“Contact” – NCCA Final Major Project: Environment & Asset Modelling Breakdown

During my Final Major Project, where myself and 3 other students worked as a team to produce “Contact”, a two and a half minute short film, I was designated with multiple roles throughout the pipeline of the project. One of these was Modeller, whereby I ended up modelling the first environment to completion, the second environment I was tasked with up-resing the assets to final production level, and for the final environment (if you can call the depths of space an environment) I was responsible for modelling various high level assets within the shots.


A final frame from “Contact”, showcasing the TV Store environment

Please click on the images to see larger versions of them

Within the piece, we had 3 main environments; the outside shop window of a TV store, a NASA interrogation / interview room, and space (specifically, on a spacewalk somewhere between Earth and the Sun). I was entirely responsible for the TV Store environment, which you can see above in a still from the piece itself.

One aspect of the project I found interesting was the compromise between modelling, texturing and shading – I would plan ahead while modelling assets and consider which ones would be better suited with more time in modelling, or more time adding details in textures and normal / displacement maps, for instance the walls of the building and the pavement.


Please click on the images to see larger versions of them

As for modelling this environment, I did the obvious thing and broke it down into groups of assets and then worked my way through the lists, using creative techniques to lessen the workload where possible. The outside of the store was reasonably simple on the whole – flat walls without cutouts for the door / window as needed, with additional meshes for the poster on the left, windowsill / trim assets, guttering and door on the right hand side. The reason I did this is was because I found it would be faster for my workflow to use a tillable texture with normal maps and displacement – something I talk about in my other Final Major posts.

For the pavement, I decided to break it down into 5 components – the road, the road trim (all of the paving stones that hold the drainage grates and separate the main pavement from the road), the pavement tiles and the lamppost.

The road was a flat plain, as depth of focus and texturing would take care of the tarmac detail. For the road trim paving stones, I made a few variants of the flatter tiles and a few variants of the taller bricks and instanced them randomly along the edge of the road, adding in the drainage grate to break it up.

Please click on the images to see larger versions of them

For the lamppost section, we wanted a relief section around the base of the post to make it feel more believable and to breakup the repeating tiles. To do this, I blocked out a large tile to represent the square of tiles which would sit around the lamppost, and then used booleans and a cylinder to create a cutout of the main decorative tile in the middle. Using this as a template for the outer square of “cutout” tiles, I modelled actual tiles to match the rounded cutout in the middle, and modelled the cylindrical tile to slot into the center, later adding detail where needed.

For the actual pavement stones, I modelled an original handful of tiles, duplicated them and altered the duplicates for variation, then created a patch of tiles. This patch would then be instanced along the length of the pavement where needed, which I did to save modelling time as well as to save on the texturing load.

Shop Interior:

Similar to the structure, the interior walls and structures were low poly to be made more interesting with wallpaper textures and shaders. For certain assets, for instance the neon sign or the standing signs around the TVs, the modelling was reasonably self explanatory. With the neon sign, I again used the same instancing technique with the clasps and any duplicated letters.

Please click on the images to see larger versions of them

The two main interesting parts of the tv store were the TVs themselves, and the curtains found at the sides. With the TVs, my team and I decided that they need to be what we classed as a Level A asset – not the standard of a front and center Hero asset, but high enough that their detail could hold up as a prime focus of the shot. After researching into TV stores, no window shop has 8 of the same TV in the window, meaning I would have to make 8 different TV models for the shot.

To efficiently create the TVs, I used modular modelling tactics, whereby I modelled a handful of various components (outer cases, monitors, dials and buttons etc.) and mix / matched the parts, sometimes scaling or adjusting the meshes as needed to correspond with each other. Another note was that because we only ever saw the TVs from directly in front, I paid no attention to the modelling of the sides or back of the TVs.

Please click on the images to see larger versions of them

As for the curtains, I decided that poly-modelling them would be a losing battle if I wanted them to look realistically draped. Instead of this, I created a routine for creating curtains which gave me many random iterations of curtains based on randomising the perimeters of Maya’s NCloth, and then decided which ones I preferred for the scene. The railings at the top were modelled normally using the Modelling toolkit in Maya.

Please click on the images to see larger versions of them

This process involved creating a proxy railing plane, and 2 large subdivided planes for the curtains, hanging vertically from one end. Each ‘curtain’ was then given NCloth perimeters based on silk, and then bound to the proxy railing using a “slide on Surface” NConstraint, which meant that the curtains would drape whilst being able to have the constraints slide along each railing.

I then selected a cluster of vertexes at each end of each curtain, constrained them to a transform, and then keyframed the constraints pushing inwards on the curtain as if they were pulled in and rippled. Finally, to get the crushed base effect, during this NCloth sim, I created a rigid body plane underneath the curtains, and at the same time keyframed the floor rising underneath (in an opposite effect to the curtains being lowered onto the floor). This final effect was then baked out and used with the modelled railing meshes.

Interview Room – Up-resing Assets:

Due to the division of labour during a group project, and due to the larger than average amount of environments and assets in the piece, we decided to initially breakup modelling between me and my teammate, where I would take the TV store, and he would take the Interview room. This was initially a good idea, but later during the pipeline we realised that other areas such as rigging and and lighting needed more time being spent on them, so because of this, my teammate blocked out / modelled the interview room to Level C (on our scale of Level C, B, A and Hero), and then from there I took anything which needed more detail and increased the asset quality to a higher level.

Overall this meant going through the assets which are more noticable in shot, for instance the chairs and the microphone for instance, and adding additional details to the existing block out, or in the case of other assets, for instance the CCTV & recording cameras, taking the blockout as proportion reference and remodelling the assets from a new mesh completely.

Please click on the images to see larger versions of them

Two other assets which I had to up-res, and in this case model from scratch, were the Tape Deck and Sound Mixer. I decided to up-res these to Hero level assets, as the shot they were included in was (a) full front and center in shot, and (b) the assets were to be animated.


Please click on the images to see larger versions of them

Space – The Final Frontier:

Whilst I wouldn’t consider space as an environment per say (due to the lack of well… an environment), there were a number of assets within the space portion of the piece, all of which were created to either Level A or Hero level quality, due to pretty much all of them being front and center in shot, or amassing to a large proportion of screen space.

Out of the assets in space, I was responsible for modelling the Astronaut’s helmet (minus the cloth simulation, done by my teammate), arguably the asset which had the most time spent on it, and the Voyager probe, which will need some explaining in a moment.

Please click on the images to see larger versions of them

First onto the helmet. This asset 100% had to be of hero quality for multiple reasons;

  • The helmet was front and center in numerous shots of the piece and had a lot of on screen time.
  • The helmet had to be integrated with a CG snoopy cap, and more importantly our actor’s head, which meant in terms of detail it had to match the detail of a real human face.
  • The helmet, specifically NASA helmets (which ours was based off), are almost iconic in their design and form factor – if the helmet looked weird or wrong, it would be very noticable.

Modelling the helmet proved difficult at first to block out, but once the main form of the helmet itself was formed, I used subdividing and the ‘sculpt geometry’ tool to smoothen out the form of the helmet and the visor, as well as the modelling toolkit to model the harder forms of the helmet.


Please click on the images to see larger versions of them

One major tool which helped sped up my modelling workflow with this asset, as well as many others in the project, was a custom widget I created, which uses the Mirror Merge tool as a MEL script with an extremely low merge vertex tolerance to flip and merge a mesh on it’s far axis, meaning symmetry of a model was instant.

Please click on the images to see larger versions of them

Overall this was definitely the hardest asset to model, and took the longest out of all the individual assets, but based off of the list of reasons why I was more than happy to spend the right amount of time on the mesh to make sure it was fit for purpose.

As for the Voyager probe, this was another interesting modelling task, but for various other reasons. In the piece, our creature has destroyed the probe for an unknown reason, meaning in the piece we only ever see the wreckage of the probe. While the asset would mostly be out of focus blown-up background parts (well, all but one specific part – the golden disk), we decided that it instead of modelling hundreds of wrecked pieces of metal which might resemble the probe, that the best route was to model the probe as if it were brand new, and then have my teammate simulate it’s destruction in Houdini.

Please click on the images to see larger versions of them

2 Shots where the destroyed probe was important to the background, as well as the Gold Disk from the probe’s shot

To begin with, I researched reference of what the probe looks like. Due to it being made decades ago, and it still being in use by NASA, accurate references weren’t easily found, so I resorted to finding blueprints online, as well as various images of the probe from the 80s, and various scientific diagrams of the probe.

Once I had the concept of what to make it look like, it was just down to the process of blocking out aspects and refining the mesh where needed. Specifically with the scaffolding branches which protrude out of the main console, I modelled individual quads of tubing, and scripted it to instance each chunk to the final length. This then meant that I could use a lattice deformer to give the scaffolding a slight twist, as per the original. This also meant that when my teammate simulated it’s destruction, chunks could be glued and broken for realistic fracturing.

Please click on the images to see larger versions of them

Leave a Reply

Your email address will not be published. Required fields are marked *