Creature Proposal
Pre-Production
Research and Inspiration
CG creatures and animals have been a staple in the VFX industry for a long time yet remain some of the hardest assets to pull off convincingly. This mainly stems from the sheer complexity that rebuilding a living, breathing creature brings. Creatures and Animals have to be designed from the ground up with working skeletal and muscle systems. Using accurate and reliable references from various different sources is vital to this process.
Figure : PureRef Design board for The Nyala Project by Nicole Jackson
Usually during this stage concept artists will gather various different resources depending on the kind of creature that they are after. For fictional creatures like the Demogorgon (Stranger Things) or a Chimera, from Greek mythology, they will gather references for various different real-life animals with anatomy similar to that of the creature, i.e. goats, lions, and snakes/serpents. This will help base the anatomy of the fictional creature in some form of reality and provide a solid foundation from which to build upon. Adding these layers to a CG creature helps sell the photorealistic look as you can see the different muscle groups and bones working underneath the skin. A famous example of this would be James Cameron’s Avatar: The Way of Water. The majority of that was CG environments and characters with the underwater sections having a plethora of various creatures and life-forms that take inspiration of real animals. The Weta VFX supervisor, Pavani Rao Boddapati, cited using Inspiration and references from real-life ecosystems like the shallow Reefs Found in Bora Bora (Greene 2022). In addition to this the filmmakers also look at exampled of bioluminescent plant-life to help them ground the fictional ecosystems of Pandora in reality keeping viewers emersed on the world.
Figure : Example of skeleton, muscle and skin system for CG creatures and Characters.
All of these have to be carefully planned out ahead before starting the sculpting workflow as it will define the order and methods used in the workflow. Designs and concepts will often bounce several people like the technical directors, producers, and directors as well as others to make sure it is in line with all the various aspects like genre, style and that it is feasibly possible to achieve. Once the final designs have been signed of the sculpting and 3d modelling processes can begin and depending on the creature this can through a variety of methods.
The main inspiration driving my creature comes from a Hellhound from Greek Mythology. They are typically described as being large supernatural dogs that are associated with fire, death, and the afterlife. Despite their hellish name they are not inherently evil but simply guardians to the gates of the underworld. The most famous known Hellhound in mythology would be the large three headed dog Cerberus, though not all hellhounds follow his design. They can be any type of dog breed with pure black fur, glowing red eyes, increased speed and agility, a foul odour, and phantom abilities such as shadow travel to and from the Underworld. Some folklore says that if you hear a hellhound howl it is a bad omen and looking into their eyes three times or more will cause your soul to leave your body, which I will aim to embody with the final design of my hound (Harrington 2017).
Figure : Illustration of Cerberus, the guardian to Hades’ domain in the underworld.
For my take on this mythical beats, I will be taking heavy inspiration from its commonly found description, forgoing the three heads, and combining it with living creatures such as known dog breeds as well as wolves and bears. The two main animals I will be drawing inspiration from are the British Columbian wolf and the Great Dane. I will also be looking at Mastiff dogs and grizzly bears for alternative facial features and physiology.
Biology and Physiology
Anatomy
The biology of the hellhound will primarily draw inspiration from the British Columbian wolf but also the grizzly bear. This is to give it a bit more muscle and physique to put It more in line with the original myth and give it a bulkier look compared to household dogs which tend to a bit slimmer.
Figure : Hellhound anatomy refence pictures.
To try and bulk up the physique I will be taking inspiration from the grizzly bear in the top left of Figure 4. Particularly the heavy build-up of muscles in and around the upper legs and neck where a standard wolf or dog may be lacking. For the main body composition and design, I will draw more from the anatomy of the wolf muscle system shown in Figure 4, keeping the body itself more streamlined and speed orientated to keep in line with the mythos. If time allows, I will try and build a skeleton system to drive the hellhound and its animation down the line which is where the dog skeletal references come in. Bears and dogs have a remarkably similar skeleton which allows me to easily combine the two together to create something faster and stronger looking.
Head
The main part of the heads that will be a focal point are the eyes and the fangs since they will be the most prevalent in the shot alongside the mouth and fangs.
Figure : Reference images for wolf fangs and the glowing eyes.
Hellhounds are described as having glowing red eyes in the folklore which is what I will be trying to emulate in my creature. In Figure 5, you can see some reference images I have picked out which highlight the kind of look I am going for. A more natural style wolf eye, with beautiful corneas and irises like in the top two images but a with a darker, red tint to them. If time allows some particles and red flames will be coming out from the corners casting a red hue on the nose like the two right images. This gives a more demonic and ghostly vibe compared to the softer and friendlier looking tones in naturally occurring wolf eyes.
The teeth and fangs will also be largely based on wolf physiology as well as regular canine teeth. The main highlights of this will the four big, large fangs that feature on the front of the snout and the smaller canines surround the edges of the mouth. In the animation, the hellhound will be shown barring its teeth to the camera to show of the anatomy so these will need to be fairly detailed particular for use in Creature FX later.
Body and Fur
Figure : Reference Images of different body and fur types for canine animals.
The core body shape will be largely taken after Columbian wolves, mastiff dogs and Great Danes since they have largest statures of the canine family. As mentioned earlier, the grizzly bear will also be used as inspirations when drawing out the muscle groups to bulk up the frame a bit compared to the smaller canines. I have done this because hellhounds are not only meant to be incredibly quick and agile but also insanely strong compared to their natural counterparts, hence the added muscle. The ears I will be using will more closely resemble the Columbian wolf, with maybe some iterations more similar to mastiff breeds depending on which ends up fitting better to the intended theme. The fur type will be a cross between the shorter haired dogs and the longer haired wolves due to both the limitations of this project and the aesthetical considerations compared to the myths. The colour of it will certainly though be as black as night since that is a defining characteristic of the original creature. The tail bones will be more akin to that of the Great Dane with a long, trimmed, slicked back but powerful tail giving it more control and stability in its movements. Columbian wolves tend to have thicker, bushier tails which is not fitting the image of a sleek, agile demon which is the reason I will be basing it more on the short haired dog breeds. Though given the required speed and strength a hellhound should possess, the legs will actually be similar to that of a greyhound dog with their large leg muscles which allows them to achieve the insane speeds for which they are known.
Figure 6: Reference images for the fur texture and length.
The look development will be mainly focusing on the black fur and getting the texture and materials to look right. This will be primarily started in Zbrush using various brushes before being moved into substance painter, Mari or even Houdini (if time allows). As mentioned above the main parts that will be looked developed will the fur, particularly how it interacts with the nighttime lighting in the scene.
Details
Figure : The Three Wise Monkeys – “Speak no evil, hear no evil, see no evil”
As part of the overall Cerberus design, I will be playing with the idea of ‘hear no evil, see no evil, speak no evil’ which is a Japanese idiom originating in Buddhism (Three Monkeys Restaurant Phuket 2025). It stems from Buddhist teachings and from the belief that if you don’t hear, speak or see evil you will be spared. This seemed like a theme that could nicely tie into the Three Headed Cerberus: The Guardian of the underworld.
Figure : Hear no evil, see no evil, speak no evil Cerberus inspiration
Taking inspiration from Figure 8, the three heads of Cerberus will have different levels of ‘damaged’ applied to it, each one a different method sensory deprivation. One head will have its ears be ‘ripped off’ leaving behind some serious scar tissue and wounds, leaving it deaf. The middle will be completely blinded, missing one eye completely, leaving behind an empty eye socket with scratches all over as if it had been clawed out. The remaining eye will have a long gash running through it, taking inspiration from Scar in The Lion King. The remaining head will take inspiration from the Shrunken heads out of Harry Potter; the jaw will be completely sewn shut with no inside mouth or teeth visible to complete the look. All of this will be accompanied by heavy s and wear and tear all over the body to show that this hound was seen some action.
Production
Sculpting Techniques
The initial Sculping workflow can vary drastically depending on the studio and the creature itself. One of the main pieces of software used in Creature creation is Zbrush because of its powerful sculpting capabilities and advanced detailing. There are several ways to start a new sculpt in Zbrush the primary two being working from primitives or using ZSpheres.
Figure : Example of a base mesh constructed using ZSpheres.
Z-Spheres
ZSpheres are a powerful tool that allow artists to create complex base meshes. They have an advantage over using primitives for creating base meshes since you can go back and easily edit and transform it at a later date.
Figure : ZSpheres
ZSpheres are basically spheres which form together to create a malleable armature which can be easily posed and manipulated to fit the desired design. They can be advantageous when creating characters with complex anatomy since it allows artists to work using a parent-based system emulation a skeletal system and marking out anatomical indicators like elbows, knees, hips, and other joints. Once the final armature is complete it can be easily converted into a sculpt able mesh by clicking on ‘Make Adaptive Mesh,’ this allows for the mesh to be edited and sculpting on using the various brushes as well as ZModeller. Using this method is generally faster and more flexible when making a base mesh however it outputs very heavy geometry. This results in the base mesh having to be re-topologized to create more even geometry on which to add fine details.
Figure : Base mesh created using ZSpheres in Zbrush for Cerberus Project
I opted to use a ZSpheres workflow for the main body of my creature as it’s a single mesh and allowed me to quickly and efficiently block out where I wanted certain aspects of the creature and how big they needed to be. This mesh was then turned into adaptive skin and manipulated further in Sculpting mode, whilst working to reference, to get a more defined and blocked out base mesh which I could easily work with. After creating a skin from this and remeshing it into workable topology I will have a good foundation on which to start to build some muscles on top and get a basic system going. This will be done using primarily Z-brushes standard brushes like Clay build up to get some depth and layers on the creatures. I will block out the different muscle groups and systems and separate them into playgroups for a better workflow and animation later on. Once I have a lower poly model with the muscle details backed in, I will start adding in finer details and beginning to work on the fur. I ended up, separating the head from this and focusing mainly on the shoulders and main body. This was the head had to be a lot more detailed than the overall body which can’t really be achieved easily using ZSpheres.
Using Primitives and Dynamesh
The other main method is to use a series of primitives to create a base mesh and make use of Dynamesh to fuse the various objects together.
Figure : A Nyala skull sculpted in Zbrush from a primitive sphere.
Dynamesh is one the most powerful tools in Zbrush since it allows you to dynamically re-distribute geometry across a model. This is most useful in the earlier stages when the model is still in base mesh form since it allows you to pull and stretch the model without having to worry about going deeper into subdivisions. It also allows for two sub tools to be merged and then combined into a single mesh with even geometry. This way of working can be beneficial when creating smaller or more defined base meshes or skeletons than when using ZSpheres. Though this also tends to create quite heavy geometry when using a higher resolution setting so using ZRemesher to re-topologies the mesh later would be recommended.
Figure : Cerberus head made from a primitive Sphere
As Figure 13 shows, the head for my dog was made from using a Primitive sphere as the base shape, and two sperate ones for the ears. This was then manipulated in sculpting mode using brushes like Move, Clay Buildup and Damien Standard, to create the outlines and main shapes of the head. A similar process like this was applied to the other aspects that were modelled separately of the body. This included the feet, eyes, ears and jaws, all of which were created using primitives like spheres or cones as a base before being broadly manipulated in sculpt mode to get the desired shape. All of these subtools were then merge into one subtool and conjoined using Dynamesh. As explained, this re-calculates the existing meshes and filling in gaps where possible, to merge it all together into one cohesive mesh (Maxon 2025). It dynamically distributes polygons evenly across the model meaning that low poly areas get denser allowing more detail to be sculpted, serving as a great starting point on any mesh.
Orthographic View and References
Figure : Using orthographic images for my Cerberus Creature in Zbrush
Regardless of which method is used to create a base mesh, using references and orthographics is vitally important. This particularly applies when creating anatomically correct creatures and is just generally considered good practice. When in Zbrush you can lock onto certain perspectives, namely, front, back, right and left. When in orthographic view mode, perspective shift is disabled which means that there is no distortion of your model which is ideal when using references images, as seen in Figure 14, (Maxon 2025). This helps you look in the base shape and outline of a mesh based on the refence work you are using. Figure 14 shows one of the orthographics that got used when I created the base mesh for Cerberus using ZSpheres. I based the body off Columbian wolves and had various orthographics to help me get details down.
Adding Details:
Once a base mesh has been created using either ZSpheres or Primitives and Dynamesh, then you can start adding details in different subdivision levels. This best works when saving your history and then using ZRemesher to create a more even and simplified topology at a lower poly count. We can then re-project the details back on using the ‘Project from History’ function. This adds in some of the details back on to the lower poly mesh. It’s important to maintain a lower polygon count since it can significantly increase the computational power used to render and process the model. An animator is not able to use a very high poly mesh effectively so it’s important to maintain a steady poly count by working in different subdivision levels. This lets us control the levels of detail within different layers which can be separately exported.
Figure : A ZRemeshed model split into different polygroups.
At this stage it is also recommended to split your model into different polygroups based on the anatomy of the creature as seen in Figure 15. This essentially groups different areas of polygons together telling Zbrush that they are related and should function as one large group. This method allows you to isolate certain areas of a mesh based on the polygroups and use brushes that will now only affect the one defined area. This can be used to help guide edge flow and topology when using ZRemesher allowing for a cleaner workflow. Certain tools like Insert Mesh Brush and Boolean brushes use polygroups to define their area of effect giving artist more control over their mesh.
Figure : Example of a low poly mesh vs a high poly mesh in Zbrush.
Subdividing simply means dividing each polygon into 4 smaller polygons within the parameter of the first one. This effect essentially quadruples the polygon count for the overall mesh making each subdivision layer denser than the last. For this reason, it is generally recommended to stay within sub div 5 or 6 so the high-resolution mesh can still be easily worked on. Each subdivision level allows artist to create finer levels of detail-like pores and small blemishes and scars that would not be possible at lower poly counts.
Subdivision Level 1
Figure : Subdivision 1
Subdivision Level 1 is typically reserved for just the base mesh and larger shapes. In Figure 17, you can see how I’ve blocked out the body into its broader shapes, with the jaws, eyes and ears being separate subtools since this allows for easier animation later. There’s not a whole lot of detail here as the density of the mesh doesn’t allow for it. However, for the same reasons this can be useful for manipulating large regions of the mesh without focusing too much on the smaller details (Novedge 2024).
Subdivision Level 2
Figure : Subdivision Level 2
Here you can see some details start to be added to the mesh, predominantly large details. Things like thick veins and the outline of the scar on the front head are being blocked out at this level to give the rough shape whilst leaving details for smaller blemishes. Further muscle definition on the neck and shoulder regions was also sculpted on this sub div level as they don’t require fine details and can be managed with broader brush strokes.
Subdivision level 3
Figure : Subdivision level 3
At subdivision level 3 were starting to get into quite a denser mesh, now operating at 200,000 polygons compared to the base mesh of 14,000 polygons. This lets me add some smaller details like larger wrinkles and skin pores as well as the flesh detailing on the head wound on the left side. Extra detailing to the scars on both eyes was also added using a variety of different brushes like DamStandard and Clay buildup to create clear and defined lines.
Subdivision level 4 and 5
Figure : Final details at subdivision level 4 and 5
For the Cerberus model subdivision 4 and 5 were for the highest levels of detail, reaching a maximum of 3.8 million polygons. These higher levels are perfect for creating skin level details like small blemishes, skin and muscle pores. On my model, presents itself mainly on the surface skin texture and the flesh details in the head wound with highly detailed scrapes and blemishes visible. These kinds of details wont actually be part of the final geometry but rather part of the displacement and normal maps, which are later applied in texturing after retopology.
Projecting Detail
Figure : Projecting high-poly details from history
At this stage the model has become far too dense to work on reasonably outside of Zbrush, applications like Maya and Houdini don’t handle well with meshes that dense. This is where retopology, which is explained below, and projecting details from history comes in. Once the base mesh has been retopologized, using a variety of methods, you need to ad our original high-level details back on. This can be achieved by reimporting the new low poly base mesh into Zbrush and creating the desired subdivision levels, capping at around 5 or 6 typically. Then on your original high poly mesh you save the history of that model in the timeline. This allows you to go back to each subdivision of your new, retopologized model, and project the history back onto each subdivision. Now you have a deconstructed, retopologized mesh in Zbrush which you can sculpt further details or hand of for rigging and animation,
Retopology
Figure : Manually Re-Topologizing a head in Autodesk Maya.
When creating a CG creature for VFX use in films/episodic as well as games, it is important to consider polycounts and topology flow to allow for easier transfer across the pipeline. There are several ways of re-topologizing a model, usually split into Automatic and Manual methods. Automatic methods include tools like Zbrush ZRemesher which creates even topology based on the polygroups defined by the user. Whilst this methods can provide good, even topology, it is usually beneficial to retopologize a model manually since it gives the artist more control over where they want their model to deform.
The industry standard software for this is Autodesk maya since is allows artists to easily quad draw polygons on even a high poly mesh without slowing performance. A mesh with good edge flow will have edge loops that flow the shape and contours of the model (Nguyen 2024). This is necessary to allow for better deformation in the mesh when animating particularly when creating blend shapes and painting skin weights.
Figure : Bad mesh example in Maya by Thomas Husband
In Figure 23, you can clearly see how on the right, how the messy topology is causing the mesh to deform in unwanted ways. Having proper edge flow here means that the topology follows the contour of the face, as is shown in the left of Figure 23. The topology is nice and even, keeping withing quads and deforms to the shape of the face meaning when a blend shape artist animates the model it deforms how you would expect it to rather than creating the artifacts we can see in the messy example above.
Figure : Example of a retopologized shoulder in maya for the Hellhound Creature
Maya is a leading piece of software for this workflow, offering a vast arrange of tools and methods to help maintain correct edge flow. For my workflow, I started by exporting the model as a GPU alembic cache and reimporting that back into the scene. The once made live, there is no poly data available, but you can still use quad draw on the mesh. This allows me to draw quads directly onto the model making it easier to follow the shape. When re-topologizing there are a couple of guidelines to follow to ensure the model deforms accurately when animating. A good rule is not to have no more than 8 edges in a loop so as to not have over complex geometry resulting in artefacts or improper deforming during rigging and animation. It is also best practice to align poly-loops with the flow and contours of the model, so the topology adheres to the anatomy of the creature as explained above. Figure 24, shows how I tried to maintain good flow around the edge of the shoulders, making sure the edges curved around the base and underneath the armpit for better deformation. There were some issues with certain areas of the model, particularly creating the jowls around the mouth since they were a bit thinner quad draw was having a hard time mapping it out and joining it up the rest of the mesh.
3D Modelling and Baking Details
Whilst creatures primarily consist of soft body workflows given their organic nature, many aspects may still require hard surface 3D modelling techniques. This is particularly present when making fantastical or humanoid creatures which would be wearing armour or clothes as part of their overall look. These would be created using a more traditional modelling workflow in Programmes such as Maya or Cinema 4D and or even Zbrush’s ZModeller.
An especially important part of creating a CG creature is using 3D modelling to control polycounts and exporting higher resolution details as texture map. Software programmes like Maya can’t handle very dense geometry very well, drastically increasing processing and rendering times. To get around this, the high-resolution details sculpted in higher sub div levels are exported as normal and displacement maps. These can then be applied as a texture to the lower poly base mesh to create the illusion of fine details.
Figure : Example of a low poly model using normal maps for fine details.
These maps can be then baked onto the model to create detailed geometry without the high poly count. In order for these to conform correctly on the model it has to be properly UV unwrapped which is done in software programmes such as Maya or Zbrush’s UV Master Plug-in. Some large studios like MPC film and artists working on large budget productions such as Disney’s Lion King can use up to 50 UDIM tiles for each creature due to the sheer amount of complexity and detail required (Guimoneau & Capellazzi 2019).
Figure : Example of UVs created for the base mesh of the Cerberus Creature
Post-Production
Rigging
Placing Joints
Once the final creature has been sculpted, with a well retopologized base mesh that has proper edge flow, will it be passed onto a rigger. This is a vital part of any creature production workflow that aims to have it be moving in the final shot. A riggers job is to create skeleton rigs which bind to the mesh of the creature allowing it to deform and be animated (Screenskills 2025).
Figure : Example of character in Autodesk Maya
To start of a character or creature rig, the model has to be in a neutral or T-pose depending on the if the creature is a quadruped or biped. This ensures that the joints will deform the mesh evenly and there are no weird artefacts present when trying to animate or deform the mesh in certain ways.
Joints are the backbone of 3D rigging; they act as the skeleton for the model and are connected by bones or controllers to create a complete rig (University of Silicon Valley 2025). Joints are placed inisde the mesh when loking at it from a orthographic viewpoint and when placed in succetion form a joint change. It is best pratice to use orthographics when placing joints so you make sure to keep as close as possible to the reference animals and real world biology.
Inverse Kinematics vs Forward Kinematics
There are two key types of animation and join placement when it comes to rigging, Forward Kinematics and Inverse Kinematics. Forward kinematics starts from the base of a limb, such as a shoulder or hip bone and works it way to the end, resulting in a linear line of progression and movement (Moulakhnif 2025). This is typically used in leaning moving joints like a central spine or when needing to create rigid movements that need to be precisely controlled. IK allows for a lot of granular control of each joint which can be helpful in certain situations but also results in unnatural looking movements.
Inverse kinematics works in the opposite direction, starting at the base of the root, like a hand or foot, and going back up the chain, allowing for smoother and more natural animation (Abrams 2021). Instead of controlling each joint individually you create constraints in between them and use a Controller on the base joint to automatically rotate the join based on the relative position of that constraint. IK systems are frequently used when you need naturally looking movements like arms or legs in a walking animation.
Figure : Using IK controls for the Cerberus Project
For my Cerberus creature I used both IK and FK joint systems within my rig. IK systems were predominantly used for the two pairs of legs as these needed to be able to move in a natural way, bending at the knees whilst remaining constrained to the hip and foot. I also ran an IK joint from the wrist to the toes and one between the wrist and heel. This would give me some deformation control over the toes and heels as it moves allowing for more natural rotations.
Controllers
Controllers are typically meshes or curves that allow animators control over certain parts of the mesh in an intuitive and visual aspect. Animators use these to keyframe their animations rather than individual bones since controllers tend to work in tandem with certain FK and IK constraints that are set in place (Rokoko 2022). They allow for an easier workflow and maintain the complex constraints between joints whilst maintaining the ability to be reset to default positions.
Figure : Controllers for Cerberus Creature Rig in Maya
Figure 29 shows the controllers used in for my Cerberus rig, separating out each side and making sure to colour code the sides for easier differentiation for the animator. For the feet I used a separate controller for the heel, paw and overall foot. The toe and heel controllers are centred on the mid joint in the foot with the foot controller being centred on the heel. These are then parented to the corresponding IK handles rather than individual joints meaning all 3 IK systems work in tandem with each other and allow for smooth animation.
The three head controllers are parented to the joint at the base of the neck, working purely on the rotational axis rather than an IK system. This is because there is only one joint in between the head and neck which only requires some linear movement. These three controllers are parented to the hip controller which means when the hip moves back and forth that the heads don’t get left behind and deform weirdly.
Driven Key
Figure : Using driven keys for the tail controller of the Cerberus Rig
Driven keys are a great way to rig certain parts of a skeleton, particularly features like a tail, since they create an association between two pairs of attributes (Autodesk 2024). This method allowed me to link the rotational attribute of the tail joint to a custom scalable attribute named tail swish. Then, by setting a key for the rotation attribute at a value of -10, 0 and 10, I can key that to the new custom attribute. This allows an animator to simply keyframe the tail swish attribute in the channel box to get the tail to move dynamically from side to side. This process can also be repeated with a tail height attribute for a similar effect. These can be extremely useful for animators as you can efficiently automate relationships between various controllers and create complex animations.
Binding and Weight Painting
Once the overall rig has been created and the various controller and IK handles are in place, the joints need to be bound to the skin. In Maya, this process binds the closest vertices based on their location in relation to the joints and then assigning influence based on this (Pluralsight Content Team 2014). This creates a base set of influence for each joint and how it would interact with the mesh. The complexity and quality of the model will drastically impact the accuracy of the initial binding. This is where weight painting comes in, as the majority of the time, the mesh requires manual weights being applied to it. This is the process of assigning weights to each vertex of the mesh, which indicate how much influence each bone should have on that vertex (Oneil 2024).
Figure : Painting weights for the shoulder joint on the Cerberus model
For the Cerberus project, I used Maya’s initial weights it calculated form the binding process and the manually adjusted the weights based on how it should be deforming. This process involved in going on every joint and making sure the wights matched what it realistically should be affecting. During this process, some notable issue arose with the way the legs and shoulder, in particular were modelled. The legs were not separated enough from the rib section meaning the deformations were not ideal with the shoulder influences those certain areas. This was minimized by painting the weights so there was a clear line between the shoulder and ribcage area. During this process, I created a basic animation to help test the various ways in which the mesh deformed when put into different poses, essentially creating a stress test for the model which can be seen In Figure 31.
Figure : Stress test animation to help with analysing weight painting.
Delivery
Once the creature has been completely rigged and weight painted, with suitable controllers and constraints will it be passed onto the animators. As part of this whole process the Rigging Technical Director will testing out the final rig in various positions where it would need to deform, checking for artifacts and inconsistencies (ScreenSkills 2024). The animators will also participate in this, with the rig going back and forth until everyone has signed it off.
Figure : Reference image for a wolf walk cycle.
Figure 32 shows the animation test created for the Cerberus project showcasing the different controllers and how they affect the model. If I where to take this project further I would look at the reference seen in Figure 33 to create a realistic walk cycle animation where the shoulder and hips move accordingly when it steps, whilst considering the multiple heads and how they would affect the geometry and the weight of the movements.
Look Development
As part of the delivery, the creature artist will also send of to a look development artist to help define the final textures. They are responsible for defining the visual appearance of the creature, ensuring its consistency and realism across different shots and lighting conditions (ScreenSkills 2024). This means that the creature arists has to mainting good topolgy and UVs across the model, as explained above, otherwise the texturs won’t look correct.
Figure : Experimenting with texturing the Cerberus Creature
For the Cerberus project, I took the low poly model into Substance Painter 3D and baked the details down from the high poly model. This allowed me to have all the fine details and blemishes without having to worry about a dense mesh when it comes to animation. For the texture I used a variety of different base materials, like organic flesh, plastic raw and paint white to get the different colours and textures. I made sure to disable normal input from the substance materials so that I kept the custom details that were sculpted previously. This was combined with a selection of black masks working with generators to some layered effects on the wounds and make it nice and bloody.
These textures would then be exported as separate maps and applied to the rigged model which the animator was working on. This allows them to see how the rig and weight painting deformations affect the textures and if there needs to be changes made to the UVs or weights applied.
Output
In a typical workflow, the finished animated and textured creature would be rendered with their chosen solution, such as Arnold or Redshift into different AOVs and passed onto a team of compositors. Prior to this, a matchmover would have gone in and tracked the desired sequence to re-create the set in 3D space and sent that data into maya. This allows for the creature to be rendered from the perspective of the final camera in the correct composition.
Figure : Turntable render of final creature
As seen in Figure 35, the final shot for my sequence will be an animated model that will be rendered out of Maya using Arnold since this will likely give the best quality render in a timely manner. The final composition of the shot will the rendered animation out of Maya comped into the tracked footage. To aid in this I will making sure the animation is rendered out of Maya using the OpenEXR format, containing various AOVS. This will include the diffuse (indirect and direct) lighting, coat (albedo, direct, indirect), subsurface scattering of the skin as well as the final RGBA pass. This will give me finer depth of control when working NukeX and getting the best-looking shot I possibly can if it were to be placed in a live action plate.