Lights From The Alley (Student short animated film teaser)

This was the final assignment for my vocational training in 3D animation and art. The goal was to create a minute-long animation featuring (preferably) 1-2 scenes and characters, and it was meant to be completed within the 8 final months of the programme. Admitedly, I was a little overly ambitious at the beginning of this project, initially planning to include 3 scenes and 4 characters (it didn't quite work out).

For this project, pre-production began with a five-page extensive script with detailes such as camera angles and movement. A short storyboarding phase followed right after, and the actual production was next.

Due to hardware limitations and the skill level of a student, working towards a stylized look was the obvious path to take. The modeling phase went by without much issue, but dealing with the textures was the first big roadblock. Originally I was planning to render the entire animation in three layers - grouping together objects depending on how far away they were from the camera, and then create a painterly filter in post-production. This wouldn't allow for too much control without a considerable amount of time spent on fine-tuning the filter, so I opted for hand-painted textures.
But even with that plan the time budget wasn't nearly big enough. I ended up setting up a "Painterly Processor" graph in Substance 3D Designer which would take 2-4 inputs - the base color map, a baked normal map of the local coordinates of the object, and optionally up to two "detail" maps which would control which areas received smaller brush strokes on top of the big ones. This approach lead to an interesting aesthetic which was perfect for the job, and would allow me to let this processor graph handle most of the texture details.
Next came lighting which was a big challenge, considering I was using an older render engine (3DS MAX's Scanline renderer) with limited ray tracing capabilities. On the *bright* side, this was a great opportunity to do a bit of research and testing on optimizing lights and creating semi-natural effects in creative ways.
Another challenge were long render times caused primarily by the amount of light objects needed to achieve proper lighting. To combat this, I employed very primative versions of most modells with the original model completely baked onto them. This took care of most unecessary light calculations, not only because the baked models had less than 10% of the original geometry, but also because lighting was baked onto the model so it could be entirely ommited from future light calculations.
Post-production went relatively smoothly, with the final video only requiring some depth of field and motion blur to give a more lifelike look and mask a few unpolished parts of each scene.

Overall this was primarily a learning experience, and most importantly a great opportunity to experiment with most aspects of production. It's a project I plan on revisiting once time allows for it.