 Hi, I'm Julian Kasper and in this video, I will show you the workflow of creating stylized characters in Blender based on my experience in film productions at the Blender Animation Studio. If you find this topic interesting, visit the Blender Cloud, where I go into far more detail by giving a step-by-step tutorial course on the subject. This tutorial is in a way a brief version of the full course. So, be aware that I'm not going to dwell on any details. There's a link in the description, but let's just dive straight in. The full character creation workflow is often not very linear and can include different steps based on the type and style of the character. In this case, I'm giving a somewhat idealized, simplified workflow. As the very first thing, it's necessary to know what the character is supposed to be. In this case, it's not going to be overly cartoony or realistic, but sort of a mix of both. This way, the style is abstracted enough to not be uncanny, but also not too detached from reality to still feel grounded. Apart from that, you should have an idea of who your character is supposed to be. Nail down individual points like ethnicity, gender, age, time period, occupation, attitude, and potentially more. Really nail down who you want to create. Then, before this leads us into doing some first concepts, it's good to do some research. This step might make you re-examine your character, but it also should inspire and guide you on creating it. There need to be references on the character itself. Maybe you have an actor in mind as an inspiration, for example. You should also gather references for the style you want to use, which are likely going to be artworks by other people, which is OK. Then there are also references on the technical execution. If you want to go for a fur coat, then find out how other people tried and succeeded before you, which is going to save you a lot of trial and error. And of course, you should always include references from real life, like anatomy and fabric behavior and more. Know how real life works before you recreate and abstract it. Once you have all of it, display it onto a collage, ideally nicely organized in size and groups based on importance and categorization. My personal recommendation is using pure ref on a second monitor, phone display, or even printing it out on paper. Now it's time to work on the design of the character. The ideal scenario is to work from an existing concept before going into 3D. This might be your own drawing or someone else's, with their permission and credit, of course. I have to admit that in my case I didn't work from any drawing, but I can highly recommend you do so, since it will save you a lot of time throughout the process. The concept should embody a lot of the preparation from before and serve as a reference for the model. It's good to start off the 3D sculpting with some simple blocking, which means laying out the overall proportions with simple shapes. This can be done by creating and placing cubes, cylinders and spheres and sculpting them roughly into shape. From here on, you add more definition and detail over time, step by step. It's okay to keep the shapes very planar and rough for now, until the overall shapes are set in stone. Definitely keep multiple objects around. You don't need to sculpt everything from one object. Also make good use of symmetry options to save time. To not really worry about running out of geometry to sculpt on, there are helpful tools in sculpt mode. One is dynamic topology, which is enabled with CTRL D, which creates and removes geometry dynamically while you are sculpting. In the options, you can further define things like the resolution and the remeshing method. There's also the new voxel remesher that remaches the entire object almost immediately with the shortcut CTRL R, based on its volume alone, which is a much faster and reliable method with much better performance. It even remaches intersecting objects if you merge them. You also don't actually need to use that many brushes to get the first sculpt going. Simple draw, crease, smooth and grab brushes are usually enough. Once everything is roughly where it should be, it's good to finally merge the separate objects, either with the boolean modifier or by just joining them and remeshing everything with the voxel remesher. Afterwards, you can paint the objects with vertex colors to get a better impression of how the character will look like. This is very helpful, but also remember to turn the colors off from time to time, since they can obscure the actual shapes of the sculpt. You can even paint in things that are not there, like eyebrows and eyelashes, to keep things fast and flexible for now. For the hair, you can again use separate objects or extract geometry from a mask, or by using the skin modifier or even curves to create separate hair strands or ponytails. Afterwards, add vertex colors to those as well. If you don't have a strict concept art to follow, you can continue to polish the head sculpt or create design variations before settling on one of them. Just make sure that you increase the resolution and polish for the final head sculpt until you are happy with the results. Different matte caps or even your own custom ones can help you see surface imperfections more clearly. And don't worry about losing the vertex colors at this step, since they can always be repainted afterwards. They just serve as a pre-visualization right now anyway. And again, keep references on the side, especially when sculpting things like hair, which can be very tricky to pull off if you're not used to it. For the rest of the body, it's basically the same thing, starting with basic separate objects to block out the proportions and shapes as a good workflow. Also try using the mirror modifier with a custom mirror object that serves as a center axis. This way you can easily keep the entire body symmetrical, no matter the transforms of each object. Also don't forget to apply the scale with control A of your objects to avoid weird brush behavior. It's good to sculpt the character in a resting pose, like a T-pose, with very straightened and angular limbs, or in an A-pose, which is more relaxed with, for example, 45 degree arm angles. The T-pose is a better resting pose for later rigging and animation, while the A-pose will give you more accurate deformations in the animation, if the character is leaning more towards a realistic style. Keep sculpting and refining the proportions and shapes, until you're happy before joining everything together. For the clothing, it's again good to extract new objects from masks and refining them further by adding folds, wherever there should be compression and stretching in the fabric. Vertex colors will again help you with your impression of the character so far, and you get to experiment with different color palettes. If you decide to change the overall proportions of the character, it's easier to merge all the objects of the body together and adjust everything together. Just be careful about any modifiers you might have still active. Afterwards, you can split them into individual objects again, with a shortcut P and edit mode. You can again create variations of the outfit, but extracting them via masks is not the only option to create them. You can also, for example, box model a shirt, subdivide it a bit, and shrink wrap it back onto the sculpted body, and continue defining it from there. Keep adding more sculpted parts of the outfits, until everything is there. Some objects can remain a bit rough, until you get to the final polishing later on. Before settling on the design completely, it's a great practice to test the current one, and see if it can be tweaked in any way. This is also important for the later rigging and animation, to have a style guide already at hand. Before work can start on that, it's important to have a better mesh to work with, instead of the highly dense sculpt of the head right now. I can recommend to either roughly remodel the head manually, from a subdivided cube, that you snap back onto the sculpt with a shrink wrap modifier, or to use the new quadrimesher in the object data tab to the side. Just make sure to have enough overall density, and place the geometry a bit more accurately around the eyes and mouth, with some manual modeling. Also make sure the eyelets have three to four edge loops on them, to be able to close them too. Add an inner mouth and some teeth and a tongue, that closely represent the later polished result, but it's okay if they're a bit rough for now. To get most of the sculpted detail back, you can add a multi-resolution modifier, subdivided a couple of times, and add a shrink wrap modifier. Shrink wrap the original sculpt onto the new one, and apply it, and you will have the detailed version on the multi-res layers. Just make sure to not shrink wrap any geometry, that wasn't there in the original sculpt, like the inner mouth and inner eyes. With a vertex group and some white painting, you can exclude parts to be shrink wrapped, by adding it to the modifier. Add some shape keys, model and sculpt on each one, to get different facial movements, and mix and merge them however you like, to create various expressions. Be careful to only work on the base resolution, not any of the multi-res layers. For this task, you essentially only need the grab and smooth brushes, to move stuff around and relax areas a little bit. Start off with the basics, like closed eyes and an open mouth. You can create the same shape keys for multiple objects, and either animate them together, or even hook them up via drivers, so that they all slide together. From there, work your way up with simple smiles, angry shouts, and very stretched and compressed expression tests. This will inform you if the current proportions work well, or need to be tweaked, but by the end, you should have some appealing expressions to show off as well. Like always, references are important, even if you're just filming yourself to know how a face is really supposed to move. You might realize the eye shape doesn't really work with all the expressions, or that the eyebrows are too high by default, or the teeth are too low and small. Just tweak those areas until it looks right. Just make sure to always get some feedback from friends or coworkers before moving on to the next big step, or if you're in an actual production, sign it off like all the other steps by an art director or whoever your superior is. To make it all look even better, you can add some additional asymmetry to insert more character and appeal into each expression, and you're done with the general design. From here on, the task is to create the actual production-ready asset and finalize the style. The sculpted character so far is sadly impossible to rig and animate in any production, so the model needs to be cleaned up. This is where the retopology comes in. Retopology is about creating a version of the model that has the lowest necessary amount of geometry with the same shapes as the original one, especially when subdividing it. Just create a new object and either start from a simple base object or just poly model it all the way, starting from the face. Make sure to use a mirror and shrink wrap modifier and enable face snapping in edit mode to keep things close to the original sculpt. From here on, it's all about extruding, ripping, cutting, filling, and merging geometry to get the topology you need. There are a few things that make up a good retopology. The edge flow is very important, which is the direction of the loops that are going around the surface. These are all made from quads, which are faces with four sides, and they're led into different directions via poles, which are vertices with less or more than four connected edges. But never go above five connected edges, otherwise they might become really visible. It's very much a puzzle where you need to create an efficient and very functional version of your objects that supports any sort of movement, compression, or stretching that will happen with it once it gets animated. For example, just like for the expression tests, it's good to add some extra loops on the eyelids since these will have to close too and need that extra geometry. Joints need a minimum of three edge loops to be able to bend while keeping their shape. If you want to exaggerate a crease somewhere in the face for certain expressions, also have a minimum of three edge loops ready, like at the laugh line or between the brows. Using the subdivision surface modifier will smooth the entire result, so be aware of stretching. That's why you should mostly rely on evenly sized quads all over the place, especially on areas that will deform a lot or are very curved. Creases and hard surfaces are an exception usually since the former needs more density to stay sharp and the latter doesn't need any extra loops to deform well. It's generally a good practice to start from the face and to re-tapologize each limb separately from there and stitch them together afterwards. Make the loops generally follow the shapes and muscles of the body to support the direction they will deform too. Make sure that most loops lead back into each other. You don't want some loops to spiral all over your objects. At some point you will have re-tapologized everything that was sculpted, which is when you will have to model things that weren't already there. But first add a multires modifier, subdivide it a couple of times, and apply the shrink wrap over it. Then click on apply base and get rid of the multires modifier too. Now your re-tapology will have the same volumes as the original sculpt when adding a subdivision surface modifier. From here on out you can model in some inner eyelids, inner mouth, polish the already existing teeth, gums and tongue, and model in some proper eyes. When you get to the clothing you can duplicate parts of the body to have a good base to start from. This is also the time to add some more definition to roughly sculpted objects or adding seams and nice pockets to pants and other details. Once it's done, make sure to have the asset in the correct world scale like roughly 1.7 meters in my case, and it's time to go to the next step. Just like the topology is the first step to some good rigging and deformations, the UV maps are the first step to good textures and materials. Just imagine you take some scissors and make multiple cuts on your cleanly modeled character, unfold it onto a flat surface and place an image on top. This flat unfolded mesh is a UV map, and that's how texturing a digital model is done. To achieve this, you can select edges in edit mode and tag them as seams. These edges will be the cuts on the UVs. Once you have them placed, you can unwrap your mesh and the UV map will be visible in the UV editor. By enabling the stretching overlay, it will be more visible if the UVs are less ideal in certain areas. Also already using the UV map to add a color grid texture onto the model will also show the distribution of the UV space better. The goal is very much to place just enough seams to evenly lay out the mesh with minimal amounts of stretching. But just like it's impossible to create a perfectly flat map of the earth without some areas being a bit bigger than others, some imperfections will be hard to avoid. Also try to avoid placing too many seams that are in the face, for example, since under certain circumstances, it can be easy to spot the seams. Hide the seams as best as you can. Once you have your UVs nicely unwrapped, it's time to adjust and organize them better. They should be laid out in a way where they take up as much space as possible from the one by one UV space. Having them placed in a symmetrical way can also be very helpful later on. Make sure the UVs are generally having the same scale unless you want some UV islands to get more texture resolution than others. It's a bit like packing your luggage and making sure everything fits in perfectly without getting all messed up. For areas like the eyes and lips, it can also be helpful to select that geometry and projecting it from the view, smoothing them out a bit and stitching them back to the rest. Whatever makes the UVs more evenly distributed. If you start adjusting UVs like this in detail, remember that you can pin UV points with a shortcut P so that they don't move when unwrapping those UVs again. Also make sure that none of the UV borders are too close together and that no UVs should ever overlap unless you know what you're doing. You can also use UV sculpting tools for wider smoothing and grabbing of areas but be careful not to cause too much stretching with it. Scale some areas with proportional editing enabled to get some more evenly sized UVs at the cost of some more stretching. It's always a bit of a balancing act between the two. For the clothing, it might also be important to have the UV maps laid out in neat rectangular patches. This way it's very easy to just slap on some fabric patterns without having them distorted or go into weird directions. Just pin the borders of the islands and enable live unwrapping. Then adjust the borders to be as rectangular as possible and the rest in between will automatically conform to it. And keep comparing it to the grid texture in the 3D view. This will be a good reference on how distorted the textures would be and how much texture resolution each area gets. Go through all the different objects with these principles and you will have some clean functional UVs to texture in the next step. To get the character some colors we need to add textures. Add a shader editor in the interface to edit the node based materials and an image editor to see and paint on the textures in a 2D view. Also as a pro tip, enable the node wrangler add-on in the preferences. I can recommend to just use the default principled BSDF shader in the nodes since it has basically everything one would need during the texturing. Then start out with the basic colors. Do this by adding and painting on image textures or vertex colors and plugging the node outputs into the base color of the shader. Most nodes can be freely mixed with the mix RGB node and each have their own advantages. Use the material preview mode in the viewport which displays everything you're working on. Control shift click on any node to see that output directly. Start out with most primary colors and then mix in secondary colors. You can also then start mixing an ambient occlusion or cavity inputs, patterns, noise and specific painted in details. Just blend the nodes together either by using screen to brighten, multiply to darken or mix to blend between two inputs. Also make sure to paint on the alpha channel which is the transparency on an image texture to use as a mask by plugging it into the factor afterwards. The well-aligned rectangular UVs for the clothing will really become useful now since images like clothing patterns can just be slapped on top and they will immediately align. Extra details like stitches can always be painted separately on top afterwards. Definitely keep all of these influences in the color as separate nodes because once you are done it's time for the shading. The other most used inputs in the shaders apart from the base color are roughness to define the glossy shine or the lack thereof, metalness if the surface is metallic and normal to add fake high-frequency details that we didn't model in already. If you want parts to be transparent like glass or ice you can use transmission and to just make them use fade away or just completely disappear use alpha. You can also use specular to tone down the glossy highlights but if you want to have it really realistically accurate increase the roughness instead. Plug your individual nodes from the color also into color ramps and hue saturation value nodes to change and then remix them for the other inputs of the shader. Try to keep everything organized well enough by clumping nodes together, framing them or even grouping them. Change the lighting conditions and test how the materials react. Adding additional lights and rotating them around manually can also help. In case of the skin it's also good to add subsurface scattering or SSS in short to give it that waxy look like the light is shining through. To give a nice fake hair look on modeled hair I can recommend the anisotropy shading which adds a directional distortion to the glossy highlights. Turning the anisotropy to 1 by itself is not really going to look convincing or even good since it just adds a spherical pattern based on a world axis. You need to create a second UV map add it as a tangent node and plug it into the tangent input of the shader. The orientation of the UVs will then influence the direction of which the highlights will be distorted towards. So all hair strands need to be laid out as flat as possible and aligned vertically. Now the shiny streaks are going horizontally across the surface instead of the fixed spherical direction of before. To fake individual hair strands of hair you can paint a black and white texture with small value differences and plug it into the anisotropy rotation. Adjust that same directional map with a color ramp and plug it into the roughness. Add color and you're done. Keep testing and tweaking your materials in both EV and cycles and optimize it towards the render engine you want to use later on for the final rendering. Once you're happy you might have a hugely complex nonsensical node tree. This can be then simplified by only looking at the necessary inputs that lead into the shader and baking everything down into a set of image textures. Just make sure you have cycled set as the render engine. You control shift clicked on the part of the node tree you want to bake and select an image texture node where you want to bake it too and it should work. To speed up the baking process set the samples to something pretty low go absolutely overboard on the tile size set the bake type to emit with a generous margin and click bake. Ideally once you're done your character should look good whether you use EV, cycles or just material colors in the viewport. Just like we wrapped up the original design with some expression tests this is the time to put everything that we made so far to the test with a full pose. This is to see if there are any last adjustments that should be done like tweaking the proportions of the body, fixing stretching, etc. Of course it's also great to not present the character in the current resting pose. The rigify adding in the preferences can really help to get some rough poses going. Add a basic human metarig and adjust the placement, size and orientation of the bones in edit mode to fit the character. Don't forget to enable xMirror to save yourself some time. Once you're done click on generate rig and there it is. Select your character then the rig and with CTRL P parent the objects to the armature with automatic weights. With this done you can do some exploration in the overall pose, experiment and see what you want to go for. Since the rig is pretty procedural there will be issues. If some deformations are outright broken you can manually go into weight paint mode and paint out some issues. Make as many poses as you want until you find your favorite in the end. Don't forget to look at references and even make references from yourself if necessary. I recommend to have a second viewport to the side and set the shading to be completely black maybe with some dark gray outlines. This way you can see if the silhouette is readable enough. The pose can then be tweaked towards a single camera angle or any perspective you want to render. Make sure the pose is dynamic enough with differences in angles enough asymmetry and whatever else you have in mind as well. In this example I wanted something comfortable and confident not a super happy victory jump or something like that but something way more contained and down to earth. Have an idea in mind of what kind of pose you want. Then apply the armature modifiers as shape keys and then you do essentially the same thing as with the expression tests. Add more shape keys for various non-destructive fixes in the deformations and keep comparing it to the original resting pose because the proportions shouldn't change of course. This is very easy when animating all shape keys to zero on one frame and enabling all shape keys on the next one. Creating variations this way is also a good opportunity. If you notice that the arms actually look a bit weird in length or the hands are too big now is the time to make last changes before committing to the end result. Also sculpt an expression that you like and polish it until you're happy. From here on out you can place some lights a nice environment and render a final image. Render multiple angles or even a turntable to show as much of the character as possible. A so-called clay render is also good to show off the model itself where it's rendered colorless in diffuse gray with only the bump and normal information at most from the materials. Add wireframes on top as well to show off the clean topology you created. Render those also as turntables and you have a complete character for production that you can show off. From here on there are of course more steps like doing an actual rig for the animation, animation testing and the rest of the production. But in terms of finding and developing the design and style of the character these are the steps you need at the very least to create a complete stylized character workflow. If you're interested in more educational blender content a huge asset library of all the Blender Open movies or you want to support these projects head on over to the Blender Cloud and say hi in the comments. You will also find a fully rigged version of this example character called Rain and various files from the creation process. I hope to see you there and thank you for watching.