Charger Pistol

Prop Breakdown

Gaston Welisch

45
1

Gaston Welisch

3D Artist

Introduction

Hi everyone - I’m Gaston Welisch, a designer and artist working in the UK.

Reference

The brief I assigned myself was: a retro-futuristic hand cannon with a prototype appearance, that could have been manufactured in small quantities, but not as a commercial product. This would be a late game possibly a rare sidearm.

I wanted the gun to be exciting to use in-game, with visual interest when in the first person, and dynamic parts that would be a good challenge to animate.

I wanted to do everything from scratch, the design, modelling, textures, sounds and animations.

The design had to stand out in a unique way, yet feel like it belongs in the game where I would implement it (Fallout 4). The gun had to make sense and look like it could reasonably have been made by a group of gun manufacturers in the post-apocalypse.

Finally, I wanted to challenge myself with modelling sculpting and texturing parts out of different materials, like metals, welds, fabrics, straps, different manufacturing processes, etc.

Design & Research

I started with some very rough sketches to figure out the parts the gun would have and its silhouette (important for a recognizable design!). I ended with some sort of break action like a Webley revolver, but with the cells/cartridges behind the gun so they can be visibly spent when shot.

To ground the design I based some of the shapes on classic early  20th century handguns, mixed in with 50’s power drills, and a hint of NASA tools.

After the initial sketches, I started properly gathering research and references for the gun. Even when making a design from scratch for a sci-fi gun, it’s important to have good references, as they help you make it feel more « real ».

You can look at your references,  see how they were manufactured and used, how they aged and think about why some parts are made from certain materials. Think critically and look beyond the appearance of your ref to analyze why it interests you and why you chose it.

During the research phase, you will start to build a design language for your project. The semantics (how the shapes/colors/textures/materials make people feel and what they say about the gun/object) have to match the story you want/that you have been assigned. Who is your object for, how long has it been used, and what?

Is it industrial, heavy and rugged, or precise, antique and expensive?

Is it meant to be used by the expendable goons or the evil overlord?
You get the point.

All design elements have to match this language, as well as the animations and the sounds. There can be a mismatch but you have to do this purposefully: as an example for a gun, the noisy cricket from Men in Black works well in subverting design language expectations for comedic purposes.

I wanted the gun to feel robust like an industrial-grade tool, yet somehow high-tech enough that it felt powerful and dangerous. So for the shapes, there is a mix of curves and larger rounded parts (more industrial), with smaller, sharp details that feel more space age.

Even if the base gun is relatively well-made and uniform because the game is in a post-apocalyptic setting, I figure some wastelands would « improve » or repair their prized possessions, sometimes in weird and inventive ways, so I designed more makeshift attachments along the stock attachments.

Blockout & testing

As you can see, the design is still very different from the final gun, and this is why it’s important to iterate a lot during the design phase. I started modelling a rough blockout that I  could test in-game to see the scale and how it would look in the first person, and it felt very wrong- it took too much of the player screen and felt awkward. I had to refine the shape of the receiver and cells twice, checking in the engine, before they felt right.

Once I’m happy with the blockout for the receiver and necessary attachments it’s the start of the whole modelling journey. I hadn’t designed all the branches at that point (although  I had a few sketches). I think I could get away with it because of the slightly makeshift nature of this pistol but I was designing something that was manufactured in high numbers, I would probably try and have all the designs for the attachments ready before modelling them.

Modelling

I model mostly in Rhino (CAD software). This is what I’m most comfortable with using to model quickly, especially with more flowing or organic shapes. The only drawback is the topology, but this is where blender and Zbrush (game changer for me) come in.
Once I’m done with the « mid poly » I export it from Rhino and open it in Zbrush.

I generate polygroups from XNormals, this helps Zbrush, later on, to figure out how to retopologise the model.
Then I use Zremesher with a high value, and “detect edges” and “keep groups” on. I save this file as it will be helpful to get back to later for the low-poly (I usually Zremesh with a  lower value and delete unnecessary loops in blender).

Then I subdivide the model until the geometry is smooth (I crease the edges if I want them to stay sharp), and finally I  use Dynamesh.

From this point, I polish some edges, and apply scratches, welds, dents,  cracks, tool markings, small screws and all the other details that will be baked onto the low poly. I know Substance painter can be more flexible for this sort of detail, but especially for things like edge wear, baked details look best for me.

The part I dislike most is the low poly – Once I have a reasonably low poly ZRemeshed model in ZBrush, I open it in Blender and start deleting edge loops. I also open the high poly to make sure the lowpoly is adapted to any changes that happened when sculpting.

I mark seams and hard edges and unwrap the model. Usually, there are issues I only notice after baking the first time. I bake in Marmoset – it’s very dynamic and easy to see where things are going wrong.

Texturing

To me, modeling and sculpting are some of the most important steps in the texturing process. You may try as hard as you can, but without a good bake and model as a basis, it will be hard to create convincing materials. Consider things like edges (are they sharp,  smooth, irregular, worn, beveled…), how parts intersect and are fixed together and how thick parts are if the surface is bumpy or smooth. These things are very hard or impossible to replicate once in Substance Painter and it’s worth getting them right before texturing.

Again, our references are a guide to help us ground the textures in reality, but the goal is not to recreate them. I allow myself a lot of space for creative interpretation.

I do all the texturing in Substance painter, with some help from Megascans grunge, and  Ayi Sanchez’s (https://www.artstation.com/kratos) excellent alpha collections. Some small notes on how I set things up in painter: I set the camera to the max focal length (500mm) so I have a flat perspective, I disable all the fancy effects like flare and shadows, and I keep the resolution at 1K (even if I later export at 4K).

This is because I have a slow laptop, but the side effect is that if the textures are effective at lower resolutions without fancy effects that means they are readable and will work in most situations. Like a lot of people, I also switch the standard HDRI to Tomoco studio to have a  more neutral lighting setup.

In general, I like to combine generators with grunge and hand painting with brushes and/ or alphas. However for this project hand painting was reserved for the most visible layers  (like scratched paint or heavy dust), because there were a lot of parts that needed to have somewhat consistent materials between them. So I started by creating a few smart materials and deciding which layers need a hand-painted touch. The detailed sculpting and proper baking help the materials feel real.

Dust and grime layer

This layer is one of the most important. It’s on top of the layer stack and unifies all the materials together. Although different materials may gather dust differently, I find it makes the object feel more coherent to have this layer be common to all materials.

For dust, I use a combination of concrete material and a sandy beach material with different masks (sand in crevices, concrete on top of objects) – materials don’t need to be exactly what they will be used for, they just need to have interesting variations for what you need – you can always change the color of a material using levels and the HSL  sliders.

On top of the dust, I add an “ambient occlusion” layer which is basically dark dust in crevices, this helps the gun read better in the engine as Fallout 4 has pretty poor ambient occlusion. Finally, I add two hand-painted layers: One dark rough layer where I paint traces of suit, burns, and grime, and one shiny dark layer for oil, glue and other greasy stains.

texturing-dust-and-grime

Wood

Wood is a complex material and this one needs a bunch of hand-painted masks – In the example, the stock has layered sheets and glue in between. It’s important to start with a strong base material – for organic things I prefer something that was scanned rather than procedural but procedural gives you more flexibility.

You can have a look at  Polyhaven for free materials, or Megascans if you’re happy to pay a bit. Then I think about wear – which parts are varnished, where would it be worn off, where would hands rest on the wood often (and transfer oils, darkening the wood) – where would dust accumulate?

If wood is very old there may be several layers of varnish in areas where it’s hard to sand off the varnish. I also paint scratches using stencils – these are rough, lighter and have a negative height value – they are super important to get right because they have a big influence on the final reflections and look of the wood.

texturing-wood

Metals

Metals are something I feel I haven’t found a proper process for yet – I use entirely too many layers to get things looking right! I start with any engraving I haven’t done in  ZBrush and add linear machining marks if the metal needs it.

Then I build on top of Megascans base metal, with wear (old, new and from the making process), oxidation, dirt, oil and heat discoloration. I find I often have to readjust the base color and roughness with an adjustment layer and a levels effect so that the material still looks right.

Don’t be afraid to have metalness values that aren’t 100% or 0% metal: partly oxidized metal or metals covered in the dirt will lose some of their metalness values in places.

Render your materials and test them in the engine! Usually, I do a first pass and correct anything I’m not happy about in a second polishing pass.

Good watch: If you’re relatively new to Substance Painter, this tutorial has a lot of helpful tips: https://youtu.be/GBg3gSGTN-I

texturing-metals
texturing-oil-filter

Animations

I tried my best to make the animations as expressive as possible. There are two poses that should change the character of the weapon: a one-handed pose with a free hand on the left, which I call the “space cowboy” and a two-handed version with an angled foregrip, which I call the “tactical power tool”.

The two-handed version is closer to the screen and more centered, which makes it appear bigger, but moves more slowly and is more stable, while the one-handed version appears smaller, but feels more nimble and less stable.

Playing games like Metro Exodus, I was always amazed how adding a stock would make the weapon feel completely different because of the new animation set and  I wanted to have a go at a similar transformation.

Once I had figured out the poses and tested them in-game, it was time to animate. I used 3Ds max for the process, starting with a standard fallout 4 rig and adding helpers to control the different weapon parts.

I bought a power drill to use as a reference to see how it would weigh and move when walking, turning, running, etc… A lot of the movements were then exaggerated to make the movement feel a bit heavier because it is meant to be quite powerful.

I also watched videos of sharpshooters handling revolvers, doing spins, and reloading. Going back and forth between animating and testing in the engine was really important for me to visualize how the movements would flow together in normal gameplay and how the slight head movements would feel in the environment.

Good watch:
This is a GDC talk about the animations of Overwatch, which have a ton of characters and the talk goes into what stories animations can tell: https://youtu. be/7t0hLZd_8Z4

A quick note on sounds

I’ll be brief with this because the sound is far from my expertise, but it was a fun part of the project! I find the sound to be a really important part of shaping how a video game gun feels, and providing feedback to the player when shooting and reloading.

You get to ask yourself the same questions as in the design phase (what do the sounds evoke?) –  especially for a sci-fi gun, you get a lot of freedom to play around.

I mixed elements of real gun sounds with more vintage electronic sounds, like static, tape noises, lamps turning on and off, springs, dial-up sounds and other sample sources.

My recommendations: Aftertouch audio tutorials: https://www.youtube.com/c/ AftertouchAudio are easy to get into to get the basics right, and free sound has a ton of open source samples https://freesound.org/.

Presentation/Rendering

I decided early on to have both static renders and video renders to showcase the animations properly – I also think the moving renders give a different idea of the materials because you can see the light reflect as the objects move.

The marmoset toolbag can import FBX files pretty easily and even has motion blur. All the renders are done with raytracing enabled and HDRIs from Polyhaven (mostly Studio 2, but the video uses more natural landscapes).

A lot of people seem to like the neutral look of studio tomoco, but for renders, I find it makes metals look a bit flat.

The thing to keep in mind with raytracing is that the shadows are calculated from your geometry, and if it is too low poly some artifacts can appear. I tend to avoid HDRIs or lights with very harsh shadows for this reason – softer lights won’t cause these artifacts.

I like to use a long focal length for renders (around 80mm), to avoid distortion, although for the video I matched the camera FOV of fallout 4 which equates to roughly 26mm.

It is a bit tricky to present a project that has so many parts – taking inspiration from product photography helped and gave me inspiration for good compositions. Browse youtube for product photography videos, they could give you ideas for lighting setups as well.

To me, the static renders need to be pretty faithful to the model, they have a bit of character but just enough. With the video, I went completely bonkers though. I added grain, vignetting, chromatic aberration, haze and light leaks. In the end, it gets the vintage spirit of the design across so I’m glad I lacked restraint.

Tips and tricks

Motivation:

Get feedback often! Thankfully modding is a very active community and  I was able to share work-in-progress pictures on discord, and people gave their input and encouragement.

Not only did this mean I had feedback from the final users of the project, but it pushed me to work on it regularly. Even if you don’t have access to this kind of community, I would recommend showing your work to a patient friend or joining 3d modeling discord servers (as an example).

Be curious:

I fell down a rabbit hole with soviet spy equipment during the research phase. If your only references are other video game guns or real-life guns, it will be difficult to bring something to the table that doesn’t feel like a microwaved remix.

Researching the “state of the art” or other examples from video games helps understand what works and what doesn’t, but the special sauce is what you bring from the outside world that hasn’t been seen before (hopefully). Stay curious and try to surprise yourself with new sources of inspiration.

Keep things (relatively) clean:

I learned this the hard way, but naming things following a good naming pattern, having a well-kept folder structure, naming layers, etc. Saves you a lot of time and your sanity in the long run. This project ran for about 7 months (in my free time) and I ended up implementing a decent-ish workflow halfway in, but it would have been very useful to have started the right way. I always overestimate my ability to remember where things are or why I did things a certain way.

Have fun:

This is especially for those working on a passion project, I had to remind myself that I was doing this project for fun. Explore things you haven’t done before, learn new things and experiment!

I sometimes become a bit of a perfectionist and worry about how things will be perceived, but if I have fun making them they have value. For the one-handed animation pose, for example, I received a bit of feedback – the pose seems to be controversial, a marmite type of situation.

But I found it very fun to animate the free left hand the pose felt weirdly nostalgic to me so I kept it.

Obviously, you should listen to feedback, but in this case, I felt it would have removed some character to make the pose more generic. Listen to that voice in your head when it’s being nice to you!

Thanks to Games Artist for the opportunity to write this article – super happy to share the behind-the-scenes of this project.

Feel free to ask questions if you’re curious about points I left out.