Making Your First Model Part 22
Adding More Complex Facials 

As a note, I was able to get skirt physics to work wonderfully on a quick test. However, I was unable to achieve those same results on Camila. Taking advice from Bandages, I used a Solidify modifier to make Camila’s dress solid as opposed to a flat, two sided mesh. However . . . This did nothing of note. This may be due to bad settings on my part . . . Or bad body placement . . . But we are still working on a model with no skirt physics as of now.

Last time . . .

Last time, we went over how to add the easiest facials . . . The eyebrows! Once you get those eyebrows done, you feel like you’re the king of the world. You just made your first shape keys and they’re looking awesome!

Well, I hate to burst your bubble, but faces are hard!

The more high poly your face is, the harder it is to do anything with it.

So, let’s struggle through our first shape key for the eyes . . . The blink.

To see if it makes it easier, I’m hiding the irises so that I can use the proportional editing with “Connected” turned off. I don’t actually know if it will be helpful . . .  But I struggled to make anything half decent for the eyes with “Connected” on last time.

Let’s talk about C . . .

There are quite a few ways to select things in Blender. The most useful thing for shape keys is likely “C”. When you press C,  it allows you to paint your selection onto the model. It’s a bit confusing at first, however, as left clicking selects the area and right clicking exits the paint selection mode. Holding the wheel mouse while painting over an already selected area will deselect that area. (I think? It’s finicky when it comes to deselecting!)

Now, we’re already running into a tiny bit of a problem . . .

I’m not the best modeler. I’m probably not even a half decent one! Something that looked awesome stationary can look disastrous when moved. That’s the problem we have here.

Let’s try to make this in a more round-about way . . . First, I moved the eyeliner around so that her eye was covered with it.

Huh! That actually looks really cool! I’m saving it . . . but it’s not the “blink” that PMX particularly wants.

I was able to move things around to get a nice shape on one side . . . but not so much the other!

Now . . . How do we fix this?

Let’s talk about Stack Exchange for a second.

If you’re ever in a bind, you have three options. Sit there and try your hardest to fix it . . . Run crying to your nearest adult . . . or Google It.

With MMD and PMX, Google isn’t the most help. It’s not a very widely used software, and most people just use other people’s parts. I couldn’t find a tutorial on what I needed . . . because no one else needed it enough to ask someone who knows to make a tutorial. So, I ran to Bandages.

However, with something like Blender . . . Google is your friend. There is countless documentation online and thriving communities. For the most part, every question you will ever have in Blender has been answered five times in other places.

I wasn’t sure if it was possible to do what I wanted, so typed “delete half of shape key” in Google. I was frustrated and too out of it to tack on “In Blender”, but it worked! I found this result and almost hooted in celebration. It worked!!

While there are a ton of resources you can find, Stack Exchange is usually the best resource.

Once I learned how to reset that part of the shape key, I was able to make a new shape from mix, then mirror that. I named them Wink and Wink R . . . and by setting both of their values to one, I was able to make a new shape from mix and name it Blink!

Now, there are still several shape keys left for the eyes . . . But let’s skip to something more important!

The Mouth

Before getting started, let’s talk about lip syncing.

How do lip syncs work in MMD? If you’re extremely new, it’s as simple as dragging a VSQ into MMD. The VSQ will tell MMD when to trigger each movement.

I was going to show you an example of a VSQ and the resulting lip sync data side by side . . . But I have to use UTAU to make VSQs, and exporting a VSQ from UTAU results in the lyrics being replaced with question marks. whoops.

So, you’ll just have to take my word on this . . .

Lip Sync Generated from VSQ in Face and Lips

Using Face and Lips allows us to analyze the lip sync data. And . . . it’s pretty bad!

For shorter notes, there is an abrupt switch over from one vowel to another with literally no room for consonants. Near the end, I had “かんぼんそん R んあ” You’ll notice that in the data, there is no shape key for “n”. The only indication that “n” was there is the gradual increase in the final “a” as opposed to beginning suddenly.

I tested this on a model that does have “n”. There still was no “n” phoneme in the lip sync data. Why is that? Well . . . heh . . . You’re supposed to make models with their mouths closed. whoops. Simply closing the lips looks close enough to “n”.

The main problem with both Face and Lips and MMD in general is just that the only phonemes you need for lip sync are a, i, u, e, and o. (Note, these are using the Japanese pronunciations. Japanese i is “EE”. Japanese e is “eh”.) The irony of it? Most people outside of MMD will tell you that “a” and “e” look so similar, that it’s pointless to make two different versions. Just use the same one! Try it out for yourself – if you make a conscious decision to not move your lips, you can switch from “a” and “e” just by moving your tongue.

While it is possible to say “e” by baring your teeth . . . That’s kind of silly!

Here’s a head scratcher, though. The basic Miku model has no “e” shape key. However, she is able to sing “e” in Face and Lips. The “e” looks very similar to the “a”. A way to make an “e” from an “a” in Blender would be to set “a” to a value below 1 and then create a shape from mix. Easy!

So, now that we know about the vowels . . . Let’s point out that Face and Lips utilized exactly zero consonants. Even if it wanted to, most models don’t have shape keys for consonants.

Now, let me warn you – if you add shape keys for consonants, your motion data will look a bit like this:

Lipsyncing well is hard. It’s a puzzle that has you ask “What is realistic? What is not realistic but seems more realistic? What shape should this sound make?”

As far as what shapes you should have, a chart here makes it pretty clear. At most, you can expect to have seven shape keys for consonants without factoring in the tongue. If you do factor in the tongue, it’s quite a bit more. However, adding a separate shape key for the tongue would allow you to add that bit of added realism without making an insane amount of shape keys.

Taking cues from Flash animation, however, you only need four shape keys for consonants.

Whether you go the traditional route of no consonants or add consonants is up to you. If you add them, they’ll likely never be used by anyone except you. If realistic lip syncs are your goal . . . You need them. If you don’t care about lip syncs, ignore them.

Now . . . For actually making the keys!

Let’s start with “a”.

Remember – your top teeth are physically incapable of moving. Your upper lip is able to move about as much as it wants to, however! The upper lip moving around is extremely important for rounded vowels.

Rounded vowels are . . . Just kidding! I won’t bore you more with phonemes!

Now . . . What makes mouths so hard? The first is simply that you have to choose if you want a realistic shape key where the jaw moves along with the mouth or if you want a simple shape key where only the mouth is affected. Once you decide . . . Well, either you’re in for a nightmare of checking to make sure everything is moving naturally . . . Or you’re just going to change the mouth around and hope for good topology!

Which will I choose? The more realistic one, of course!

To do that, I simply selected part of the chin and used connected proportional editing to lower it. Then, I did the same to her bottom lip and her teeth so that they were lowered also.

However . . . I’m not particularly fond of it in side view . . . So I will redo it!

But if I made you stick around for that entire process, you’d be pretty bored pretty quick!

I’ll just finish up the shape keys . . . Then tune in next time for when we sanitize our display panel!