With the Pezzio direction and initial previs established, I wanted to test one of the biggest potential production risks as early as possible: character animation.
The film is not intended to become a character animation piece, but the current opening depends on a human and a dog.
Before developing the storyboard much further, I needed to establish whether I could control both characters efficiently enough for the animation I actually required.
Defining What the Characters Actually Need to Do
Before approaching the rigs, I reduced the performances to the minimum actions required by the storyboard.
For the human character, I currently need:
- seated pose
- standing pose
- transition from sitting to standing
- head and torso movement
- simple arm gestures
For the dog:
- seated pose
- head movement
- looking from side to side
- tail movement
- potentially subtle ear movement
- a separate standing pose if required
If I need the dog standing in another scene, it can already be standing when the shot begins rather than requiring an additional transition.
This allows the characters to support the idea without creating unnecessary animation work.
Initial Rigging Tests in Maya
My first approach was to work in Maya.
I experimented with HumanIK and began developing controls using NURBS curves.
I tested controls for:
- COG
- feet
- legs
- arms
- chest
- head
- IK
- FK
Some parts of the setup worked relatively quickly, while others created problems with constraints and how the controllers were connected to the existing skeleton.
For example, I found that some NURBS controllers constrained correctly while others appeared disconnected. This required me to investigate how the skeleton had been characterised in HumanIK and which joints were actually being controlled.
Although troubleshooting these problems improved my understanding of the rig, I was also conscious of how much production time I was beginning to spend building and repairing a system for relatively limited character animation.
Visuals
- Maya HumanIK screenshot
- Full skeleton
- NURBS controls
- Foot/COG controls
- Chest/head controls
- IK/FK tests
- Screenshot of one of the failed/disconnected controls
- Working version after fixing it
Try to show problem → solution, rather than ten nearly identical Maya screenshots.
Testing Existing Rigs
I also explored whether using characters that were already rigged could reduce the amount of setup required.
This was particularly relevant for the dog.
I found assets that already contained:
- skeletons
- skin weights
- basic deformation
Because the dog only requires a limited performance, I started questioning whether creating a completely new control rig was necessary.
A simpler existing skeleton may already provide enough control for the shots I need.
This reinforced an important production principle for the new project:
I do not need the most technically sophisticated solution. I need the simplest solution that gives me enough control for the shot.
Visuals
- Original dog model
- Dog skeleton
- Weight/skin screenshot
- Sitting pose
- Standing pose
- Head/tail joints if visible
Discussing the Workflow with Anis
While working through these rigging problems, I discussed the project and my workflow with Anis.
At this stage, I had already changed the FMP to Pezzio. The conversation was specifically about how I should approach the production problems I was now encountering, particularly character rigging and animation.
His recommendation was very clear: commit much more strongly to Blender.
One of the main reasons was not simply that Blender offered another rigging solution. It was that significantly more people around me could help with Blender.
By using a workflow that tutors and other students were more familiar with, I would be able to make much better use of feedback sessions and peer support when technical problems appeared.
This was particularly relevant after spending considerable time troubleshooting rigging problems independently.
For the remaining production period, access to support is itself an important production consideration.
The decision therefore became less about asking:
Which software can technically do this?
and more about:
Which workflow gives me the best chance of solving problems quickly enough to finish the film?
Visuals
- If you have notes from Anis → include them
- Otherwise you do NOT need to invent visual evidence here
- You can transition directly from final Maya rig screenshot → first Blender screenshot
Learning Blender Around the Needs of the Project
Cinema 4D has been my main 3D software, so moving more seriously into Blender required learning a new workflow.
However, I did not want to pause production in order to complete a general Blender course.
Instead, I began learning the tools directly around what Pezzio requires.
So far this has included:
- interface and navigation
- object and edit modes
- modelling
- cameras
- curves
- modifiers
- bevels
- rigging
- bone placement
- snapping
- parenting
- skinning
- weights
- Rigify
This approach has allowed me to immediately apply what I learn to the FMP rather than learning tools without a clear production use.
Visuals
Here you can show your learning progression:
- First Blender scene
- C4D/asset imported into Blender
- Modelling test
- Curve/modifier test if relevant
- Camera test
- Character imported into Blender
Don’t fill this with screenshots from tutorials. Prioritise your own file showing the technique applied.
First Rigify Test
One of the first systems I tested was Rigify.
I began with the metarig and repositioned the bones to match the proportions of the character.
This required me to understand Blender’s Edit Mode, bone placement and snapping much more precisely than I had needed during my initial exploration of the software.
After positioning the metarig, I generated the Rigify control rig.
The rig itself generated correctly, but I encountered another problem: the controls moved, but the character mesh did not follow them.
This initially made it look as though the generated rig had failed.
After investigating the setup, I realised that the problem was the relationship between the mesh, armature and weights rather than Rigify itself.
Working through this helped me better understand the complete hierarchy:
Metarig → Generated Rig → Mesh → Parenting → Weights
Once the mesh was correctly connected, the generated controls successfully deformed the character.
Visuals
This section should be almost a mini process strip:
1. Character mesh
→ 2. Rigify metarig
→ 3. Bones positioned
→ 4. Generated controls
→ 5. Rig moving but mesh not following
→ 6. Fixed working rig
And ideally:
short GIF/video of the working character rig
That communicates far more than six paragraphs explaining it.
Testing the Dog in Blender
I also began investigating how to use the dog’s existing skeleton rather than automatically replacing it with Rigify.
The model already contains weights, so the priority is now to determine whether the existing rig provides enough control for the limited performance required.
If it does, retaining the existing skeleton would be considerably more efficient than rebuilding the entire rig.
Again, this is a deliberate attempt to keep the production solution proportional to what appears on screen.
Visuals
- Dog in Blender
- Existing armature
- Weight relationship
- Head movement test
- Tail test if you have it
- Sitting pose
Current Position
Moving to Blender introduces a learning curve, but the decision gives me access to a workflow where I can receive significantly more support throughout production.
More importantly, the rigging tests have already helped me define how much character animation the project realistically needs.
My objective is not to create sophisticated character rigs or demonstrate advanced character animation.
The characters are one component of a wider motion design system.
The next priority is therefore to use these working rigs to begin rough character blocking, continue developing the previs and determine whether the complete film remains achievable within the revised production timeline.
At this stage, the most important measure of success is not how technically complex each individual system becomes, but whether all of the systems can work together to produce a coherent final film.