What I did:

Following on from the walk cycles and lip-sync work in the previous post, animation continued for the remaining scenes.

With several scenes involving moving vehicles, wheel rotation needed to be addressed. Since none of the car models was rigged, the wheels were separated from the body of each car, and drivers were added to mathematically link wheel rotation to the car’s movement. The axis and calculation values needed to be adjusted for each model, depending on wheel size and orientation. The one exception is the Rover SD1 – the low-poly nature of that model made separating the wheels impractical, so wheel rotation was omitted for that vehicle, and shots were framed so that the wheels were not in view.

Along the way, creative decisions were made to serve the narrative – in the garage crash scene, the dashboard was given an emissive material to provide a light source that illuminates John’s face from within the scene. A subtle detail, but one that makes a difference to how dad’s emotions are read in that key moment.

With all scenes finally complete and rendered, I roughly comped everything in Adobe After Effects and exported my first draft animation. Then attention turned to the snagging process.

The Snagging List

After the first beta draft and discussions with my supervisors, I had 45 items on my snagging list. Some were minor, such as timing tweaks; others were more involved. Having a structured list was invaluable; without it, this phase could have become a mess. Trello kept it manageable, and working through it methodically meant nothing got missed.

Clouds

One of the more interesting additions at this stage was the introduction of volumetric clouds to the garage crash scene. I used an OpenVDB file from JangaFX, imported it into Blender, and added a Principled Volume material to get the right look. The clouds ended up with a slightly menacing quality, which felt appropriate. By this point in John’s story, a metaphorical storm is very much rolling in. It’s one of those happy accidents where the technical solution ended up serving the narrative.

Getting the clouds working wasn’t entirely straightforward – Blender 5.0 doesn’t have a built-in cloud preset button, so I imported a VDB file and built a material. Once the density was reduced and the material was set up correctly, the result worked well.

Slow Motion Roundabout Scene

To add weight to the near-miss moment on the roundabout, I wanted to add a slow-motion effect on that shot. Rather than re-animating at a higher frame rate, which would have been tricky because of all the separate elements involved, I used Pixel Motion frame blending in After Effects, which interpolates frames to create convincing slow motion from existing footage. For this particular shot, with its relatively straightforward motion, it works pretty well.

Keys Animation

The scene in which the keys are dropped onto the hall table posed a challenge. The keys are parented to a keyring, making a rigid-body simulation impractical. The solution was to adjust the origin point of each key and parent them to the keyring, then manually keyframe the animation.

Finchley Grange

There was feedback (from my partner and my sister) that the nursing home needed to be more clearly established as a care home rather than a hospital. The environment itself (carpet, curtains, wardrobe, personal effects) reads as residential rather than clinical, but the concern was valid.

The solution was a brief exterior shot: a freestanding wooden sign reading Finchley Grange, surrounded by planting and trees, with a building in soft focus behind it. The name is a quiet personal nod to the fact that my dad grew up in Finchley, and, at times, still believes he lives there. There are many personal details in the animation, such as dad’s room number (35), which is his birth year. Only immediate family will get the references, but they’re important to me.

The Reveal — Final Material Pass

The reveal material got a final pass at this stage. The pure chrome wasn’t reading well enough in the final sequence – it was tonally too similar to the badge before the fade out.

Before:

On my supervisor’s suggestion, I added an Emission shader, then mixed it with the existing metallic material using a Mix node, with a factor of around 0.2 and a warm white emission colour. The result is subtle; it’s not glowing, it’s just present.

The ‘IAM’ letters hadn’t been bevelled because they’re not bevelled on the actual badge, whereas JOHN had been, since it was the basis for a boolean representing embossing. As a result, the reveal letters appeared visually inconsistent. I selected the edge loops and added a bevel to bring the whole phrase into line.

To help with keying in After Effects, I added a bright green plane behind the 3D lettering in Blender to create a crisp effect. I also added a light sweep to enhance the chrome.

After:

Audio

Sourcing sound effects and voice-over clips was one of the most unexpectedly difficult parts of the production. Finding voices that don’t sound generic is genuinely hard when you’re working with free AI text-to-speech resources – most free VO clips have that robotic quality that immediately pulls you out of a scene.

ElevenLabs was really good, as mentioned previously. The level of control it gives over voice generation – tone, pace, character – meant I could generate VO clips that felt more like real people. It was also used for some sound effects.

Since the first draft of Martin’s score was delivered in late March, the animation has been cut to music, which has been invaluable for timing. However, the scene lengths changed considerably between alpha and beta following supervisor feedback, which meant Martin had to rework sections of the score to fit the revised cut – an additional ask that he handled without complaint. The second draft, however, had taken on a slightly rock quality in the final scene, which I didn’t feel was right for such an emotional and quiet ending, so I asked Martin to pull it back. The revised version is much more fitting, and I’m really happy with it.

Scene timing differed quite a lot from Alpha to Beta:

I made the decision to remove the Erasure reference. Dad did love the album “The Circus”, but it wasn’t sitting right with me. The Abba reference made it through though!

Asking Martin to supply the music track and sound effects as separate files proved useful, allowing me to make independent decisions about each element. The car engine loop in the roundabout scene wasn’t great, so I replaced it with my own SFX sourced from ElevenLabs, along with timber scraping noise, and V8 engine sounds for the Rover SD1. Dad adored that car in particular (as did I), so I wanted to use an authentic sound for that one. I also added some tyre screeching from Uppbeat.

The EXR Pipeline with the Benefit of Hindsight

This wasn’t without its moments of panic! At one point, almost the entire composition appeared to have gone black. I panicked! I thought I’d messed up a setting which had ‘rendered’ my EXR renders unusable! It turned out that I’d forgotten to apply the Colour Profile Converter effect alongside EXtractoR, which is needed for EXR files in AE. Thankfully, I had this blog to refer back to and see what had worked in my previous testing phase. Phew!

One reflection worth documenting: I rendered far more scenes as multilayer OpenEXR files than was strictly necessary. The original thinking was visual continuity – I wanted to be consistent across the entire film. In practice, the Cryptomatte isolation of the badge was only needed for a relatively small number of shots. The EXR files are significantly larger than PNG sequences, and, in hindsight, planning in advance which specific shots needed to be EXR files would have saved a considerable amount of disk space.

That said, once the pipeline was established, it was unbelievably fast and so much more efficient than animating mask paths in After Effects. Isolating the badge using Cryptomatte took a fraction of the time I’d anticipated. A reminder that time invested in solving a pipeline problem pays dividends when it comes to actually using it.

Stages of the Cyrptomatte workflow:

What I learned:

This phase taught me that the snagging process is as much a creative stage as a technical one. Small decisions – a bevel here, a material tweak there – collectively make a significant difference to the overall quality of the film. Sound design is also far more time-consuming than anticipated, and tools like ElevenLabs are genuinely game-changing for a solo producer working without access to a voice cast. Having stems supplied separately by the composer gave a level of post-production control that I’d recommend to anyone working with a commissioned score.

The EXR pipeline reflection is probably the most useful production learning of this phase – plan your pipeline requirements shot by shot before rendering, not after.

Next:

The film has been soft-launched to family, friends and peers for initial feedback ahead of formal user testing. A supervisor meeting is scheduled to review the final cut and confirm readiness to approach the expert proxy testers. The Beta presentation is also in preparation.