Pulkit Jain talked about the Vintage Bus Ticket Dispenser project, shared the SubDivision workflow that was used for the project, and thoroughly explained the modeling process in Maya.
Hi, my name is Pulkit Jain, and I'm a 3D Artist from Canada. I did my diploma in Digital Animation from Durham College, Ontario, Canada, and have been working as a 3D Artist since 2016. I have worked for quite a few companies, majorly in the Advertisement and Marketing industry.
In this breakdown, I will tell you my workflow for my project Vintage Bus Ticket Dispenser.
Vintage Bus Ticket Dispenser
Being in the advertisement industry and modeling game assets, I usually deal with tris topology and hence wanted to expand my horizon and polish my SubDivision modeling skills. So I talked to a few senior professionals who recommended I go through something different for my next project to better show my SubDivision skills. This particular prop was one of the few that were shown to me by them, which caught my eyes. It had all the elements that I needed for my next challenge, from a robust foundation to multiple small details, only to top it all up with a hyper-realistic vintage metal texture that I haven't been able to find before in any reference.
From my previous projects, I learned to take time in pre-production, and thus, I took my sweet time finding all the references that I could. One website that always helps me get the best result is iCollector.com, an online auction platform, the best place to find vintage props. I also happened to find new reference images later in the project while I was texturing, which gave me a better idea of how this prop should be textured.
Modeling and Sculpting
I start all my projects mainly with blockouts where I place geometrical objects in places to give me a rough idea about how I plan on using them for Midpoly. Since I strictly followed the references that do not always have an orthographic view, blockouts help a lot. However, these blockouts need not be perfect and often change later in the stage as required. On the other hand, I use the mid poly to create high and low poly models.
Through my previous project, I learned that while for game assets, high poly is usually made first and then the low poly, it isn't always the case while modeling SubDivision models. So here I created my low poly and then smoothed them out for high poly.
While the elements like rotatory dial, handle, etc., were relatively easy to work on, modeling the dispenser's body was tricky. The body consists of three major parts – the front, back, and side, and all these three parts have curves and odd shapes.
I started with the side (handle area). At this stage, I rely on as low topology as possible and stick to the shape. I kept switching to Smooth Preview to check my progress; this is one of the methods that I learned by talking to various artists. In SubDivision modeling, one needs to start with extremely low poly topology and then smoothing the mesh to get the result. I followed the same technique everywhere throughout this project.
For the back (the painted part), I started with two shapes as per my blockout and merged them. I didn't give any consideration to the rectangular holes that I decided to model later during low poly. This stage was all about shaping the foundation and keeping the geometry as smooth as possible without increasing the topology.
Next was the front, which was easy except for the edges where the topology curves. Also, I need to have holes for the screw. Even though details like these could have been baked, I decided not to, to test my modeling skills. "Everything should be modeled, and there should be no triangles"; that was my goal.
Once done with the foundation elements, the rest of the pieces were comparatively easy to model. For instance, to model the dial, I started with the top plane of a cylinder and worked on one slice of it to create a hole. After that, I used "circularize components" to get the result and then duplicated it to get others followed by "extrude."
After this stage comes cleaning. It's here that I started working on the low poly. I fix all the edges and mesh and increase topology to create screws holes and glass holes for the back. I also bevel and chamfer in this stage, keeping in mind where to have hard and smooth edges. I frequently used Smooth Preview to make these decisions. At this time, I was also considering how I will be UV mapping. This thought at the back of my mind often helps me to layout the mesh.
Before ending this stage, I made sure all the surfaces are perfect and smooth before asking for reviews from friends and seniors. Some of the topologies always need to be adjusted, which makes this process essential to have.
For high poly, the majority of elements were modeled in Maya. Only the cogs were made in ZBrush because of the boolean. I find ZBrush boolean easy and fast as compared to Maya.
I went one by one to smooth out the elements for their high-resolution counterparts in this stage. One thing that slipped my mind was to add wear and tear on the model as it's vintage, which I later realized during texturing and so decided to do at texturing phase.
The thing to note, wear and tear via texturing is not always as good as with baking.
Retopology and Unwrapping
Since I was planning UV mapping while creating lowpoly, I didn't have to go back and fix the mesh in Maya that often. I usually rely on Maya for UV mapping, but I find Rizon UV much faster, so I used that for this project.
Once I imported all the UVs, I planned which elements should be grouped under the two material sets that I decided to have for better texture resolution. I decided to go with two sets early in the project because of all the details that I was adding while modeling. I found I could have the back-painted part with all the high-resolution numbers as one set and the front portion as the second set.
Once I completed this early UV mapping and created texture sets, I imported low and high poly into Substance Painter for baking. I tend to do test baking to fix any distortion that might occur due to mesh and UVs. I take my time in this process as it's vital from a baking and texturing point of view. After I got the satisfactory baked result, I went for texturing.
Texturing details took me the longest as I was working on and off on this. Before I start texturing, I tweak Substance display settings as I find the default settings dark and uninteresting. I play with Environment Exposure, color profile (I use ACES_Standard), etc., to set up a suitable environment for my prop. Not many artists test their textures in Iray Renderer in Substance Painter, but I tend to, as it helps me visualize how my textures will look when rendered. I also play with post effects while doing these test renders every now and then.
My Substance settings:
I first started texturing the front with a base material from Substance. Then, I worked my way up to add metal edge wears, one regular and another one comparatively rougher, followed by dirt and scratches along with some rust that disguised as more prominent dirt. I stacked these details below the overall "roughness variation" to amp up the vintage look. Next, I added one final layer of another material, "Iron Rough," to add sharp scratches to the corners. At this point, I realized that I'm missing my personal touch as the material was exciting but lacked uniqueness. I remembered metal plates that I had at home back in India and tried finding some sort of the same texture. I stumbled upon one at textures.com. Though it neither was high resolution nor an alpha, it did the job for me as I used it as a Height Map to give it a different feel.
One thing that I was troubled with was also the logo, which I failed to get online. Since texturing already took a significant amount of time, I was lazy to spend time in adobe PS or AI to create the logo myself from scratch. I had to work my way around and decided to test my photo editing skills. I took the reference image that I already had and craftily rotated, moved, and cropped the logo from it and turned it into Black and white for the alpha, which I use as height and dirt texture for the logo.
The rest of the materials were standard Substance textures. I didn't spend much time on them as those elements were relatively small, simple, and straightforward.
Rendering and Lighting
All through the texturing, I kept doing test renders in Iray. It turns out, whatever post-production effects we use in Iray, also show in Substance while texturing. So this hack helps in planning final renders, which I personally feel is a big plus.
For the final renders, I decided to go with Marmoset Toolbag. As I had a clear picture of how my renders should look now, thanks to Iray, I copied the tweaks I made in Iray to Marmoset. I even imported the same studio HDRI image into Marmoset, which gave me the desired same render. I used a simple three-light setup with front, back, and fill light, which I tend to move according to the camera angle. I further play with depth of field and bloom, making the renders pretty and also showcase your skill as an overall artist.
I then took these renders to Photoshop and adjusted brightness, contrast, saturation, and levels. I also played with filters like "unsharp mask" to sharpen the details and gaussian blur to take the final composition a notch higher.
One thing to mention is while rendering, I found the scene less interesting and so decided to model a few extra pieces to fill the empty place in the scene. The creation of paper and the leather piece on the floor was done at the end and used a different third texture set. Since they were secondary elements, I didn't invest a lot of time in them. For modeling the paper, I extruded the edge from Helix and used this curve to "curve warp" (under deform) to get the shape. I created the texture for the paper using paper texture images (as Height Map) from the internet, and for the leather, I used the standard leather material from Substance Painter.
Overall, 3D Modeling is interesting and relatively easy but knowing the workflow and presenting the model is one of the most critical aspects of 3D Modeling. For this end-presentation, I advise artists to work on their baking, texturing, lighting, and rendering skills. Being passionate about modeling myself, I understand why the majority of the artists don't spend time in these areas. Still, the fact remains that the model only comes to life when all aspects of model creation, that is, texturing and composition, are done successfully. Different departments might be modeling and texturing in a studio, but having extra knowledge and skill is never a waste. Having post-production capabilities in an artist only makes them more desirable to have in a studio.
My last advice will be to keep an eye for new tutorials and hacks, talking to as many professionals as possible to know the secrets of model creation, and never run away from appreciating good work by fellow artists. The 3D industry is quite vast, and everyone has their share of challenges and tips, and hacks. Having a decent productive circle only makes existence in this industry much more fun and exciting. You happen to learn new ways to do the same stuff more effectively and with new motivation.
One tutorial that I suggest to people who wish to learn the workflow for the game prop will be the Military Radio by Simon Fuchs.
You may find these articles interesting