1-An Introduction to 3D Face Modeling
Creating a believable human face in 3D modeling has been one of the ultimate goals of many artists. Creating an average human face is nothing new or rather an ancient topic. However, to achieve the most realistic and natural-looking possible, we have to make the best out of every step we do: collecting references, planning before modeling, topology, texturing, rendering…etc.
This study only focuses on advanced aspects of modeling such as workflow, diverse approaches, topology, and common errors in head modeling. I will skip through the fundamental things like how to use tools in specific software like Maya/Max or step-by-step instructions, and pay more attention to upper-level things such as overall workflow, a different approach of modeling, topology, common mistakes …etc. As well as other things that you should put your mind on while working to create a realistic head model
2-Pre 3D Face Modeling
Before delving into any actual 3D face modeling work, always prioritize conducting research and planning. This is critically important for almost anything, not just realistic head modeling.
While some experienced artists create without reference, it’s crucial practically and for production. Using concept art or photos significantly influences models. High-quality references are key. Mistakes here can cause issues later, like models fitting the front view but not the side. Also, a model might match the orthographic view but not the person in perspective, leading to time-consuming adjustments and frustration if reference images aren’t as helpful as expected.
To create an accurate reference, photograph the person from a right angle. Incorrect angles will require extensive adjustments. Consider the camera’s focal length; longer focal lengths provide better reference photographs.
A longer focal length narrows the view angle, similar to the orthographic view crucial for 3D face modeling. However, due to view angle distortion, an exact orthographic reference photo isn’t feasible, resulting in 3D models appearing slightly broader than reality. This can be corrected through additional editing in the perspective view. Additionally, consistent and even lighting is crucial to prevent detail washout or concealment caused by overly strong or weak lighting.
Getting reference images into the 3D app:
We should not import the raw photo image straight from the camera into our 3D software, yet, because of the deviation during the photograph process we have to check whether these images precisely fit each other. We should correct our reference images in Photoshop (using lense profile correction to eliminate the lens distortion) or any other image processing application.
Draw horizontal lines across key elements such as eyes, nose, mouth, and eyebrows to check if our reference image is consistent and correctly scaled.
3- Edge planning & Topology in 3D Face Modeling:
We should always plan our topology before face 3D modeling or else we will get lost in the maze of edge loops or lose control of our polycount and end up wasting poly on unnecessary parts. We have to sketch out the main shape of the head, and then based on that we plan our edge structures. Topology is one of the key elements in 3D modeling in general and human face modeling in particular.
Why is that?
Topology and edge loops play an important part in building up the head volume, face characteristics and making up what our head model can do/deform in animation. Unlike other parts of the character, the head has much more screen time with many close-up shots that involve complex facial expression animation like smile, sad, anger… etc. all bring up different shapes and deformations stages of the face.
With a sound and intelligent topology, our head model can deform well into any realistic expression. If we have a messed up topology, our head model can not do a single facial animation. Even if it can, the expression will look unnatural and distorted badly or require heavy skinning/ deform rigging work. Another plus flint is that the nice topology will make it possible to build the model with less geometry but still look realistic and accurate yet easy to edit and maintain.
3.1- Human Face topology
What is a good head topology? How should we know it is good or bad ? Base on what margin?
To answer this question we should go back to our anatomy book and references to look at human facial anatomy. The shape of a human head, winkles, and expressions are all formed and built up by muscles, the shape of the skull, and the amount of fat under the skin. The muscles that we need to keep in mind that play an important role in forming one person‘s facial characteristics ( nasolabial folds) are:
– Zygomaticus major and minor
– Levator anguli oris
– Depressor anguli oris
-Levator Labii- Risorious
But these muscles don’t help in determining the face’s shape or the topology of the surface or the nasolabial folds. Still, we should know about them as they are the source of what causes emotion and how facial muscles work underneath the skin:
“Imagine a thick sheet of foam rubber lying on a table, with strings attached to its underside. When one or another string is pulled, part of the sheet slides and wrinkles. The strings are invisible to us – all we see is the wrinkle on the surface. To get it exactly right we must study the outside shape of the sheet, possibly also the table, but not the strings themselves”
Steven Stahlberg – “D’artiste: Character Modeling” – Ballistic Publishing 2005
The foam rubber is the skin surface which our model is, and the strings are muscle, and the table is skull. Focusing on the movement and form of the outer skin, rather than the muscles, guides our topology decisions based on prominent facial muscle-formed wrinkles. While understanding muscle and facial anatomy is advantageous, overly detailed planning of edge structures for each muscle and wrinkle could hinder the realism of the head model, leading to a dead end. We should exaggerate the wrinkles on the reference images and build our topology.
This is a basic example of sketching out our topology on reference images. We start with the most identical wrinkles and then build up intersected edge flows. Since most human faces are about 99 % similar in anatomy, the topology would not change dramatically. These are some good and bad examples of face topology:
Looking at those good topology examples, you can see that each model has a slightly different topology but is still based on the same concept of edge flows and planning.
”Humans are genetically similar to 99.97 percent or something like that. Yes, the lines in the face look different, but we all have the same underlying topology, the potential for those lines. It’s like lines in the hands of our palms, all slightly different but basically very similar – for instance, you’ll never see someone with the lines going in the opposite direction, from the thumb to the pinkie.”
Steven Stahlberg – “D’artiste: Character Modeling” Ballistic Publishing 2005
“On the guy, the typical doubled-semicolon shape created by the nasolabial folds as they run into the chin, together with the next wrinkle parallel to it. Almost every single person on the planet shows something quite similar. (A few have the nasolabial fold hooking up a bit lower on the chin.”
Steven Stahlberg at www.cgsociety.com
It‘s also better to know that there is no such looping structure in our human face. As we can see in those references, breaking our face structure into edge loops is only for the sake of sufficient simplification of polygon topology. So that the face’s topology in 3D technical terms is approximated and there is no absolute perfect topology for the head; it‘s dependent on our reference and the purpose of the 3D modeling.
3.3- 3D Face Modeling Rules:
The rule for human face modeling is the same as for other objects:
– Keeping a suitable tris count (not too high for easy editing, not too low for better 3D face modeling definition).
– Use quads as much as possible and use triangles when necessary, triangle are not recommended but it doesn’t harm having a few triangles to terminate our edge loop to avoid adding extra unnecessary loops, but keep in mind the fact that the model might not deform nicely when animated and smoothed. Still, it will save us unnecessary edge loops and faces.
– Avoid poles (where vertex share more than 4 edges, it’s impossible to avoid pole in our model but we should hide these in those areas that are less seen in the model because the areas that have poles are not smoothed properly)
– These are some areas that usually have poles and they are acceptable.
Polygon should always be chubby or square, not long thin, or diamond-shaped (some are acceptable to form wrinkles ). Square & chubby polygon hold up nicer when being deformed.
– The edge direction should not run across the curvature line of surfaces.
– The worst case is topology running at 45 degrees to the curvature line of the surface form that the diamond shape deforms which will look terrible later in render and smoothed.
”The most challenging part is finding the balance between having enough detail to achieve what is needed and building as light a model as possible”
Francisco A.Cortina “D’artiste: Character Modeling”- Ballistic Publishing 2005.
The more edge and polygon we have, the more accurate and realistic our model is but the more complex we can edit and maintain it. This applies to both still images and animation. With still images sometimes it’s okay but it’s a good practice to keep our model as light and low res as possible. If we want to have extra details, the critical thing lies in the texture part—especially typical and displacement maps.
Main topology lines in the 3D head modeling:
Green lines are mainly to help polygon flows form the shape of the body part based on surface curvature while Red lines will play their role during particular extreme expressions.
Note: Red line loops will vary in different head models based on animation requirements; the more extreme expression & wrinkles the face has, the more extra red lines must be added to the face to build up the deformation properly.
4- Different Ways to Execute 3D Face Modeling:
There are many ways of modeling a human head, and everyone finds their own comfortable way to model. These are some primary methods that I sum up:
4.1- Polygon Modeling:
Polygon modeling is the most popular, essential, and straightforward 3D face modeling method. Within polygon modeling, it has different ways:
– Start from a early shape ( cube or sphere… etc.): some people start by modifying a basic mesh like a cube or sphere. We block out the overall shape of the head first and delete some unnecessary face and edge loops to form a basic topology. Then we start going into detail modeling by adding more edges and faces around areas like eyes, nose, mouth….etc
Start from curves/lines and separated parts: Some people prefer to draw curves (mentioned above as green lines) that run along the surface’s curvature of body parts in reference images; these lines together will form a basic head shape. After all, extrude the polygon around the lines and start building separated parts of the face, connect them together when done, and carry on editing those gaps (adding red lines) to form a completed face.
Polygon modeling lets us have direct control over edges, faces, and vertices; it’s fast and easy to keep track of the 3D face modeling process however it requires a lot of editing work and sometimes does not work efficiently with surfaces that need high smoothness and accuracy. However, with human head 3D modeling it works best in my opinion.
4.2- Sub-D Modeling:
SUB-D : this can also be called polygon modeling. Not many people choose this method to work with.
In the beginning, it is the same as polygon modeling. Once we have our basic head model, we can go into detail by adding edge flow to areas that have more detail (need more edge/vertices) than other areas like eyes, nose, and lips. N-gone and unterminated edge loops are allowed at the base mesh because we must smooth the whole model to convert all the polygons to quad. The advantage of this method is the hierarchy workflow it allows us to easily edit the vertex at the base level if we want to change the overall shape of the head while don’t need to worry about the detail/extra edge loops created by those detail areas
The other advantage is the low – res version of the model doesn’t need to be perfect in topology when we go to a higher level, all these N-gone will be all quad and smooth without losing any detail.
4.4- Zbrush/Mudbox Sculpting Modeling
Sculpting: an alternative way to model a human head is using 3D sculpting software like Zbrush or Mudbox. We can start from a very basic rudimentary base mesh and then sculpt and add detail to the model. After you finish the sculpting proceThe advantage of this method is that we can detail our head model with millions of polygons, on the plus side: the process of modeling and topology is separated so we can concentrate on one thing at a time, modeling first, topology later! Derived from the high-resolution model we can generate normal maps, cavity maps, displacement maps, etc. These textures will help us a lot in texturing works later.
The disadvantages are: This method will take a lot of time, and we have to create both super high-res models and low-res models. In addition, we need to have good sculpting skills to sculpt an excellent high-resolution model with details. But these are only a few disadvantages I can find for this method. It is really good and is being used by many artists nowadays. Sculpting has been becoming mainstream day by day.
Among all the 3D face modeling methods I mentioned above, I am comfortable with the basic polygon modeling which I will use to create the base mesh in Maya. Then I will bring it into Zbrush for sculpting and re-topology. Then I can export the lowest-level mesh in Zbrush back into Maya and apply a displacement map & normal map generated in Zbrush to make it look like the original high-resolution model.
Nevertheless, we can use the retopology of the head whether it can be done in Zbrush/Mudbox or your 3D package to create a lighter/topology-correct model but still keep the similar look of our high-resolution sculpture model.
5- Details & tips on 3D face modeling work:
However, despite the 3D modeling method we are using. Below are some tips & tricks and details we should pay attention to.
5.1- Eye areas
“Eyes are the window of the soul,” they said. Eyes are the critically important element in painting as well as 3D. Viewers always look at the eyes of the character first. The eyes make the most characteristic out of the character’s face. So it is never redundant when we spend extra time on the eyes.
It is better to use a sphere as a reference to model the eyelids or else they will look unnatural and intersect with the eyeball. The eyelids are decided mainly by two curves that are formed by the upper and lower eyelids. We should also have an extra edge loop inward to make the thickness of the eyelids. The angle of the eyelids’ thickness also has to pay attention to, so that they can catch light better.
The eye’s canthus should be modeled with a real polygon (i.e., not just texture/bump map), or else our eyeball won’t fit in the eye holes. Some people make the eye’s canthus separately from the face, but this will make it look like dirt in the eye’s comer so it’s better to have everything as one union mesh object.
It is not recommended when making an absolute circle edge loop around eye areas. Instead, we should have a line go out from the outside eye corner to help build up the volume of the upper eyelid. These small details are not noticeable from a distance, but in close-up scenes they make the eyes look much more realistic and soulful.
5.2- Eye Balls
Many people underrate this part and they only use one single simple sphere for the eyeball. This makes the eye look robotic and dollish. Looking at a realistic eyeball, we can see that the eyeball looks much more profound and glossier.
The key is we have to mimic the real anatomy of the human eyeball. The most used method in modeling realistic eyes nowadays is breaking the eyeball into two separate parts: cornea and eyeball.
– The cornea is like a sphere but more bent forward at the center, and its transparency only catches specular, reflection, and even refraction.
– The eyeball is a sphere but sunken inward at the center to form an iris.
This detailed 3D modeling can make our model’s eyes believable.
The water part between the eyelid and eyeball should also be modeled as a separate mesh object to have a realistic-looking water ring around the eyes.
5.3- Eye lashes & tear
It is vital to get the natural shape of eye slashes right. The many features of the eyelashes determine the shape.
“An important thing to remember is to correct the amount of bunching and randomness in both length and placement. The lower lid’s lashes tend to bunch in smaller numbers, while the upper lid’s lashes bunch in greater numbers. The small membrane on the inner corner of the eyelids, or the ‘caruncula lacrimalis’, is one of the most important elements of the eyes.”
Francisco A.Cortina – “D’artiste: Character Modeling” – Ballistic Publishing 2005.
Many people use an image plane with eyelash texture on it. But this will make the eyelash look flat, and not good enough in close-up shots. I do some extra study of different methods in making eyelashes. Some people use Hair & fur in Maya or the Hair plugin to make this, but it’s still not really food for close-ups. I discovered that the best way to do eyelashes is: polygon eyelashes! Despite what method we use, the output should be a polygon. Polygon eyelashes have a better look with more depth feeling and cast more excellent shadows.
We can use Maya paintFX or a third-party plugin to make the eye slashes and convert them to polygon. I am comfortable with using Maya paintFX hair to model the eyelashes and then convert it to a polygon. I can still render the eyelash properly in polygon form while keeping the construct history of PaintFX, I can modify the randomness, the thickness, and the number of strands).
“Both nose and lips contain some of the most subtle variations in shape on the faces. As with drawing and sculpting faces, the key is understanding the planes and angles they flow based on the underlying bones and muscles.”
Francisco A. Cortina -“D’artiste: Character Modeling” – Ballistic Publishing 2005.
Through research, I found some excellent topology layouts for those areas:
The edge around the nose area forms a round curve that starts from the top of the nose and goes along ala and terminates at the area where Philtrum meets the nostrils. The mistake we often have is that the edge loop doesn’t go around the nose but a grid-like edge loop instead, it makes the nose look square and big.
In the lips area, we should have at least 2 edge loops go out at the lip’s corner or else the model’s mouth looks like it is being cut. Also, to emphasize the shape of the lips, we should add an extra edge loop around lips areas because without this edge loop when we smooth or subdivide one more level, the lib shape will be smoothed out.
In addition, we model the head with a closed mouth but it’s better to have some extra face loop inside the mouse to hide the border edge when the character opens his mouth.
At the chin area, we should terminate the edge from above with a square polygon’s area. This topology will make us adjust the shape of the chin more efficiently or just in case our character has a double chin. We should also have an edge loop starting from the chin and terminating at the ear, this will form our character’s jawbone more nicely.
The ear is a very tricky area in the face. It has a spiral structure with a lot of detail. If we are not careful, we may spend too many polygons on it. The easiest way is to model the ear separately while considering that its edge flow needs to meet the head topology. In the end, we can sew the ear to the head model, we may have some triangles around the sewing area, but they hardly ever deform while animating, so it is safe. This separated ear model can be reused for any other models in the future.
6- Optimizing in 3D Face Modeling
We must consider whether our 3D face modelings are good enough to move on to another stage, like UV or sculpting. We should continually review our topology before moving to the next step to check for unnecessary edge loops, triangles, badly deformed faces, etc. This will make our work much easier later.
– UV mapping: It is necessary to keep the model’s topology as symmetrical as possible unless we must model the character with half of his face altered. This will significantly help us in UV and texturing, whether in Maya or other software.
– Animation checklists
– Review our topology to check whether our character’s face will have an excellent deformity when expressing.
– If our character talks or opens his mouth, we must hide the open edge of the inside lips area or model the mouth’s inner interior.
– Putting an extra edge loop in the upper eyelid area so we will have enough polygon for the eyelid to cover the eyeball when the character closes his eyes.
– If the character has a frowning or glaring expression, the forehead area will form some wrinkles. Unless we want to do it by texturing, we should add some extra edge loop around the forehead area to have enough polygon to make those wrinkles deform. But this is not always necessary if we don’t want such realistic deformities.
7 – Final thoughts on 3D Face Modeling Work
Modeling human faces is a common topic in 3D graphics. Face modeling is a simple task, but the most difficult things lie in the harmony of details and shapes that make up the unique characteristics and nature of the face. A small change can lead to a vast difference. Sometimes it’s difficult to evaluate whether the face looks right or wrong. If it looks wrong, which part should be edited? A realistic and successful face model depends significantly on texture, shader, and lighting. However, these parts can never be successful if they don’t have a solid, steady base – the model itself.
Let’s take a look at our most recent 3D face modeling work.
ArtStation – Windaria – Showcase ArtStation – Subverse – Taron