I'm not so good at box modeling, so I tried to make these models with modifiers as much as I can. My target was having the freedom to change and adjust the shapes at any time, to fit the references as better I can. But, especially in the front of the joycons, I made a kind of retopology to avoid booleans bugs, and this kind of thing, as I have shots with the camera very close to it.

I used the main models a lot, especially the joycons, so I made just one model and instanced it every time I need.

On the scene with the “bouncy ball”, I made a simple rig to control all the models. If I wanted to change the object, it's just changing the collection, and the animation on the rig was preserved. This is good too, for example, to share animations between different characters.

I did model and rig this character too, but I thought it would take the focus off the main scene.

I wanted to do all this project only using Blender. This workflow is sooo good because I don't have to render viewport previews all the time and render more frames than I need.

Blender's video sequencer has pretty much every tool that I need to accomplish this montage. I can make the cuts, change the cameras that I have in the scenes, time remapping, etc.

As I said, I liked this workflow, because I have the previz without a render, and made just the final render, with the right number of frames that I need.

I decided to make the color grading and sound design in Davinci Resolve. It has the best tools, in my opinion, for this kind of thing and has a complete solution with the free version.

I made the final touches on the editing and sound design in Davinci Resolve too. I tried to match the cuts to music as better I can.