Exploring the Blender-Unreal Engine Pipeline
This week, I investigated how to transfer the animated scenes from Blender to Unreal Engine 5. My intention was to complete all the animation in Blender and use Unreal primarily for materials, lighting and rendering.
However, I wanted to preserve the original animation, including object transformations, deformations, character rigs and cameras. I began testing FBX exports, which offered a possible solution for static geometry, skeletal animation and cameras.
However, certain animated objects did not transfer correctly, particularly when they involved more complex deformations or Blender-specific features.
I then experimented with Alembic (.abc), which allowed me to export baked geometry animation.
Although some tests were successful, I encountered problems with subdivision, scale and imported geometry.
I also investigated how to transfer animated cameras into Unreal’s Sequencer, as maintaining accurate framing was essential for the film.
As the technical challenges accumulated, I explored whether Python scripts could help automate the process, including the creation of Level Sequences.
Although I did not establish a complete automated pipeline, the experiments gave me a better understanding of the limitations of different export methods.
Here are some screenshots of what I managed to import:



Meeting with Ruby
This was what I shared with Ruby:
On 22 October, I met with Ruby, a motion designer, who shared career advice and reviewed the current animation.
She responded positively to Pezzio and gave me several useful suggestions.
For the opening, she recommended adjusting the camera easing so that it accelerated more quickly into the movement.
In the floating objects sequence, she suggested that the elements appearing after the megaphone should enter more quickly, almost as a single burst. She also recommended adding subtle continuous movement to certain objects, such as the stars.
For the city sequence, Ruby suggested giving the buildings more personality rather than simply sliding them upwards.
She also encouraged me to synchronise the building animation with the filters and allow the sequence slightly more time, as the interaction was difficult to read.
For the café sequence, she recommended making the location and its accompanying card more prominent through adjustments to scale and composition.
Finally, she suggested reconsidering the clock animation and exploring a more cohesive interaction within the calendar itself.
Testing ZXY Rigs & Refining the Building Animation
Following Ruby’s feedback, I began experimenting with the ZXY add-on to introduce squash and stretch into the buildings’ animation.
I really liked the initial results, as the deformation gave the buildings more personality and made their movements feel more playful. However, setting up the rigs came with a few challenges. I discovered that they needed to be configured using the base geometry rather than applying them directly to the complete building models.
The additional rigs also significantly slowed down viewport playback. Because I was animating while the scene was running below its intended frame rate, I didn’t realise how fast the movements actually were. It was only when I exported a playblast that I noticed the buildings were appearing far too quickly.
At the same time, I realised that I had pushed the squash and stretch too far, causing some of the buildings’ geometry to become visibly distorted.
Initial ZXY animation test: unexpectedly fast playback and excessive deformation:
To evaluate whether the extra deformation was worth keeping, I looked back to Ruby’s version and this new one. Although the version without deformation was cleaner and easier to manage, I felt it lacked some of the personality I wanted for Pezzio.
I therefore decided to keep the squash and stretch, but reduce its intensity and refine the timing. I’ll work on this next week and this will allow me to preserve the playful movement without compromising the buildings’ shapes.
Reflection & Next Steps
This week involved a combination of technical research and creative experimentation.
My Unreal Engine tests highlighted the importance of establishing a reliable production pipeline, especially when working towards a fixed deadline.
Meanwhile, Ruby’s feedback encouraged me to think more carefully about timing, secondary motion and visual hierarchy.
The ZXY experiments also taught me an important lesson about evaluating animation at its intended playback speed, rather than relying on a slow viewport preview.
Moving forward, my priority is to refine the remaining animation, improve the overall pacing and begin developing the final visual look of Pezzio.