Home Artists Posts Import Register
The Offical Matrix Groupchat is online! >>CLICK HERE<<

Content

In this post, I wanted to go into how exactly I arrived at this peculiar art style I'm using for LLRR. I'll discuss both positives and negatives about it, as well as what I might do differently once this project is complete. This is within the context of Blender.

First, let's talk character creation.

Around 4-5 years ago, I wanted to figure out how to make anime characters in Blender. I'll go into my full motivations on this someday (hint: it was not to make erotica lol).

Being one human being, I decided it'd be best if I could save myself a lot of time using an addon. At the time, the most popular Blender addon for anime characters (and humans in general, I think) was Manuel Bastioni Lab, named for its creator. You can find it here.

Here's a sample of what the default anime characters look like (the user guide has a lot more info, including many realistic humanoid characters you can make):

Does anything look familiar?

I actually used an older version of the addon, made for Blender 2.7, so there were some small differences, but the above is very close to what the add-on made for me. Fun fact: I updated after making Amy and Flin, so all the other female characters have redder lips than Amy does because that's something the addon changed. I've personally never really been a fan of lipstick, so...I made sure the difference wasn't that big.

Anyway, I...did not really care for the style that much. But, it was still a lot closer to anime than any other addon at the time, and would save me a TON of time since the characters came out fully rigged with a wide array of shape key expressions (a "shape key," often called a "morph target" in other software, is a saved distortion you perform on a mesh, such as the grin in the above shot).

Anyway, so, I decided to make some adjustments. Thing #1 I did not like was the head and face shape. In particular, I struggled with creating characters that didn't look like teenagers. Take a look at how Amy used to look:

Like a little girl, right? Or maybe a teenager if you see the rest of her body, but I was aiming for a young adult. So, I stretched out her face to achieve that.
Granted, I think the "chestnut shape" of the face is a key part of charming anime design that I kinda butchered when I did that, but I also succeeded in making her look older. This is an example of something I will definitely work on once I'm done with LLRR. In fact, I daresay facial structure is the single most important thing I'll need to work on.

Here's the first render of the first character I made: Flin!

Oh, I made ALL the hair from scratch, by the way. It constituted essentially half of every new character's design time (which, to be fair, makes sense for anime. Hair is often the key distinguishing visual trait between anime characters). It is funny to note that both Flin and Amy's hair have remained almost entirely untouched since I made them 4+ years ago.

Anyway, at first, all I did was remove the shader's natural outlines (relying instead on Blender's "Freestyle" outline drawing), stretch the faces down a bit, and add hair. But eventually, I arrived at this:

There are 4 key things I did to get to this point:
1. I replaced the eye texture. This is...mandatory.
2. Lowered the Freestyle outline drawing to just 1 unit of thickness, making the outlines quite thin.
3. Deleted the black "eyeliner" along the bottom of the eyes for all characters. Ended up being very surprised at how much I preferred this change.
4. Recolored the eyebrows to be a given character's hair color instead of black.

Next: The Shaders.

I adjusted the shader used for the skin so that it just went down to two tones. However, sometimes I use three tones for certain objects. It's more random than I care to admit. Here are some examples of shaders using 2 tones and one shader using 3:

Like I said, most things use 2. Julie's shirt uses 3. The funny thing is, MB-Lab pretty much only gives you two shaders: the skin, and eyes. I ended up just copying the skin one and recoloring it for LITERALLY EVERYTHING YOU EVER SEE. Most things aren't textured, though some things are (like Eric's shirt).

I did make a few adjustments to the shader, but on the whole, you can see how the MB-Lab addon ended up powering all the core elements of this art style. The fact is, I really liked how this shader looked and how easy it was to use.

Other funny thing about the shader: It ignores light completely.

Well, unless I tell it not to. Technically, the shader is an emission shader, meaning that literally everything in the above scene is emitting its own light. This is a common way to achieve cartoon styles in 3D. But that doesn't mean you can't achieve shadows:

This is an old picture, made before I ever started LLRR. Ugh, that posing is awful, and Flin's still got that weird "eyeliner" outline...but the shadows are pretty cool, right? Loved how clean they looked.

The problem is, relying on lighting for shadows takes away a lot of my creative control over how a scene looks. Something I really like about my shaders is that I can always tell them to add a little shadow along the jawline to give it definition or something, regardless of where the character is or what direction they're facing.

Observe the following:

Amy's face and neck are shaded on the left side. Flin's neck is shaded on the right side. This might make sense if the light source was in between them...but then, what about that strip of Flin's shirt marked "what the fuck" that's also shaded on the left side?

In short, the shading here is not designed to imply a light source; it's to imply definition. I think it works pretty well (and it's used in anime, just not as ubiquitously as I use it)

But, they don't cast shadows against the pillows they're leaning against, or the bed frame. Little things like this definitely bring down the quality a step or two, and it's something that I'd...like to revisit, when I finish this project and re-assess my art style.

I think relying on lighting for shading leads to some very cool qualities. The tradeoff is that if I do that, I'll either need to fuck with light placement in unnatural ways sometimes, or otherwise have a way to override it when I want to, because many shots done in 3D do not make ANY sense in the actual 3D space they're in. Maybe I'll do a post on that sometime.

Still, shadows are cool, and I have only used them precisely one time in LLRR. It was very recent, and it definitely added to the image quality in a subtle way:

Stella's shadow is far more noticeable than the little shadows cast by Eric's drum kit. But even small shadows go a long way towards making something look like it's actually touching the ground instead of floating above it (or vice-versa). They can create a strong sense of depth.

But...I gotta cut corners. Shadows are an easy one to cut. I'll revisit them once I finish this project.

Outlines.

Somewhat already covered outlines, but not properly.

There are a number of interesting techniques employed in 3D in order to create outlines. In the case of Blender, there are two main tools you're likely to use: Shaders or Freestyle. MB-Lab relies on shaders, but as I mentioned earlier, I turned them off and turned on "Freestyle," which is Blender's word for this special outline-drawing tool it has.

Leaning on Freestyle led to a kind of "Flash animation" quality to my work that I liked, though admittedly, it's not super anime. I'm of the opinion that relying on shaders for outlines is probably the best way to achieve that more traditional hand-drawn style, but it's also possible that by tweaking the settings on Freestyle, you could achieve something effective there as well. I'll be experimenting a lot with both when the time comes. Also, different anime have different-looking outlines. Some are black, while some are a darker version of whatever color the object is. The latter would practically require a shader.

You know, I never do a very good job of showing you real behind-the-scenes shit. Screenshots of the actual Blender work 'n stuff. This post is getting very long, so I'll try to do this more next time, but let's end with something fun. Check out the 3D scene of the above rock concert:

Don't worry, I don't usually look at this freaky shit. Instead, I rely on material view, which looks very similar to the final render (minus outlines) and works in real time:

Isn't that crazy? I dunno, you might not find this very impressive, but I've always been fascinated with cel-shading, and how it can make 3D objects look very 2D. I think one of my big complaints with Koikatsu is that it strives to...not really do this so much.

Anyway, all the way to the right, on the bottom, are the shape keys for Flin. Well, half of them. Those shape keys represent a massive part of why I loved Manuel Bastioni's addon so much, as Flin automatically came with all those. Granted, I've since done plenty of adjustments to them myself over time, but they've saved me a ton of time (and are very fun to work with).

That said, I think another change I might make in the style revamp will be to rely on facial rigs instead of shape keys. Animation is a LOT easier to handle with a rig, as is saving (and more importantly, copying) specific poses or expressions.

One last thing, and that's how I can make custom outlines:

As you can see, outlines don't need to be drawn wherever they make sense. They can also be drawn where I tell them (those green lines). However, you can also see that this is restricted by the mesh's topology, or edge flow. I can't tell Freestyle to draw a face on Flin's arm unless the vertices play a game of "connect the dots" to make a face shape there.

...I don't know why I used Flin's shirt as an example, when the microphone he's holding is a way cooler example. Whatever. Here's how that looks up close:

Good topology makes for much easier detail lines!

Anyways, I think I've droned on long enough. Hopefully this shows you how this bizarre style came to be. I don't even know how many of you are interested in Blender, but I think this shit is so cool! I'm sorry if I sometimes do a poor job of presenting it. If you're at all interested in taking up 3D (or 2D) art, I totally recommend you get into it! This time next year, I hope and plan to be showing you much cooler shit.

Until next time, thanks for reading! This makeout sequence I'm never gonna use is your reward:

The End~!

Files

Comments

Azalies

This is a wonderful insight into what you do!

flinpaltwell

Thanks!! There's always so much I wanna talk about, it's hard to decide what to include. Glad you enjoyed it!