S.H.M MK1

Interests

I have always been super interested in pursuing Sci-Fi Hard Surface artwork, and when I was recently approached with the opportunity of joining the Splash Damage Team as a Hard Surface Artist.

It felt like a no-brainer to create some Deathloop Fanart -This also gave me the opportunity to work on projects I never thought I’d be able to!

Goals & Inspiration

I’m a huge fan of stylized and outlandish FPS games, such as Wolfenstein and DOOM which I reference heavily in my own artwork (For instance my AtomFuse Artwork).

So when I recently acquired time to play Deathloop – I was amazed by the look and design that the Devs had put into the game and felt compelled to create something inspired by the incredible artwork I had seen.

In order to really stretch my skill set, I decided to create three projects with the intent of having the end result look as if it could be placed into the Deathloop Universe.

In this breakdown, I will focus on the last project of this series – S.H.M MK1.

References

Since I had already created a gun and a new version of the Hackamajig, the mech would represent an extra challenge since there are no existing mechs within Deathloop, therefore I would have to draw upon design elements that already exist in the game and add my own design knowledge, in order to create a mech that felt like it belonged in the universe.

As is standard for almost every project, I start by gathering references that I can use as inspiration for the asset.
It felt logical to first gather as much as I could from the game to serve as a solid basis for the art style & design.

Even elements that at first seem irrelevant could be helpful in later stages when trying to push the design, which is why I gathered a mix of concept art, in-game screenshots as well as portfolio pieces.

Once I had gathered this material I then began to gather references ready for when it came time to blockout the model.
I knew that I wanted the mech to be able to traverse the terrain of Blackreef and be able to effectively hunt down a player – it should be able to strike fear into those who cross its path.

With this idea in mind, I knew that making the mech bipedal would make sense, with powerful legs able to leap across large surfaces.

Many of the references I gathered during this stage were based on the mechs from the Maschinen Krieger universe, which looking back I feel made the final asset look more functional since these mechs were originally designed to be assembled by hand (Scratch Kits).

Classic cars from the 50s/60s were also heavily referenced since their sleek design and rounded smooth shapes are featured heavily throughout the Deathloop universe.

Now that I had enough references it was time to start the show and move on to the blockout!

Blockout

My first priority when making a blockout for an asset is to establish the proportions and silhouette that will carry over into the final result.

Many elements will change between the first and final step of the asset, but the overall proportions and scaling between will remain nearly the same if the blockout is created correctly.

Keeping things simple at this stage is useful since it means, changes can be made quickly and alternative ideas can be created, without time being wasted.

Once I had experimented and found a general shape and outline for the mech that I felt worked, I moved on to the 2nd stage of the blockout.

At this point, I try to flesh out some of the smaller elements that would be featured on the mech which serve to give the asset some unique qualities, e.g. Turret gun arms and springs on legs.

As stated above there are still elements that would be changed moving forward, once I begin refining details, but this is to be expected.
I generally feel like this stage is complete once I can ‘squint’ whilst looking at the design and my brain can start to fill in the details.

High Poly

This stage to me is the most exciting since I feel the most adept at this point in any project, since modeling is where I began my career!

As well as this, I think a lot of character can be added to the project at this stage, as much as texturing is used to give an asset the final touch, I feel that without a solid high poly to base textures off, the final outcome will never be as good as it could be.

For my personal projects, I try to stick to traditional sub-d modeling for as much of the project as possible for a couple of reasons.

Firstly it’s an extra challenge to tackle difficult shapes while keeping the topology of an asset intact.
Secondly, by using this workflow creating the low poly is made super easy since I can simply toggle off smoothing and clean up extra loops.

I keep things relatively loose at this stage, working on each section pretty sporadically in order to remain fresh and keep up enthusiasm; the priority is to have each section be as detailed as one another and keep the meshes free from shading errors which can cause baking mishaps later down the line.

I often take screenshots at this stage in order to be able to reflect on progress once I’m away from my PC, having regular breaks and going on short walks, I find invaluable when trying to create an asset, since it allows you to see elements of a project that might have previously been unseen.

I use VRay and procedural shaders to preview what a potential end result might look like, this is something that is very helpful since it can keep you on a solid path forward and be a symbolic ‘shining light’ for the project.

Low Poly

Once I’m satisfied with the overall look of the high poly, I start to move onto the retopology stage.

At this point, I try to be more logical in terms of how I approach the rest of the asset development.

I split all the objects into groups and rename each with numeric values and appropriate underscores, this way once the low poly is complete I can just change the suffix on all the objects and know that the baking will be mostly free from errors with few manual tweaks needed.

I go through each object and remove turbosmooth modifiers and extra support loops to create a lower-resolution mesh.
Since this was a portfolio project I wasn’t concerned with the overall tri-count, especially since I knew I wanted to have some closeups of the mech and needed the extra resolution.

For an in-game asset, certain elements like springs and bolts would be either removed or heavily optimized in accordance with how close the player would be to the object.

This process is usually the longest in the pipeline since it involves a lot more experimentation with how far certain elements can be optimized before too many details are lost.

I went by the rule of thumb that anything directly in front of the viewer’s eye line would be less optimized than meshes that were further from the viewer, so the legs would receive the most detail whilst the hatch on top of the mech for example would be optimized more thoroughly.

I also went through and deleted any objects that would be mirrored/duplicated since this would save a large amount of UV Space.

At this point I began to start visualizing how many texture sets I would need for the asset, I settled on 6 texture sets which whilst being a large amount, would give me enough resolution and texel density to work with and hold up at close distances.

UVs

In terms of UVing, I don’t reinvent the wheel.
I use the basic tools within 3DS Max in order to create UVs for each element. Since the majority of the meshes use planar angles I can flatten and break meshes by 45 degrees and then relax each UV shell in order to avoid distortion.

An important note is that I also tend to split UV shells by smoothing groups in order to avoid normal shading issues as well as UV seam borders.

Overall I try to work efficiently at this stage without rushing in order to have the best start possible once I begin texturing.

Texturing

With the UVs complete and the texture sets assigned, I export the disassembled low poly from 3DS Max and import it into Substance Painter. Initially, I keep the baking resolution for the textures set at 1K in order to be able to speed up the baking process.

These test bakes are important since it allows you to be able to identify any errors/skewing that may occur without having to wait an increased amount of time for a high-resolution bake that ends up being unusable.

Since the high poly and low poly meshes were both named identically, there were only minor tweaks needed in order to have a bake that worked and could be used in the texturing pipeline.

Once I’m happy with the test bake, I increase the resolution of the bake to 4K as well as setting the subsampling to 8×8 in order to achieve the best result possible.

I try to keep the texturing process as simple and logical as possible, by trying to replicate how materials are produced in real life.

I start with the base of each material and then build up layer by layer, for example, the main shell material is composed of a basic steel base, with a shellac paint added on top; some roughness grunge breakup and then finally a layer of rust which is shared across all texture sets, in order to unify them all.

Keeping the layer stack clean by naming each layer in a logical order, I feel is important in creating believable textures since it allows you to keep track of how each material has been built up and avoids creating unnecessary layers which can really slow down a Painter scene.

I go through and apply this process through each texture set individually, often unhiding them all and taking a step back and viewing how all the materials ‘gel’ together.

It can be easy to work in a vacuum in this process and forget the bigger picture, I often try and reference real-world materials whilst working and really question each decision, I have a small checklist of questions I ask myself during this process:

• Does this layer contribute to the design?
• Is this layer PBR compatible?
• Does the layer tell a story?
• Does it look good?

The last question seems really obvious, but I feel as an artist sometimes it’s easy to get caught up in what is technically correct but then forget that we’re creating art, and sometimes rule-bending and tweaking are necessary in order to accomplish a great piece of artwork.

Rendering

Next I move onto rendering & presenting, this has to be my favourite stage in the pipeline, since if all the other stages are done well, then presenting the final asset is made so much easier, since the hard work has already been done.

I chose to use Marmoset Toolbag 4 for the renders, since I wanted a chance to use some of the new Ray-Tracing effects that are in the latest version, it had been a while since I had used the software!

I went back into 3DS Max and assembled the final low poly mesh by duplicating and mirroring over the components that I separated prior, before exporting and importing into Toolbag; along with the textures.

I exported the textures from Substance using the Unreal export template, since it reduces the overall number of textures due to the RMAO packing algorithm.

The Marmoset scene setup is extremely simple, with just a basic HDRI used for the main lighting and two lights that act as accent lights to highlight certain angles on the mech.

Each Render angle uses a separate camera with varying focal lengths in order to showcase certain aspects of the mech, exaggerate the scale of the mech and replicate the in-game camera used in Deathloop I used a focal length between 85-100mm in order to create a wider angle look for the render, whereas the closer up images use a much shallower (35-50mm) lens to really capture as much detail in a small area.

As mentioned before I utilized Marmoset’s ray tracing ability in order to enhance the renders, the difference was night and day!
Each render is saved out with Alpha transparency so that I had flexibility later in the post to be able to tweak background colors when needed.

With everything set up, I render each camera angle and save out TGA’s, ready to throw into Photoshop for some post work.

Post-Processing

After rendering, I place all the raw renders into a Photoshop document and begin the process of post-production. I think this can be an underutilized stage in the process when creating high-quality renders, since it is easy to assume that taking images directly from the engine is the asset complete; in the case of a game-ready asset this is most often the case.

However, for a portfolio piece I feel adding some extra flare in the post is worthwhile, and in a production environment, there are often layers of post effects/VFX added once an asset moves further down the pipeline.

I added basic adjustments to each render which included some level adjustments and Colour Lookups, as well as a high pass filter to sharpen some details.
I also rendered an AO and Specular pass in order to composite in the post, which was stacked on top of the other layer adjustments.

I tried to imitate the rendering style of some other Deathloop artists – namely Robert Deleanu. This included adding in the trademark grunge style bordering as well the vignette style background for each render.

I repeated this process for all renders and saved each out, and that was the entire process complete – Around 50 hours of work all condensed down into several images!
(Why does that sound so haunting??)

Outro

I was pretty happy with the final end result for this project and was especially happy to have all three projects wrapped up and looking like they could be playable in Deathloop.

This project was a stretch in terms of designing something not only from scratch but having to incorporate an existing design style into it was another challenging feature.

I would like to give a shoutout to all the artists and devs who worked on Deathloop, it was the awe-inspiring artwork and visual fidelity that inspired me to tackle these assets.
I would also like to give a personal shoutout to James Meader who provided super helpful feedback throughout the entire process!

And a final thank you to GamesArtist for allowing me to ramble about mechs!

Thanks and keep making cool art everyone!