How Nana prepares their model for Live2D (Part 1)
So, with many people aspiring for having their own Live2D model nowadays, whether it is for streaming or chatting, people seek out for artists and riggers in the world wide web to make their dream become a reality.
The method I use probably doesn't have a name, since it's something I came up with for myself. It basically boils down to drawing the parts separately even before the "cutting apart" phase, which many artists do as an approach, since it's the closest to how usually illustrations work. For that purpose let's try giving this method a name, since I have a rather dry humor, lets call it, the "Backwards Method of Nana".
The Backwards Method of Nana
This method does as implied, many things differently as artists usually would do when drawing an illustration. And that's how it was meant to be, since we are not talking about a normal illustration here, instead it's going to be a Live2D model, which can be compared to 2D ragdoll animations, a term which should be familiar for people from the animation industry.
Because this is essentially Live2D models are when it comes down to it, my approach doesn't require any sketches, but the lineart is the sketch itself and is broken down to many different parts (such as 'top eyelash', 'bottom eyelash', eyewhite', 'iris' etc.).
You may require a rough sketch still, if you are not accustomed to work in this fashion, but even so, once you are starting to work on your lineart, I would suggest you to start working "backwards" from this point onwards.
What are the advantages of this method?
So, one might ask: "Nana, this method takes so much time in this early phase, plus I can't really give my artist any feedback until they have already spent so much time drawing on separate layers, why would anyone do that?" There are a couple of reasons why this approach is smart and how both parties can benefit from that.
Fitting and making small changes
Despite what you may think. The fact that this method enables parties to be able to have meetings even before the art is done is a big reason why I think it's the best approach to set up a Live2D model. Sketches are usually not defined enough and once layers are colored, changes can be proven to be rather tricky (especially if the coloring method is painting).
With this method, you can show a defined work in progress and layers can be easily manipulated and edited. Don't like the positioning of the ears or are they too big or too small, is the arm too slim, the torso too large - you can make those changes in a jiffy and show the results in seconds. It's also not a hassle to actually redraw the lines and for most established artists, there's no need in requiring a sketch at this phase.
Make Use of Different Colors
During this phase, one big bonus one has is to be able to use different colors for the lineart. In my personal case, I prefer to use 2 colors, black and red. Red is used for corrections before they are fix and lines which will later be hidden (before they are fix).
It can come in handy because the model at this stage doesn't have any colored layers, which can make it very confusing to look at it. Another method you can use in this process is simply making the lines which will be hidden see-through (reduce transparency) before you start add colored layers / color your layers (whichever applies for your personal style and approach).
Making Decisions Before Rigging
Especially applying for people who use the free version of Live2D, saving on resources is a very important factor when making a model. However, another point which has to be taken into consideration is what features this model is suppose to have, examples as such could be:
Are there going to be any bounce physics in the chest area (aka "booba physics")
Are there going to be animated joints
How is the model going to move and tilt when moving left and right
Questions as such should be asked beforehand and can possibly be answered with this method, since they determine on how the lineart (and that for layers) has to be separated.
The result could be something like this. This example features the lineart of a model which has separated arms and legs, and a head with different layers of hair and a face with eyes drawn on separate layers (top eyelid, bottom eyelid, iris). Once everything looks okay, we can now start coloring each part of the body and arrange them accordingly. After that, we can finally add some clothes to the model and think about features.
This method keeps track of how many parts we are going to have, the limit of the free version of Live2D is 100 parts. If you are going to use the free version of Live2D, or if your rigger is using a free version, this is a good method to keep the project in check.
Once the base colors have been added to the current model, you can test the joint parts which will be used for the final model. At this stage, it is very easy to test those places.
Another thing which can be done once the base colors have been added is to adjust the details of the base model one last time. In this addition, the hair has been modified a little bit by making the hair a bit wider, as well as adding 2 more spikes on the side. The eye lashes have also been slightly corrected, they are wider now and the eyes have been moved slightly more to the center.
Now it is up to the customer whether they want a 'naked' base model. Generally I always suggest doing that, sure, this means you will have extra layers on the model, however, it also means that you can later on exchange clothe layers in the rigged model by replacing the existing ones. This is where the whole dress up doll concept comes in and in my experience it is a really nifty thing to have.
Once the Colors and Outfit Have Been Defined
Once everything have been put together, the result should look somewhat like a normal illustration. In this live example, you may have noticed that once again, many small tweaks have been made on the model (head size, hair color, eyes, face).
As you add new elements to the model, you will find ways to optimize its look and feel. since we've been working on separate layer groups for each element of this model, making such changes is rather an easy ordeal.
This might be a good moment to point out some fundamentals regarding character designs. There's a saying that generally a good character design follows several rules which I will list here:
Avoid using too many different colors: generally try to work with around 3 difference main colors and distribute them evenly (example here, black, white, red). This will make sure that your design doesn't look messy and instead looks like a well thought out design.
Start from a simple design and then built up from it: It is a good idea (unless you have a trained eye), to start from a very simple concept (aka. simple hair style, clothing, etc.). Once you have the general idea down, you can think of how to improve that look by making small changes, one at a time. This will make sure that you'll end up with an interesting design, but won't end up with one which looks like an overloaded mishmash which doesn't work as well as intended. You can also add small details to your design, which will make it visually more interesting (examples: a mole on your character, markings, small labels, etc.).
Have the courage to remove things which don't work: During a design process, you may have many ideas which you think would look awesome on your product, however, most of the time you'll reach a point where you will question your previous decisions. Have the courage to remove ideas, redo changes or adapt ideas as you go.
Compare yourself with the best but don't get frustrated if you are not as good: It is a good idea to constantly compare yourself with people who are better at something than you are, this is a great way to figure out how to improve your own skills. However, don't just see the skills of someone else and get frustrated that you cannot do, what they can. Chances are, that they have worked hard to be where they are now, and the growth of each person is different in learning new skills. Take me for instance, the showcase here is the result of over 25 years of drawing. It is by far not as good as some professionals, but chances are, that it is better than many others' work who may not have had over 25 years of dedicated training. Never forget about that fact and don't lose courage.
Franchise Matching Models
Another advantage with drawing a model on separate layers beforehand is obviously the fact that swapping out parts can be done in a fairly short time. In this example, I've created a second model based on the first model and the layers were created in roughly 3-4 hours in total. Since the base model is the same, one could even try to base the second one on the first rigged model.
For franchises or groups, this method can proven to be beneficial. You can use the same base for the characters and as a result, the art style stays consistent across all models.