Model Optimizer and Converter v1.3.0

[loktar]

after optimize a model, the program set first bone's position to zero, please fix this bug.

Hmm yeah this happens sometimes, I'm not sure when/why but it doesn't always happen IIRC. But I'm not planning on updating this anymore, sorry. Thanks for the rating!

 

[loktar]

Three things:

• I MIGHT update this in the future, but I'm not sure yet.
• The code for version 1.2.1 has been lost somehow (guess I forgot to commit it to GitLab, and when I switched computers I got it from GitLab instead of moving it from the old computer - I still have an old laptop and an old PC, I'd probably check those first), so potential next version will probably be lacking its updates. I would leave 1.2.1 in the bundle until all its updates have been reimplemented.
• I think I've found the cause of the bug where the first node loses its pivot point (coordinates), in model.cpp (line 651) I have if (newObject->objectId > 0 && newObject->objectId < static_cast<int32_t>(pivotPoints.size())) where the first condition should be newObject->objectId >= 0! Oopsie!

 

[loktar]

Some caveats to the above:

• It's been over five years. I've basically forgotten some of the specifics of C++, most notably memory management, typecasting, and more. As such, I no longer understand much of my own code.
• By the same token, I no longer really understand the ins and outs of the MDX/MDL formats, contributing to my lack of understanding of my own code.
• I also don't remember the nature of the changes made in 1.2.1, so if that's truly lost, I might not be able to reimplement those fixes/additions.
• My primary objective, if I do choose to make an update, is to add support for v1100 and v1200 models, and secondarily to fix bugs. As such, new options are unlikely to be added. I would do my best to at least re-add the options added in 1.2.1.
• The "Delete Same-ish Keyframes" will either be removed (if I do release a new version), or be replaced with a (functioning) "Delete Identical Keyframes" option.

 

[Dartz]

It even removes overexposed vertices without generating errors, what a brilliant tool man

 

[loktar]

It even removes overexposed vertices without generating errors, what a brilliant tool man

Thanks What do you mean by overexposed?

 

[Dartz]

Thanks What do you mean by overexposed?

Most Warcraft III models have multiple vertices positioned at the same coordinates (X/Y/Z). While this can make some aspects of modeling easier, it unnecessarily increases the final file size. When removing these vertices manually, one by one (a really long and boring task), it is often necessary to redo the UV mapping. However, MOP automatically eliminates all these overlapping vertices without requiring the UV mapping to be redone, saving a lot of time and work, so... one more time, thank u so much buddy

 

[loktar]

Most Warcraft III models have multiple vertices positioned at the same coordinates (X/Y/Z). While this can make some aspects of modeling easier, it unnecessarily increases the final file size. When removing these vertices manually, one by one (a really long and boring task), it is often necessary to redo the UV mapping. However, MOP automatically eliminates all these overlapping vertices without requiring the UV mapping to be redone, saving a lot of time and work, so... one more time, thank u so much buddy

Ah, I see! My pleasure, I'm glad people find this useful

Edit: just checked old pc and laptop, looks like v1.2.1 source code is definitively lost

 

[loktar]

I'm so confused.

• Last commit on GitLab was June 20, 2020
• Files on my hard drive show last modified August 9, 2020
• The MDLX parser files on my hard drive are different than the ones on GitLab, but some diffs seem to be older and some seem to be newer?
• The MDLX parser code is suddenly showing a whole bunch of errors, which obviously wouldn't have been the case back then. Maybe it's because of the new Qt version or newer MinGW version?
• I tried running the code on my hard drive, and I got an error about an unknown animation tag on a model that works just fine in 1.2.1 - this should not happen on 1.2.0 at all and none of the diffs explain this!

What's going on xD Did I work on it not realizing I had lost the 1.2.1 code or something? Why is it not working as expected? WTF.
Edit: yeah it doesn't really look like I can continue this project, unfortunately maybe I can figure something out but don't count on it.
Edit 2: okay, I was able to make the parser compile by switching to MinGW 7.3.0 instead of the latest version, but now MOP itself is refusing to compile and I don't know why.
Edit 3: I'm installing Qt5 instead of the latest version, hopefully that will help. If so, I will probably work from the code on GitLab instead of whatever weird version I have on my hard drive.

 

[loktar]

Good news, I installed Qt5 and used MinGW 7.3.0 instead of the latest version and I was able to run the 1.2.0 code successfully and optimize a model. Hopefully I can add some features and fix some bugs. We'll see.

Edit: okay, despite successfully building, the parser is still showing a bunch of errors. Might have something to do with having two versions of Qt installed or something, idk.

 

[loktar]

Looks like I was right about the cause of the first node losing its coordinates... fixed! ^^ Still don't know why all the errors and warnings are popping up... it compiles and runs fine so I guess I can ignore them, but I really don't like it.

In any case, I won't further spam this thread, there's a changelog and to do list for those interested which I'll keep updated. Still not 100% sure this will result in a new release. But there's a chance it will, so if you have any bugs to report or features to suggest, now's the time

 

[loktar]

Hey everyone, I uploaded a beta version of v1.3 here. I was hoping some people could do some testing of the stuff mentioned in the changelog

 

[loktar]

This is by far the biggest update since I added MDX support! The full changelog is in the description, but here are some highlights:

• v1100 and v1200 support (thanks @Retera and @eejin!)
• Improved HD support
• Delete Identical Keyframes (formerly Delete Sameish Keyframes) now actually works as expected
• New options: Calculate Extents, Fix Negative Extents, Insert Missing Opening Tracks, Delete Unnecessary Keyframes, HD MDL Compatibility (RMS or Hive) and Convert Bones To Helpers
• Stop, pause, resume and skip buttons
• Option to recreate original folder structure
• Bug fixes:
  • First node no longer loses its pivot point/coordinates
  • Processing no longer continues after closing the program
  • Fixed memory leak causing memory usage to increase with each model processed (thanks @Retera!)
  • Fixed some potential crashes
  • Fixed bugs with several options

Please let me know if you encounter any bugs or if you have any ideas for more options!

 

[stan0033]

so you finally updated the tool. What motivated you?

 

[loktar]

so you finally updated the tool. What motivated you?

I'm not sure I'd been thinking about it for a while. One day I was looking through the source code and found the cause of the bug where the first node's pivot point was lost, and then I decided to fix it, and from there I couldn't help myself and did the other stuff too.

 

[Dartz]

I’ve noticed that during the optimization process, some models may develop small hitches or choppy movements in certain types of animations — but this issue can be easily fixed by increasing the Precision value from 3 to 5.

 

[PurrsandMlems]

awesome tool, recently used it with precision 5 to trim unused anims that hive's error checker reported on a model.
saved me an hour of doing it by hand, works like a charm.

 

[aloH]