NPR Learning Path

Written by Laurent Marechal - 03 september 2015

Classified in : Blender, Links - Tags : Blender

Site d'origine : http://blendernpr.org/npr-learning-path/#learn

Warning: Very long article without photo, please bookmark.

Table of contents:

Introduction:

The path to greatness has many junctions. Not many potential artists want to take them. The master artists that create jaw dropping artworks have gone through most of them, even returning to the same junction just to take another path. The journey will take many steps, mastering one section then moving on to another. Without going through the painstaking processes, many will not even see themselves maturing into a master. Despite this, their paths are not often discussed. What are the paths that any great non-photorealistic (NPR) artist has to venture to become master of their own?

This article will talk about learning methods, tips and hints to shortcut these processes. At BNPR, we love to teach you “how to fish” compared to “giving you the fish.” This is the kind of article that you will look back after many years to evaluate the journey you have taken.

To go the NPR journey is to go the same journey as the artists who learn photorealistism (PR) but with twists of perspective and thinking paradigm. The same terms in PR will be differently applied in NPR, often times simpler and with less math. For example, ambient occlusion (AO) is multiplied with grey color in PR; in NPR, AO can be done with any algorithm, AO can have hue at any saturation as well. Another example that is obvious for NPR artists is how we think about Sub-Surface-Scattering (SSS). In PR, SSS is light bounces under a surface; in NPR, SSS is color relationship between lit and shadow. More of these differences later.

Some beginner artists want to limit their skills, which is alright. Some only love to 3D model a character but are very bad at shading the character. Some are really great at NPR shading and compositing but hardly can 3D model anything. In the long run, any master must acknowledge that one must go through everything, or at least test everything to have an inclusive perspective of what it takes to be a master artist. Then there are those who argue a lot about art. Well, you know where they are and who they are. Even if you don’t make artworks for others to see, you should make it for yourself to test your abilities. Thinking is easy, I can think of the best looking anime character and turn table him in various poses, making it happen is a whole other world, hence this article.

Check points:

Becoming a master is a long path, please take it step by step.
PR terms can mean different things in NPR.
It is okay to limit yourself as a beginner.
To become a master artists, you must break walls of doubt & try something new. In short, unlimited.

Learning best practices

If you want to improve your skills tremendously in a very short amount of time, you have to learn smart. By smart, it does not meant it is without hard-work.

Always do sequential learning. Multi-tasking will burn out your focus. Learn one topic at a time and put a time limit. At the end of the time limit, you must test your new skill by making few artworks. When you are sure you have mastered the task then you can claim mastery. Else, add a little more time to learn the skill.

Learn a task in 7 days. On the 8th day, test your new skill, if you find you are still lacking practice, learn for another 2 days.

It is very easy to get discouraged when the task is hard to master & you have extended your learning period few times, but there is help. Start by asking. Please be thorough and polite in the process. In the Blender community, people really love to teach you and give you different perspective on the same issue. Stay on the positive side of the learning path. You cannot come out like a crying baby wanting milk. In the process, you will find yourself getting mentoring from many artists. Some mentors are for keep, your relationship with them will shorten your learning curve. Be appreciative with them mentoring you, because they are not doing it for profit.

Another way to learn very fast is by forming a study group. The BNPR team have been doing this for the past 4 years. We learned a lot from each other, plus we know the strength of each team member. In time of need, we leverage each other’s strength. This is what we call synergistic learning.

Should one get paid education? Yes and no. Yes because you’ll get those rare courses which will fulfill your knowledge gap faster than learning from pre-planned learning path. It is wiser to say that, do not stick to only 1 group of mentors. Each group will have their own blind spots, When you follow their learning path to the dot, you indirectly learned their disability to discover better workflow or work faster. That is called “learned disability.“

No you don’t have to get paid education to be a master. In the past few years I have been learning directly from industry leaders. How to model faster, how to see styles and how to work as a team. Those are better learning opportunities than the sterilize education that normal education path will give you (you can see those with young institutionalized educated artists). Do not fall into the “monkey house syndrome,” where you cannot smell the feces anymore (reality: a lot of famous studios in this category).

Check points:

Learn 1 thing at a time.
Put a time limit to master 1 task.
Work very hard with in the time limit.
Ask questions thoroughly and politely.
Seek mentors.
Form study group.
Do not get trapped in learned disability.

Embark on an epic journey (The extreme basics, UI)

This topic has been discussed so many times, yet, there are few more deeper subtopics on it that almost never mentioned. Learning Blender’s User Interface (UI) is great because you can see how things are organized in windows and the logic of their arrangement. You also start to see what each window does and the names of their elements like tabs, panels, header and sidebar.

Some part of the UI is useless for speed. The tools shelf in Object mode for example only slows you down. The hidden menus in edit mode (Special, Vert, Edge, Face menu) need better shortcuts (I use 4, 5 & 6 to call them up). Selection type also needs faster shortcut keys (I use 1, 2 & 3 to switch between them quickly). What you can learn from these 2 examples is always try to find places in the UI that slows you down then find fixs for them.

Screen layouts is also a good way to help you work quickly because you can switch to them quickly with Ctrl L/R arrow. That does not mean it ends there. Video Editing screen layout for example has never been a good enough layout for me. I changed it to work like in After Effects or Sony Vegas.

Not all Blender themes are created equal, especially those darker themes can really ruin your eye sight. Some have contrast so low that you need to squint to be able to read the labels. Some are too colorful without being meaningful. A good Blender theme will help your eyes from getting tired too easily. Blender theme Energy is designed with that in mind.

Blue hue will make you more alert but it will burn you down in the long run. It will also ruin your circadian rhythm. Avoid blue colored themes.

After you are comfortable with the theme, and what is where in the UI, you can traverse the UI in very unique workflow. If you are just beginning, most stuff will start to make sense after a few weeks. Just hang in there until that time.

Check points:

Shortcuts with one button to press is easier to press than those with two or more buttons. Always think speed.
Create and optimize screen layout.
Pick a good Blender theme so that you can work longer without tiring your eyes.
Train to use the UI for your specific workflow.

Get organized (Project and file management)

There is a lot to be said about this topic, but largely they are within 2 camps.

First is the project structure side of things. This includes, but is not limited to:

Project folder structure.
File naming conventions.
File versioning.

Second is the human side of the production. They are:

Production rules.
Scheduling and task assignment.
The feedback loop.

Let’s demystify them one at a time.

Project Folder Structure:

There are 2 major ways to structure this: workflow and file types. The keyword to select which method for folder structure is more humane. By that I mean which folder structure is easier to work with.

If you are using file types, in general you will know where to save files; in the long run they are harder to comprehend as all same files type being inside a folder. The file type method is best used for archiving.

Asset folder: 2D folder, 3D folder, etc
Document folder: Scripts folder, Reference folder
Production files folder: Blendfile folders, Scene folders
Output folder: Previs folders, Rendered raw folders, Final Output folders

For myself, I lean more toward workflow folder structures. For example, in an animation project there are few stages of production:

Pre-production.
Production.
Post production.

Inside pre-production folder you will have script development, references folder. In production folder you will have individual reusable models folders with reference files; character folders; sets folders; props folder; animation folders; code folders; and ready to be rendered production folders. In post production folder you’ll have sound folders, raw rendered folders, and various quality output folders. This folder structure is best done with asset linking. This way, more people can work on more areas of the production at 1 time.

File naming conventions and versioning:

Before diving deep into this topic, there is 1 prerequisite. Always name your objects, groups, material inside your Blendfile and properly name your Blendfile. With team of more than 2 artists, it is often very easy to not know which file has what and do what. The basic file naming convention is like this:

ProjectShortName_AssetType-ShortDescription_Version.FileExtension

For example, a ruined house model for Project Falcon will be named as:

PF_HouseRuin-2Stories_V12.blend

Note: Use Underscore (_) on different topics and Hypens (-) on subtopics.

When someone is working in a new revision, a new copy must be created. It will be named as

PF_HouseRuin-2Stories_V12-2.blend

Note: Always put version numbering last on the file name. Easier to press + to up the revision when needed.

The extra -number suffix means someone is using/working on the file. Everyone else should refrain from editing the original revision.

Note: Always make a new revision copy before start editing to claim a task.

There are many version control systems for online collaborative production. Often times, they are extremely technical to setup or cost too much. For small teams, it is best to use simple versioning as stated above. You can use either Dropbox, OneDrive or Google Drive. For bigger teams, self hosted production repository like OwnCloud will be useful. Security of your production files must be considered as main priority when you host files online. You don’t want hackers releasing previews, assets files, or the full movie ahead of time.

How many versions of a file should you keep? Keep as many as your storage allows, but make sure there are not more than 10 versions (backup then delete the older versions in repository), else it will be hard to navigate.

Production rules:

Every artist has his own ideas on a production. Without rules those ideas will cause chaos. With strict rules, files will be very organized at the expend of annoyed artists. An annoyed artist will harm a production. More lenient rules will be more friendly to artists but will take more time to organize files. Hence a balance of both must be reached. The rules must be made easy to follow and should be culture in any production.

Here are few rules:

Properly name everything & put files in correct folders.
Finish the base work quota. If something comes up, please consult team.
Claim tasks & make temporary files before working on them. Report done tasks.
If you discover better/bad workflow, please notify the team.
When facing difficulty, do not wait for help, please seek help instantly.

Scheduling and task assignment:

Scheduling tasks can be done as simple as few notes on a piece of paper. For online collaboration, tools like Google Spreadsheet will be extremely useful. In the spreadsheet, you may need to add these information:

1. Task name.
2. Who is doing it.
3. Progress of the task.
4. Short remarks, comments & critic on the task

If your production is done using the Agile development/production method (short sprints and results by feature), a Gantt chart may not be needed. It is very easy to fall into unnecessary over organization. The most minimal work scheduling can be done like this:

Week 1: Concepts and understanding scripts.
Week 2: Environment & Characters exploration.
Week 3: Building Environment & Characters.
Week 4: Environment texturing, Character rigging.
Week 5: Week 4 part 2.
Week 6: Props, special effects and animations…

When scheduling, always add few weeks extra to allow for unforeseen production hiccups.

The feedback loop:

Do at least 1 weekly meeting. Each meeting should be set to a time limit, because meeting can be time wasting and unproductive after a certain time length. The objective of a meeting should be to get a grand outlook of the whole production, what have been done, what workflow works and don’t work. It is not only for critics, but also for suggesting how to make production faster and align making stuff to the result envisioned by the director. Alternative to weekly meeting, take about 10 to 15 minutes every day to check others’ progress in the production. This is also the best time to do short critics on artworks.

These meetings can be done via text (IRC style), voice or video (Google hangout, Skype call). Always discuss on what to do next. It is advised to dwell little time on wishful idea thinking (except on pre-production stage).

Check Points:

Project management and file management can be done in smaller scale.
Use a more humane project folder structure.
File name & versioning should go hand in hand.
Production rules should be culture in any production.
If you are working alone, start to organize your work folder as if it is a grand production.

Populate your world (Modeling)

3D modeling is one of the most fulfilling thing to master. You can follow the traditional learning method or learn it by shortcuts, resulting in less effort, shorter learning time and better models.

To master modeling fast is to know where modeling tools are hidden in Blender. They are in 4 big menus: Special menu, Vertex menu, Edge menu and Face menu. Those tools with shortcuts are more often used. Each tool has its own sub-modifier keys, be sure to look for them on the 3D view window header after you press the command. Testing each tool will help you understand them faster, because doing is the best learning method.

Note: There are stray modeling tools which can only be activated via shortcut keys or search result. Be sure to find them.

A good modeler always models the fastest and keeps good topology overall. Modeling is often done in 3 passes.

1. Basic shapes.
2. Primary details.
3. Secondary details.

By using reference images, you can rough out the model using simple shapes. The simplest and most powerful method is to Edge Model. Edge modeling method is done by only roughing out the shape and silhouette of a model by using edges. This captures strong silhouette and details in the least amount of time. Even a very complex monster (like those from games) can be rough out within 10 minutes.

When that is done, we can focus on mesh topology and edge flow. To learn edge flow quick is to think of how the mesh will change shape when animated. Be sure to find useful info graphics online to help you with good topology and how to do 3 to 1, 4 to 2 and 1 to 2 edges flow. Also learn anatomy from good anatomy books, they are great to help you produce better silhouette for your characters.

Do not limit yourself to modelling only what you like. You may love to model human characters, but please try to model hard surfaces as well. A character is pretty useless without an environment to pose in. Other than basic modeling, there are many shortcuts to more advance modeling methods. They involved knowing more of Blender’s tools. For example, to model a believable pillow, you can use physics or you can use sculpting method. Using modifiers as your modeling tools is a great way to model faster as well.

For NPR, shapes are often augmented to emphasize or de-emphasize a feature. Shading will also change your topology. These are 2 extra things we as NPR artists must pay attention to.

Beyond modeling, every artists should learn compositional layout. Learning layout helps in building strong composition, it guides eyes to focus on the essentials in the shot.

Check Points:

Learn the basic modeling tools by testing them.
Model in 3 passes.
Edge modeling is the best method to capture strong silhouette fast.
Learn topology, edge flow and anatomy.
Do not limit yourself to only modelling what you like.
Use tools outside of edit mode to model.
Learn to do scene layout.
Do a lot of modeling, track your speed.

Know the elements of the world (Mesh data)

Mesh data is part of vertex shader. These data are needed for mesh viewing, and they are parts and pieces to build a shader (both viewport and rendered).

Mesh data is not limited to vertices, edges and faces. It is inclusive to:

Normals
Texture space
Vertex groups
Shape Keys
UV Maps
UV seams
Vertex Colors
Sharp edges
Crease edges
Bevel weight
Freestyle face and edge mark
Weight Paint

We need to understand mesh data because it is used in:

Mesh modifiers (Vertex Groups, Sharp Edges, Crease Edges, Bevel Edges, Weight Paint)
Shading* (Normals, Vertex Color, Weight Paint)
Texturing (Texture Space, UV Map, UV Seams, Weight Paint)
Contour rendering (Freestyle and Edge; Freestyle Face Mark, Freestyle Edge Marks)
Mesh selection (Vertex Groups)
Rigging (Weight Paint, Shape Keys)

For NPR, please pay extra attention to shading part*. For example, normals can be further edited when you have normal editing tools. You can use vertex colors to do all sort of shading tricks. Weight Paint can help you vary thickness on your inverted hull for outline drawing.

Check points:

Mesh data are building blocks to almost everything in Blender.
Understanding mesh data will broaden your perspective on how things work in 3D (actually any 3D DCC).

Shade your world (Shaders & Materials)

In OpenGL, the shading stage is called fragment shader. This is where all mesh data is combined to produce different shading behavior. This is where cults of PR and NPR are formed. For NPR, we pay extra attention on vertex color, vertex normals, light vector, and few other mesh data derived parameters.

I don’t have to speak much about this because there is a very well known and well done presentation about it.

Video Link: Guity Gear Xrd Art Style

What I’m going to add is the importance of art style. Seeing style is a skill you must train to master. From experience, the more you train the better you’ll be able to see them. Having a good eye isn’t the only step to mastery, one must be able to produce what one sees. Knowing is believing; seeing is acknowledging of its existence; doing is making it exist. You must reach the later in the quickest possible time. Hence, tools like material nodes, the classic BI material UI and how they work, must be known like you know the back of your hands.

At the early stage of your education, please start by identifying the settings and nodes. Try to understand how each works and the general use cases. After that try to make diffuse shaders and specular reflection separately. After you are familiar with those, try to add vertex color and normal editing. Much of these are mentioned in many blog posts and downloads on this website.

Once you have a good handle of how things work, try to think about making shading components in layers. With layering shading as the base paradigm, you start to see many more possible shading ideas. The main goal is to go foolishly hungry and dive in head first. Just try a lot of shading styles and keep notes.

Shading will involve a lot of Pixel Math. From the structure of your shaders to combining render passes, pixel math is everywhere. The most basic pixel math is blend modes. Beyond that you have to create your own node setup by really understanding the tiny details of how math is producing the result. You can feel the color, but it is better to master the underlying math that generate the blending effect.

Knowing is better than just feeling the result.

Lighting in NPR shading is just a tiny part of the whole. In NPR we can always go with 1 sun lamp setup. Most time we also have lamp rigged together with a character. If you are going with a more 3D NPR, color palette is more important than lighting. Which will be the next topic.

Check points:

Have a good eye to see styles.
Know your tools to make the styles you see.
Do a lot of shading studies and take notes.
Learn Pixel Math.
Use minimal lighting.

Hue hue hue they said (Mastering Color relationship)

Bad color relationship is everywhere, but no one should get a divorce.

Joke aside, the sore thumb in the room is often color relationship. It can make rendered stills or animation looked like it was done exclusively in Blender or BI or Cycles or 3D. The common problem is as follows. The relationship between THIS COLOR and THAT COLOR is only a change in value. This problem stems from the over emphasis of hue as the only important parameter in color, which blinds us to neglect other color parameters. Any n00b level color tutorial will teach you color relationship like complementary colors, split complementary colors, warm and cool colors, and analogous colors. Again, these are only hues.

What you need to understand, in any color space, color is 3 dimensional. Color is consist of HUE (color of color), SATURATION (intensity of color) and VALUE (brightness level of color) parameters. Proper color understanding is the ability to change more than 1 color parameters to depict “bright to dark” color relationship.

Examples of proper bright and dark color pairs (with all 3 color parameters changed):

BRIGHT and DARK
BRIGHT and DARK
BRIGHT and DARK
BRIGHT and DARK

Note: Some of these colors cannot be printed, as they go beyond the CMYK subtractive color space.

PR can use the whole spectrum of colors; what makes NPR looks NPR is the ability to use a very limited amount of colors. A properly planned color palette will make a render looks harmony. By harmony I mean the colors looked like they were a result from a single light source and they relate well to each other.

Check Points:

Value only as color relationship is weak color relationship.
Color is 3 dimensional.
The start of proper color relationship understanding is the ability to change all 3 color parameters at 1 time.
A limited color palette is key to produce NPR result.

Turning fabrics into fashion (Texturing)

Color can be applied in few ways on a mesh. For non-UV, there are two ways to add colors. The color can be from the material settings which are the result of shader and lighting algorithm. Another way is via vertex color painting. For vertex color painting, the amount of vertices will determine the detail density that you can have on the mesh (sort of like pixel amount for raster images; in this case, it is vertices). Vertex color painting has a limitation, it does not have alpha transparency.

To get alpha transparency, we need raster textures with alpha. The limitation of this is you need to unwrap the mesh to flat plane(s) to be able to map the 2D image(s). For NPR production, we do not need extreme high definition textures. Therefore, a simple image format like PNG with alpha channel is sufficient.

Creating good UV layout is a set of skill that all NPR artists must master. Pay attention to UV stretching and how compact each island is placed next to another. All of the UV settings are on the header of Image Editing window. I encourage you to play with each setting to get a more intimate feel of how advance UV editing in Blender is.

Texture and vertex painting in Blender both have almost similar tools, with texture painting having more features than vertex painting. Images painted with texture painting will appear on the mesh and on the UV image editor. Painted vertex color will be saved on the vertex color layer.

Both texture and vertex painting use brushes as their main tools. Settings under each brush are very well documented in the Blender wiki. With a little bit of tinkering and understanding of those settings, one will be able to create very creative brushes.

They are more creative ways to use the textures that you have painted. In general, your texture can be used as the surface color, yet that is only the start of how painted texture is used. Painted texture, either raster images or vertex color, can be used for:

Masking.
Blending factor/threshold.
Shader modifier and more.

When combining UVs from few meshes into one big texture, the resulting texture is called a texture atlas. With that you will have fewer raster files. Loading one file into memory will be faster compared to loading multiple smaller files scattered in your HDD.

These are the common problems with texturing:

Too many layers of texture on the surface, resulting in slow and heavy file loading.
Textures used are either too high or too low in resolution.
Poor UV layout causing uneven density, resulting in extreme texture stretching.

Do not forget that you can also use external programs to paint your texture. The hassle with that will be exporting the UV layout, and later fixing the seams of the painted texture inside Blender.

Check Points:

Vertex color level of detail is determined by the amount of vertices.
Texture painting requires UV unwrapping.
Learn how to create brushes.
Painted textures can be used in creative ways.
Create texture atlas for faster file loading.
Pay attention to common problems with texturing.
DO NOT forget to save painted texture.

Giving the dog its bones (Rigging) & Breathe life (Animating)

Rigging and animating are very tightly related. Hence, they are under the same topic.

Rigging

Rigging is the process of making tools for animation. At our disposal, we have armature which is consist of bones. Next we have empty, constraints and drivers. With these we can make a mesh move, flexing its arm and sneaking in the shadow.

It is very easy to get overwhelmed with the fancy stuff provided in advance tutorials on this topic. Your rig does not need every feature seen on complex character rigs. It is smarter to start simple. Start by adding a bone, get comfortable with it then add more bones in edit mode. Make sure you understand the differences of parent connected and parent keep offset. Also pay attention to bone parenting hierarchy.

Note: The container of the bones is called an armature.

When the mesh is parented to the armature, you have the option to get auto weight. That process is also known as skinning. Skinning is a complex task and will take days on elaborate character models, but the basic is simple. Skinning assigns vertices to a bone in the armature, much like how we assign material to faces. The different weight paint colors will determine how much the bone will change the shape of the vertices assigned when posed.

Note: It is always the best practice to name every bone in your armature. It makes identifying them easy.

A single tab in properties window is designated for constraints. That’s is how important constraints are. Constraints are the first step to rig automation. The very first and most used constraint that you will have to learn is Inverse Kinematics or IK for short. From that you can explore other constraints which will be wonderful for your rigging process.

Note: Parenting is a special sort of constraint.

At times you really do need complex rigging. Rigging a face for example requires the usage of shapekeys, but shapekeys alone are bad interface. We can use mesh objects to drive shapekeys. The mesh(es) can take any shape, facial control can have sliders or knobs or hovering directly to the part of the mesh where it is changing.

There are times you need to use a slider to automate a set of poses and motions. You’ll be using a mesh to drive an animation curve. If you are making games, this type of automation will be used a lot in your game logic.

It is mentioned in the beginning of this section, rigging can get very complex very fast. My advice is to peel it like an onion, one layer at a time, do not cry and wear goggles.

When learning rigging, treat it as doing small studies. Isolate the problem then solve it one at a time.

Check Points:

Start simple.
Learn the basic UI needed for rigging, because they are everywhere.
Study skinning/weight painting methods.
Be crazily enthusiastic toward constraints.
Automate objects with constraints and drivers.
Modifiers and object properties can also be constraints.
Go back to modeling if your mesh is not optimized for animation.

Animation

Let’s assume that you have the perfect rig for the job. It has all the requirements that you need. Now, the fun (also frustrating) part, animation.

Where do you even begin to learn animation?

There are two parts: a. Inside Blender, b. Outside Blender

Let’s start with inside Blender. Get familiar with the Dope Sheet because it feels more natural, similar to how traditionally frame by frame hand drawn animation. You set the key poses, Blender helps with the in-between. When the in-between is not to your desire, it is time to visit the Graph Editor. In the Graph Editor you can change the interpolation methods and do more advance in-between. If the animation was not done by you, starting in the Graph Editor might be the wrong place. Seek first the Dope Sheet and communicate with the one animating it.

Use the Action Editor (switch mode in the Dope Sheet) to make recyclable animations.

What if you have the problem that needs an extra modification of the animation you have done? That’s the task for the Non-Linear Animation Editor (NLA). NLA is a mixer and remixer of the already done animations. Such task example is adding a pre-animated motion with another pre-animated motion, like adding smiling animation to a walking animation to a waving hand animation. You can reuse a lot of animations by mixing them this way.

Let’s go outside of Blender. The first thing you have to learn is capturing strong poses. By strong it does not mean flexing muscle poses, it means the best poses to the emotion or condition of a character. For example, imagine this: A little girl, hugging a pillow, toddling down stairs with tall steps.

When you have strong poses or keyframes, it is time to add motions. These are the rules of motions that can be found in a lot of well known animator must have books.

For NPR, even with a nice NPR shading, if the motion shows that it is CG, it will remove its NPR-ness. Smooth in-betweens must be removed, that means basic interpolation must go out the window. Remember Dope Sheet? In traditional animation, a shot has a big dope sheet accompanying it. In the dope sheet, keyframes are marked, exposure time for each frame is mentioned. This means each image will have a set of screen time. That breaks smooth the animation that CG is famous for.

To enhance your NPR animation, it is advised to add extra animation to each keyframe. Those will act as errors that occur naturally in traditional animation.

Note: Another way to get good at animating in NPR is to watch & analyze animation frame by frame.

For small teams working on a budget (time and money), motion capture (mocap) will help you off the animation burden by a ton. Mocap will require you to invest in

space,
mocap software,
mocap hardware.

These investments are often cheaper than human cost. There are 2 ways to use mocap:

capture motion data to be used directly in animation (minimal to extensive cleaning)
capture motion data to be used only as reference (least cleaning, used in pose to pose)

Which workflow you use will be determined by the style you commit in the design and look stage, or decided by the art director(s).

Mocap tools in Blender still need improvements, there are other software that will bridge the gap. In the end, you have to scout around and ask.

Finally, if your animation process is a struggle, and your rig might be the problem, go back to rigging or even go back to remodel the mesh in better topology. It will be painful to do, but you’ll learn a lot. Trust me, I have been there.

Check Points:

Learn animation tools provided by Blender.
Learn the rules of motions.
Remove smooth in-betweens.
Add extra motions as natural errors seen in traditional animations.
Analyze motions frame by frame.

The Awesomesauce (Post processing)

Remember previously when I mentioned that you need to learn mesh data? That is learning the I/O or the input and output for mesh manipulation. This time the I/O is rasterized (in the future it will be realtime non-raster).

In the post processing stage, these parts are involved:

Render Layers Panel
Render Passes Panel
Freestyle Panels
The compositor window
Video sequence editor window

A rendered frame can consist of many images. Render layer determines what is rendered as a layer. Within a render layer we have render passes. These render passes can be used to enhance or to be combined into the final output with different pixel math algorithm. All the combining processes are done within the compositing node editor, while using render layer and render passes as input sources. Pre-rendered raster images can also be the input sources for post processing.

For NPR, we have Freestyle. Freestyle is a geometry-based-post-processed line art renderer. It is a separate render engine. Since Freestyle is geometry based, when the scene has high polygon count, the rendering time will be long. To cut rendertime, we can select which object layers will be included in the render. This will cut the amount of mesh seen by Freestyle. In the end, Freestyle can act as a layer of line art.

The basics of Freestyle is simple. A render layer has 1 view map. A view map is the geometry data as seen from the camera. If the camera changes location, changes point of view or a mesh changes shape, a new view map will be needed to find the lines. Hence, Freestyle will update view map on every frame of the render. If you have 2 render layers with Freestyle, Blender will create 2 view maps. There are pros and cons of having more than 1 view maps, which will not be discussed here.

Let’s get into Freestyle’s big picture. First, Freestyle loads the geometry that it can sees into RAM. Then it selects which edges will be rendered. After that, it will style the detected edges. For an in-depth understanding of Freestyle, please get BNPR’s Freestyle Level Up course. Within as little as 2 hours, you will start to understand the workflow to get the most delightful looking line art.

All screen space special effects are often done in post-process. Screen space effects are sun ray, bloom, glow, soft lighting effect, color gradient for mood and more. Here on BNPR we have Edge Nodes which also falls into screen space effects. Some styles may need a lot of post processing, some only need a few effects. To work fast, please make node groups of these effects. With time, you will improve the node group for flexibility and style.

There are also object space (or 3D surface) effects. For NPR we use very little of them. For example, depth-of-field can be done with simple render layer blurring; bokeh can be simple floating billboard textures; mist can be blurry alpha transparent texture. If your style is more toward 2D, please avoid using 3D space effects.

A post-processed images can be stored as image sequence. Whether you want to combine them to a final image or keep them as separate layer is up to your workflow.

The final stage for animation is combining all images, effects, voice, sound and music in the Video Sequence Editor (VSE). Blender has a limited featured VSE, but often times it is well enough for the job. VSE has layers which act as holder for sequence of images. How you arrange the image sequence is depending on the transition effects, sequence of the shot and exposure time of the frame.

Note: VSE is also used to plan animation in pre-production stage.

What file format is the best for the final video output? It depends. If you are doing the project for digital cinema, then you have to read image and audio formats for digital cinema. If you are only sharing on Youtube or Vimeo, MP4 will be adequate. If you are selling the video in your webstore, use the most popular file format of the time. If you need DRM, then you have to research more.

Check Points:

Think render layers as layers in 2D raster image editor, render pass as sub-layers.
Freestyle is geometry-based-post-process line art renderer.
For 2D style, avoid using object space effects. No limitation if you are doing 3D style NPR.
Keep rendered frames at highest quality possible as your storage allow.
Combine the rendered images sequences in VSE to produce final video output.

Rinse and repeat (Sharpening your saws)

To stay at the top of the game, you must stay sharp, all the time. The act of renewal is important, even your body renews itself very often to stay alive. Here are some tips to keep you motivated and in top shape in for a very very long run.

Nothing kills faster than the lack of motivation and drive. The solution is to stay curious all the time. How? Scout for NPR related things often. Learn how to do things in new ways. See things from a new point of view. Get motivated when you see things you just noticed, even if you saw the same image thousands of times.

Do studies, do them very often. Studies are meant for calibrating ideas and execution. What is possible and what is not with that particular workflow or paradigm. Studies are best to get those “AHA!” moments. Please do this so often that it became a habit.

From the studies, make notes. Please make your own notes nice, preferably in graphics. Keep it simple yet interesting. From time to time, please read your notes, or maybe improve it with new ideas or perspective. List new found pros and cons.

In NPR, the ability to see style is the utmost importance. Style isn’t just shading. It can be shapes, movements or color palette. See how other productions simplify workflow for style. When you start noticing them, equipped with your own skills, try to recreate what you just saw. Do not give up if what you have created isn’t up to par. It just means you have not deciphered the graphic well enough. So back to the drawing board.

Make artworks often and get each done. Artworks in NPR often are very simplistic. Just like cooking, the amount of ingredients you use do not reflect the final dish. It is the synergy between the least of ingredients which create more than the sum of its parts. Finishing an artwork means you have gained that extra satisfaction.

“This does not look good, but it is done. These are the parts that I need to improve and these I have mastered.”

One should not grow in a bubble. In fact, learning alone is counter productive when motivation is low. Get into constructive discussion often. You will gain extra knowledge while making new connections. Sharpening each other in the process. That’s the essence to grow exponentially fast.

Check Points:

Stay Curious.
Do studies.
Make notes.
Learn to see.
Make stuff.
Discuss.

Closing

Being a master is a long journey, it is best conquered with companions. There is one advice that is often said but seldom meaningful: “Tears, sweat & frustration are part and puzzle of life. Don’t give up because breakthrough might be just few inches away.” Keep digging Simon. 😉

If you find this article useful, please share it around. If you find anything to improve, just comment away. We encourage you to tip bitcoin, donate directly to BEER development fund or buy courses in our store that fund BEER development as well.

This lengthy article has been a collaboration between Light BWK, Lee Posey, Amir and the awesome members of BNPR Facebook group.

Comments are closed.

CoinCoins.net