HOME | DD

Tung-kun In reply to vasilnatalie [2016-09-21 18:37:26 +0000 UTC]

Thanks! I'll give you the link here: sta.sh/01kdl6r6oc7q

There might be the same glitches on my other models, so feel free to check them if you like.

👍: 0 ⏩: 1

vasilnatalie In reply to Tung-kun [2016-09-21 19:51:41 +0000 UTC]

Your knees actually look pretty good with manual animation in both PMXE and MMD.  It's normal that existing motions need to be adjusted to the model.  What I'm seeing is that the model isn't having any weird knee motion until the model's feet are well into his butt.  Like in the picture above.  That's an issue you're going to see with any MMD model, and it doesn't matter, because nobody animates feet into butts anyways.

You do have some weight issues around the knees, but I doubt that's what you're talking about.  At sharp knee bends, you can see that the sides of the knees get pointy.  I believe that this is from the calf weights extending too far up the thigh, particularly in the back of the thigh.  Calf weights should extend only up to the knee bone.

The place with obvious problems is the feet.  The ankle bones are placed way too high, which leads to a really bad looking foot, particularly at sharp angles.  You're also doing really weird things with your feet bones, with strange rotation appends and doubled bones.  This is advanced stuff, and if you don't understand the skeleton you're using, I'd recommend that you prune it to the simplest possible.  You can try to learn this more complicated stuff later, when you're more comfortable with bones.  Right now, you've got a strange and complicated system of appends, parents and weights that's actually stretching and compressing the foot as the ankle rotates.

👍: 0 ⏩: 1

Tung-kun In reply to vasilnatalie [2016-09-21 21:57:04 +0000 UTC]

It's good to hear that at least the knees aren't looking too bad, because I previously had plenty of issues with the knee bending correctly. Some of the weight issues also come due to the fact that the ripped models are actually pretty low poly and I increase the vertex count with this plug-in: bowlroll.net/file/73529

And without a suitable base to look at (at least I haven't found a decent game rip model from GTA or something like that) it would be harder to rig the bones correctly.

I took a closer look at the feets and it indeed has its issues. Thanks for pointing the mistakes out. I'm still pretty a newbie to MMD, so any help or useful suggestions is highly appreciated.

Btw do you probably have a good base model which might work with game-rips? I'd like to take a look at it.

👍: 0 ⏩: 1

vasilnatalie In reply to Tung-kun [2016-09-21 22:16:00 +0000 UTC]

I'm not sure what you mean or want from a base model.  If you want a template for learning MMD bones, a structure to copy, take a look at Animasa's Miku that comes with MMD.  This is as simple as it ever gets.  (Add semi-standard bones from the plugin if you want.)  In order to work with MMD, you don't need more bones than that until there's something specific and unusual that you want to do.

You might consider if you actually want to increase the vertex count.  Any automated system of increasing vertex count is going to come with as many problems as it solves.  Try comparing a base-count model to a subdivided (more vertices) model sometime, side by side, and see which one actually looks and moves better.  I used to think that lots of vertices were important for detail; after doing a lot more modelling, I now believe that a good modeller is one who tends to use fewer vertices.

👍: 0 ⏩: 1

Tung-kun In reply to vasilnatalie [2016-09-23 11:21:24 +0000 UTC]

What I meant with a base model is a model which has no clothing, so you can take a look at the model how the bones are supposed to be rigged (I tend to follow Wing's Helpful Rigging Guide before, but this is not exactly correct for game rips, so I have to guess by myself). Base Models are not to be confused with Nude models, which have all anatomical functions (and can be a hindrance due to probably hiding some vertices which need to be properly rigged).

Increasing the vertex count the first time for Bully models seem to be necessary due to the fact that the models are very low-poly (which makes it awkward to rig and animate properly, but I agree that there shouldn't be any unneeded vertices and will take into consideration about how many vertices I should use.

👍: 0 ⏩: 1

vasilnatalie In reply to Tung-kun [2016-09-23 15:22:12 +0000 UTC]

I think I see what you mean.  No, I don't have something off-hand.  But you should be able to make something relatively easily, maybe out of a nude model (doesn't matter if it's got any interesting creases, those are almost always 100% weighted to the lower body bone anyways).  Most of the process is just the deletion of unnecessary bones, and doing that is a good exercise to get more familiar with the bone structure.  Probably by the time you'd made a template like that, you would no longer need it   Animasa Miku would still be a good choice, for how much she's defined the default MMD bone structure.  If a bone's weighted to the skirt or tie, just delete that bone, and delete the skirt/tie.

There's no single good rigging technique for game rips.  For instance, DOA models have a lot of weights that most MMDers don't know what to do with.  DOA rippers tend to drop the weights from multiple bones onto single bones, but that's kind of a waste of the excellently rigged DOA models.  Other games' models aren't rigged well enough, and rippers have to decide whether to go to the trouble of creating new weights to improve motions, or just leave the models awkward.  But if you're ripping a model for use with MMD, then it's probably to take advantage of the library of user-created motions, so you want Miku compatibility.  Trying to emulate Miku's structure as closely as possible is never unwise.

👍: 0 ⏩: 0