WZ-10 Pylon

Prop Breakdown

Richard Gysler

render2_fix-scaled
250

Richard Gysler

3D Artist

Introduction

I’m Richard Gysler, from Vancouver, Canada, and I’ve been working mainly as a 3D Modeler for over 10 years.
I also frequently do texture work, though my work throughout my career has often been more modeling-focused.

I’m most attracted to hard-surface modeling and am an enthusiast of military weapons and vehicles.

I’m currently living and working in Fukuoka, Japan.

Project

I originally modeled this prop as part of a bigger project – a full WZ-10 helicopter model. With any personal project, I tend to choose a subject that will challenge and force me to problem-solve through difficult or intricate shapes.

Doing so can improve your skill set in a way that may not be possible at a job, as at work you have to rely on your most efficient methods of performing rather than exploring new workflows.

Goals

In this case, I wanted to make a hard-surface model with rounded surfaces, as they tend to be the most difficult for hard-surface subdivision modeling, and so I chose a helicopter.

Once that was completed, I took this prop from the helicopter model and chose it to focus on improving my texturing abilities, as I feel they are lagging behind my modeling skills. My goal was for a realistic and accurate portrayal of the prop, and to make it appear as if it was in regular use and often handled.

References

Collecting detailed references was a bit of a challenge for this project. It’s a Chinese military helicopter, so due to secrecy, there isn’t an abundance of close-up photographs or information about it or its parts, including this prop.

In some cases, photographs had certain parts blurred out, which was particularly frustrating. Military enthusiast forums turned out to be the most useful.

I found a forum for Chinese military enthusiasts, and another for Pakistani military enthusiasts, which both had a thread specifically for this helicopter.

In these threads, I found plenty of images that I could not find with a simple Google search, and was able to create a large PureRef scene with photographs from various angles and distances.

On the texturing side, there weren’t any photos that showed fine or micro details, so I had to go searching for similar objects to surmise how the finer damage and dirt might look, and in what sort of patterns they would appear.

I collected many images of aircraft props that are mounted on aircraft exteriors to see how the elements damaged and dirtied them as they flew through the air. Also, I tried to identify the various metals in the prop, what they would commonly be treated or coated with, and how damage may appear on them.

I also collected images of the smaller parts, like plugs, bolts, and screws, to make sure I modeled them accurately and didn’t miss out on any details.

Modelling – Blockout

I modeled this project in Maya using subdivision modeling. It’s not really a game modeling workflow, as I’ve worked more in film (which uses subdivided models), though it could be relatively easily translated into a game model.

Essentially, it’s a high-model with proper, subdividable topology. I think subdivision modeling is a great skill to know regardless, as it’s another tool you can use and in some cases can be a quicker way to model a complex shape, which you can then smooth the un-subdivided version to convert to polygons.

wire1-scaled

Also, making subdivision models with full quads makes any other hard-surface modeling feel easy, to the point it feels like cheating when not subdivision modeling and making proper topology.

It will also give you the most beautiful and pleasurable wireframes as a bonus. And of course, if you want to work as a modeler in film it’s a necessary skill anyway.

When starting out a model, I first start simple with a blocky silhouette and focus on having everything in the right place.

So no rounding edges or beveling, unless it’s a large bevel that helps in creating the silhouette.

No small edge bevels. If you can find accurate blueprints and/or measurements of the object, it will save time in blocking out the model and allow for a more authentic representation of its shape.

I completely disregard topology this early on; trying to model a complex shape with proper subdividable topology concurrently is unnecessarily difficult.

It’s easier and more efficient to get the shape right first, and then either fix the topology or rebuild over the shape with proper topology using quad-draw (Maya’s retopology tool).

Always prioritize shape over topology, and do whatever you need to to get that shape: ngons, triangles, dirty booleans.

As long as the shape is there, it can be converted into proper topology easily, and proper topology isn’t necessary until the modeling is complete.

High-Poly & Topology

If you add details too soon, it will be much more difficult to go back and make changes to the shape, so it’s important to take time and be sure your blocking-out stage is accurate.

Once I’m confident all the major forms are in their right place, I start gradually adding more details: adding more divisions to round out curved surfaces or edges, large bevels, indentations and holes, etc.

For very complex surfaces on a part of an object, it’s often easier to isolate them and model them separately, and then reconnect them later if necessary. Again, whatever method you need to create the shape, no matter how dirty, will be fine, as you can clean it up later.

For subdivision modeling, you need to protect edges with an added edge loop on either side to prevent the smoothing algorithm from completely smoothing the shape away.

This should be saved for the last step of any piece you are modeling, as once you add these edge-loops any future changes become immediately harder and more time-consuming.

For sharper edges, you can use 1 edge protected by 2 supporting edge-loops.

1testtt

For smaller rounded corners I use 3 edges to create the rounded shape, and add 2 supporting edges on either side to lock it in place when it is smoothed.

2testtt

For larger rounded or curved edges I use 5 edges for the shape and 2 supporting edges to protect it.

3testtt

It’s better to build rounded edges with an uneven number of edges because then you can kite (reduce) them down to a single vertex.

test4

Making clean and efficient topology can be a whole book of its own so I can’t go too in-depth here, though my one suggestion would be to make sure you know what every edge-loop is doing.

For example, I often see models where a supporting edge is needed in a certain area, and to place it there the artist runs an edge loop spanning through the entire model.

This ends up being an unnecessary increase in poly count, and an unnecessary complication in the model’s topology, as the chances are you will have to deal with the edge-loop elsewhere on the model at some point.

Ideally, every edge should be accounted for; you should know why it is there and what duty it is performing on your model. If you are unsure what a specific edge-loop is contributing to the topology, it probably is unnecessary.

To avoid unnecessary edge-loop complications, in most cases, an edge-loop can be redirected and ended so that it doesn’t cause chaos throughout your entire model. If your project allows triangles, this becomes an easy task.

Many projects will demand quads only, and then this becomes a bit of a complex 3D puzzle, which I’ve come to enjoy.

You can view the topology of my model to see several areas where I’ve kited (redirected) edge loops from continuing too far and complicating the topology.

test555

For the creasing and wrinkles in the cables, I first built simple tubes that had the basic shape I wanted for them and took them into Marvelous Designer. In Marvelous, I created a simple rectangular fabric and pulled it over the tubes.

Once it was over the tubes and sewn, I gradually stretched and pulled the fabric until it covered the entire shape, and had interesting wrinkles.

I played a lot with the fabric’s settings to get this to work well. Once completed I exported a high-poly mesh, retopologized to a lower poly count in ZBrush, and baked the finer details back in on the normal map.

UVs

For UVing and Baking, there isn’t anything fancy going on here. The model spans 3 UDIMS, with the two major pieces of the model in their own UDIMs, and the bolts, cables, and clips in the 3rd UDIM.

The rest of the pieces are fitted in the first 2 UDIMS.

There are no objects with UVs spanning across multiple UDIMs. Aside from the wrinkly cables, there’s no other high-poly to low-poly baking happening in this project.

test666

Materials & Shaders

I start the shading and texturing process by look-deving base shaders and textures, finding the right combinations of color, roughness, and metallic values to try and create a realistic and clean base, maybe with some simple roughness maps included.

As I used Arnold for rendering, I was continuously going back and forth between Substance Painter and Maya, doing test renders in various light settings while comparing the results to reference images to gauge the accuracy of the materials.

This is important as there will be discrepancies between Substance and other renderers in how color, reflections, and normals will appear.

I did some research to try and find what metal the prop is made of, and settled on aluminum, as it’s often used for aircraft equipment due to its lower weight than other metals.

A common treatment for aluminum is sandblasting, and the look of sandblasted aluminum seemed to match my reference images quite well. So I used a brushed aluminum shader from textures.com, and a sandblasted aluminum shader from Megascans layered on top to create the base shader.

I tweaked the colors and roughness values a bit to try and match the reference.

substance_details1

Texturing

Once the base shaders looked accurate to the reference images in color and reflection, I started adding details to the textures.

When texturing I try to keep in mind having large, medium, and small texture details to ensure the model will look interesting from any angle or distance.

I spent most of my texturing time on making interesting roughness layers on the main metal material, as it does most of the heavy-lifting visually for this prop.

Thinking of how this prop would fly through the air, I used a different general base roughness layer for each side, and also a different layer for the insetted areas where water would gather.

The front and back sides use a more dense dirt splatter texture to simulate direct collision with dirt particles in the air. The sides use a more cloudy, streaky texture to imitate dirt particles blowing along it and gathering more sparsely in different shapes.

The inset areas on the side have a streaky, liquid texture to show how water would gather and stream down.

The top contains a more dense, streaky texture to show water trapped and streaming between the prop and the helicopter wing it is normally mounted to.

I used a mask to isolate crevasses and added a sort of fine, dirty roughness map to create surface-level dirt trapped in areas where the wind wouldn’t blow it off. These were used as a main roughness base layer on top of which further unique details would be added through dirt, stains, and damage layers.

For damage and wear, the sand-blasted aluminum layer is masked off in worn areas to reveal the brushed aluminum underneath.

For the mask, I used a workflow I found where you create several masks on empty folders or empty paint layers, and use anchor points to reference them wherever needed.

I found this workflow to be very clean and efficient.

edge_wear_mask
test888

For the wear masks, I reference damage seen on the reference images and also imagined this prop in use and where damage might form.

So it has heavier damage sparsely showing on edges and corners where collisions might occur, slight fading on rounded edges and areas that would be hit hard by wind and rain, general scratches, and fading where handling might occur.

After adding these details procedurally I added unique, hand-painted damage to increase visual interest.

test999

Creating dirt was a similar process to the wear in that I first created more general procedural dirt and then increased visual interest with unique dirt formations in select areas.

Again, this was mostly done by analyzing reference images and trying to mimic the dirt formations found in them. I made a procedural crevasse dirt layer, and created more variation in it by adding another crevasse dirt layer that was more hard and clumpy, but sparse, to show more thick dirt gathering in some areas.

I also made a dirt spray layer to show dirt gathering from direct collision with dirt particles in the air, so the dirt gathering is weighted more towards the forward-facing surfaces, though it also includes some crevassed areas.

I found this layer in particular added a lot of appealing detail to the overall dirt layer, rather than just having generic, cloudy dirt.

dirt_spray_mask

Further dirt layers included smears around the bolts that I noticed in reference images, possibly from when bolts were being added and removed.

Also, liquid streaks in the insetted areas, general dirt spots, and streaks and splatters weighted to the lower part of the model from water and dirt spray when taking off and landing.

After these dirt layers, I added a fingerprint layer, showing finger oil stains on areas where the prop would be handled.

Also, I added a grease layer. Mechanical props tend to be oiled and greased to function smoothly with less friction, and this can pool up in crevasses near mechanical parts.

test10000

The grease layer is a brownish-yellow, glossy layer, broken up by rough bits of dirt which have stuck into the grease.

Finally, I added a dust layer creating a coat of dust on the top-facing surfaces.

dust_mask

I used the same process as above for all the materials in the prop, but I’ll just focus on the main metal material since it comprises the majority of the work.

In summary, I focus on having small, medium, and large details, creating procedural details broken up by varying layers and adding unique details, and matching reference images as closely as I can.

Once all the details are added, it’s mostly tweaking and balancing all the different dirt, damage, and base layers to get a look that matches the reference images and is attractive.

At this point, I will finalize a lighting and render setup in Maya, so I can make changes and compare them directly in separate renders to see if they’ve improved the overall look or not.

A lot of this is subjective to the artist and the look they are going for.

Lighting

For lighting, I spent a long time testing out different HDRIs and light positions and setups to get a look that I liked.

I experimented using 2 HDRIs; it was something I hadn’t done before but saw another artist doing, so I tried it and liked the result. I used studio lighting HDRI to get some nice highlights and overall attractive lighting, and then also used a bank vault HDRI to get some color and more indirect lighting.

Then I added a backlight to make the edges pop with some nice rim highlights.

Rendering

I added a table and tarp I downloaded from Megascans to try and make the render look less like a typical 3D render and more like a scene or a photograph. Generally, having simple background and foreground items included in the render can make it feel more believable.

Then I used a separate HDRI to render out a background image, which I blurred slightly and compiled to the final renders in Photoshop. In Photoshop I did some simple tweaking with exposure and contrast, but nothing too intricate.

For this project, I didn’t render out separate render layers to give me total control over every single element of the render.

test111

Conclusion

I think this concludes the summary of my work. I could go into more depth though this article would get very long. I hope someone finds it useful. As I said before, I started this project with the goal of improving my texture work, so I’m flattered to be asked to write an article about it.

Feel free to contact me at the links below with any questions or work inquiries. Also, a friend and I have started a freelance partnership that we are trying to get going, so please check out our business page for MILSPEC604.

Richard Gysler.

Website

LinkedIn

Email