For our Collaborative Project with the various courses such as Games Design and 2D Animation, we formed a small group of three people. 1 from Games Design, and two from Visual Effects. Our goal for this project was to combine live action footage alongside 3D assets and environments to create a blend between the environments, My role for this project was to 3D model and texture several assets, and to assist in various tasks related to visual effects. My tasks so far were to model and texture and tube train, model and texture a bus, and to model and texture a pipe sequence with a short encounter with a mouse.
For the first part of my role, my task was to develop several 3D assets which relate to the London Underground, and Transport for London. Assets of particular note were: the London Bus, the interior of a Tube carriage, and a short camera track travelling through underground pipes, with an animated mouse model.
We obtained the necessary footage of various spots around London, and studio footage captured in front of a green screen. Originally we wanted to film inside of an Underground station, but the staff did not allow us to do so. The plan slightly changed to substitute the station with several shots from LCC and the surrounding areas. As Robin had obtained a free environment for the entirety of London, this was a good scene to insert 3D assets into. The idea was for the camera to move around the environment, and enter a drain hole to where the pipe sequence began. Also, the original plan was to have the bus cross a bridge as the camera panned over it, but this was changed to have the artist perform on top of the bus composited with green footage footage in Unreal Engine 5.
London Bus Asset
For this model, the plan was for the bus to be seen in a short panning shot over London bridge as the camera shifts through the environment. Due to the angle and distance to the model, I kept the asset fairly low-poly with large and distinct textures. The idea to have a 1980’s version of a London bus, the RouteMaster, was to spark more interest and create something that had a classic ‘London’ aesthetic.
As stated earlier, the needs for this model shifted multiple times, from a quick pan over London, to a shot over a bridge, and then to a close-up shot with the artist composited over top. To this regard, it was difficult to know exactly which shot to model and texture the asset for.
Drain Cover Asset
For this asset, Robin requested a drain cover so I gathered several references to London city manhole covers, and I attempted to find a design which fit our intentions, this being several large gaps for the camera to fly through as the camera entered the drain-pipe sequence. As the camera would be very close-up to the asset, I modelled this to be as realistic as possible, whilst also maintaining as low a polycount as I could for the sake of computing requirements. These are aspects which I always try to observer when developing any asset, and obtaining a sucessful bake for this asset was at first slightly difficult, but after fixing areas of the high density mesh, these problems were alleviated.
When texturing this asset, I wanted to maintain the worn and scuffed feeling that was persistent in the track, and also because no drain cover lacks areas of grime and rust. Whilst I spent a fair amount of time to get this asset to look realistic, I was also aware that it would only appear for a split-second as the camera moves at a very quick pace through the pipes.
I checked the model and textures in Unreal Engine 5, and after making sure all the packed maps worked correctly, I exported the model and textures to our level designer.
Modelled & Rigged Mouse
For this sculpt, I gathered several references to a typical mouse that might be found on the London Underground. I completed the high-resolution sculpt, and retopologized the model using 3DCoats remeshing tools, as well as creating teh UV’s in the process. After this, I moved to rig and animate the model in Maya by attaching joints and painting the weights. As this was a fairly simple model, and the scene was very quick, pinpoint accuracy was not essential but the animation worked as intended. Due to the reason outlined below however, the model did not get to the texturing stage.
After a detailed discussion with the song’s artist, the idea to have a mouse in the final film was rejected. The was not really that bothersome, as it reminded me that clients may not necessarily want certain elements in their videos. However, the process gave me more opportunities to test my rigging and animation capabilities even if the mouse sequence was not featured in the final film.
AI Generated Images
For this area of the project, we wanted to create alternative AI art generated sequences. We each had several frames to develop a strong visual effect which would shift frequently. To achieve each look, and maintain a consistency across each short sequence, we gathered fractal and abstract imagery and blended the art styles into our footage.