Making of 'Mules Gold'
Part 1 - Characters: By Neil Parkinson
I had designed the original cigar smoking character for a short college project. I never thought too much about the character till I was putting together an animation portfolio whilst freelancing in London, I had drawn a model sheet for it and a friend really picked up on it. It kind of snowballed from there and I kept refining the design and coming up with further characters (Fig.01).
Having worked with Lee for several years on projects at Electronic Arts UK, we decided to pool resources on a joint venture outside of work. I already had two cowboy characters built, so we came up with a Western short film idea. Knowing how long these things take we thought we would work towards small goals. The first was this teaser image.
The characters featured are Smokey Joe (right) and Rattlesnake Jack (left).
Rattlesnake Jack was very heavily influenced by Clint Eastwood in Sergio Leone's spaghetti westerns. For Smokey Joe I had Robert Mitchum in mind with the long, droopy, hangdog features. The most important element was having the characters look like they were worn weathered and lived in - in particular Rattlesnake Jack; I gave him big crow's feet next to his eyes to suggest he had been squinting at the sun in the desert, I also gave him a scar across one eye that suggest he has a history beyond the scene being depicted.
The characters already existed in some form before we started this image. They had been built mainly to be animated and so they didn't really bear up under close scrutiny in a static image. Jack had a long duster that was rigged with bones. However, this didn't pose or read well in the shot so I decided to re-model his upper torso with more detail but without the long coat tails. Joe was very basic so I decided to add more detail into him for the shot, in particular clothing folds.
When modelling I always use the Poly subdiv proxy tool in Maya. I have the subdiv setting at just 1 to keep the model from getting too heavy early on.
I always start with the head. I will model all the features and even sort out the UV mapping with the Proxy still attached. I will then use the lower poly proxy as a base to create the blend shapes. This saves me from having to move a lot of verts around and keeps the mesh quite smooth (Fig.02).
Once all the blend shapes are created on the low poly mesh I will create blend shapes from the higher poly one and once this is done I can delete history on the head and delete the low poly mesh (I always keep a backup file with the proxy still attached in case I need to create more blend shapes). At this point I add hard edges to the mesh in places such as the creases, although I'm careful not to change any verts in the mesh.
These characters were built to be animated, so I did a full set of blend shapes to enable lip synch animation. I also have bones in the face for the brows to help in reducing the amount of blends needed. Because of the size of Jack's Cigar I had to model the displaced flesh into the blend shapes (Fig.03).
When modelling the body of the characters I model the character in a T-pose for easier rigging (Fig.04).
As mentioned, I use the subdiv proxy tool. I use this to get the basic form built, as well as higher level detail, such as belts and clothing. I then delete the history and model straight onto the high poly mesh. At this point I will be adding details such as cloth folds and seams (Fig.05).
Because of the high level of detail in the models there wasn't the need to put loads of additional information in the textures. And because they were to be animated, any painted creases and shadows would stretch and look strange in the rendered frames. I kept the textures simple, only putting in a slight indication of material such as leather or linen. The rest of the look was created via the addition of specular and bump maps (Fig.06).
As mentioned earlier, the characters were modelled and rigged to look good in any pose so they could be animated. See Fig.07 for the topology of the characters in a test pose. Once we were happy with the characters we got them into a scene and started creating a world for them. What better place to have them in than a saloon...?
Part 2 - Enviornment, Lighting and Rendering: By Lee Sullivan
Before we even modelled a poker chip, the first thing we did was research every type of film, book, website and photograph that featured a western saloon of any kind. After gathering hundreds of images we came up with a prop list of the things we needed to model, as well as a layout and an overall structure for the saloon itself. As we intend to use the space as the primary location for our short film we wanted to make sure the saloon was what we required for the entire story to help minimise re-work at a later date.
Polygon Modelling: (Fig.09)
The saloon was modelled in Maya using simple polygon modelling techniques. To minimise render times we strived to keep the poly count as low as possible whilst retaining a high level of fidelity in anything that was going to be seen up close by the camera or that one of the characters would be interacting with, either in the teaser poster or at some point during the short.
Art Direction - Texturing & Shading: (Fig.10)
Before we created one texture map it was vital that we agreed on the look and feel of the saloon we were creating. While doing so much research we realised that there were many different types of establishment aimed at just as many levels of cliental. The difference between the one in say Unforgiven and Tombstone may at first glance seem relatively small, but on closer inspection details like the wall coverings, furnishings, and the type of wood the floor is made of, all contribute to the visual style, atmosphere and feel of a place.
In our saloon we wanted things to look worn, weathered and dirty. The place wasn't in the biggest of towns, wasn't very glamorous and featured only the most basic amenities. There should be dirt on the floor, the walls should be stained with cigarette smoke and everything should look worn and well used.
All of the texturing work was created in Photoshop and authored in black and white using a combination of reference base textures from various sources and photographs we took ourselves. We used Maya shaders, mostly a combination of blinns and lamberts to achieve the right mix of materials, as well as a couple of the new Mental Ray Mia materials to simulate glass and metal. In addition to the base textures we also used a small amount of bump and specular maps in places to help achieve our visual style.
Poker Table Texture - The Process: (Fig.11)
Here are the various stages we followed in the creation of any texture - in this case the main poker table. The first step was the laying out of the UVs and creating the base textures, which in this case were felt and wood. By overlaying the UVs again we were able to work into the texture creating scuffmarks and scratches on the wood, as well as adding spillages and stains to areas directly under the player's drinks and cards. All of this extra detail was created on layers in Photoshop using a variety of blend modes such as hard light and overlay.
Lighting the Environment: (Fig.12 - 13)
Early on we made the decision to light the project in Maya using Metal Ray with Final Gather. To keep render times low we wanted to avoid using global illumination, and because the lighting style was going to be fairly high contrast, and also in black and white, it just wasn't a good fit. Because the render times were still fairly high the work in progress lighting was done with final gather switched off and with a series of ambient lights with a low intensity in its place.
Because we wanted to maintain a high contrast level we started with a directional light and only added more illumination where it was required. As you can see from the images the lighting rig consisted of:
- One directional light coming through the window casting the main shadows
- An area light in each window frame to increase the over brightening
- An area light stretching out from the window onto the floor, giving added bounce light
- A low intensity point light with a linear fall off to increase illumination near each window
- A fill light with in front of the poker table to help pick it out slightly from the background
Final Environment Lighting: (Fig.14
Here is the final render for the environment and props. If you look closely you should be able to see all of the above lights' influence and how they each individually contribute to the overall look and feel of the saloon.
Environment Lighting with Posed Characters: (Fig.15)
Once we had lit and created the environment the next step was to import the characters and pose them up as we had originally intended. We wanted the poses themselves to reflect the characters' personalities and the differences between the cool laid back nature of Smokey Joe and the Quick tempered, impatience of The Rattler. As you can also see from the render, once we placed the characters in the scene we felt that they weren't really standing out as much as we wanted, and as they were the real focus of the image the lighting on them needed to reflect this.
Hero Lights - Character Only Lighting: (Fig.16)
On closer inspection of the render we felt that, whilst the figure on the left was receiving a really nice rim light from the window, our character on the right was way too flatly lit and really needed picking out from the background. In lighting, as with painting and photography, it's all about getting the audience to look where you want them to look; to have them focus on the important things and not let their attention be drawn to background elements and set dressing.
To achieve this we created a separate lighting rig to affect the characters only, which included both a new key and rim light for the character on the right. A small fill light to help illuminate the cards as well as an old fashioned Hollywood fill light on the villain's eyes to make him look more intense. We also added another global rim/fill light to them both, the table and the props that were on it.
Final Render: (Fig.17)
Below is the final render from Maya. Whilst in a sequence of frames we would divide things into render layers to make things easier and more manageable; because we were working on just one shot it was easier to have it all in camera. As you can see, the additional lights really helped bring the characters away from the background and focus the viewer's attention on the two of them and their card game.
Render Passes & Compositing: (Fig.18 - 20)
Whilst we did keep both the characters and the environment in the same render layer, we also rendered out a depth of field pass and a smoke and effects pass. These were both composited in Photoshop along with another layer, including a vignette to add that aged photograph/distressed film look. As we have mentioned before, we wanted the salon to have a real atmosphere and all of these helped greatly in achieving that style.
Final Render Composite & Colour Adjustment: (Fig. 21)
Once we had all of the layers in place the only thing remaining was a slight adjustment of the levels and a shift in the hue and saturation. Whilst from the beginning this was always imagined as a black and white piece, as we felt a shift to sepia would be too stereotypical, at the very end a slight amount of colour we thought helped really tie it together as an image.
Fig. 21 - Final Image
We hope you've found this article useful and that it has given you an insight into the process we went through to achieve the teaser poster for our short film, "Mules Gold". Please check out our blog at http://wishingwellstudios.blogspot.com/ and follow the links to both our personal websites to see other projects we have collaborated on, both personally and professionally.
Catch you down the trail!
Lee & Neil