOVE UNUSED textures

Right now:
Out of curiosity I've working on a script that will scan my SSD drive full of 1.5TB of vars...
and extract all references paths into CSV...

That way I will see what are the most popular , common assets.. and focus on optimization of them in a 1st place!

 

[mcwiki]

ALSO I think its important to REPACK var files without COMPRESSION. For faster reading it (unpacking on the fly in the game)
I've noticed that some vars are just "stored" -> without compression.
and some are actually compressed

 

[lbgtime]

I'm not the most tech-savvy person, but I think what you're doing is super cool! I'm rooting for you.

 

[animetiddyenthusiast]

if you have working python code that is just kind of slow I recommend giving it to github copilot or gemini and ask to translate to C#. I've done this and gotten insane speedups for large file processing. You can also go for cython or ctypes but that takes more manual work and AI can pretty efficiently translate between python and C# now