Working on a Dower Chest in ZBrush, Blender & Substance

Introduction

Hello! My name is Amelie Riviere, I finished my studies a year ago, and I've been working at Rebellion as a foliage artist since. I have two degrees, one in Graphic Design and one in Game Art and Management. 

First, I studied graphic design for three years at ESMA in the south of France. I knew back then that I wanted to make video games, but to be honest it felt like an inaccessible industry. Nevertheless, I think studying graphic design definitely helped me to have different perspectives and references in my work. I had to study a lot of different graphic designers and illustrators that are not necessarily known in the gaming industry. It was also where I learned color theory, traditional drawing, and sculpture. Plus, it made me work with simple concepts such as composition and shape instead of diving heavily into details. I knew it wasn’t for me though, as I don’t really like the idea of working for advertising and I was looking for something more free and creative.

After three years I decided it was time to move on and I went to Rubika Supinfogame in Northern France for five years to study game art. It was definitely what I wanted to do and I’ve been obsessing with getting better at my job since. It was a way more inspiring environment and I learned so much during those 5 years.

The final year was wholly dedicated to working on a game prototype within a team of 9. I had to finish my graduation project during the first lockdown in my tiny apartment and it was really tough. This is when I started working on the dower chest because I really wanted to work on something visually rewarding and keep myself busy. 

I finished my studies one year ago and I’ve been working at Rebellion developments since. I’ve always been a botany enthusiast and I had the opportunity to specialize myself as a foliage artist. Nature is really what inspires me and I’m trying every day to reproduce it the best I can in video games. Even though I really love my job I also like to improve my skills in other subjects and I’m working on other personal projects in my free time.

The Dower Chest Project

The original chest was one of the very first pieces I put in my portfolio. It remained there at the bottom as I added more works, and every time I was reviewing my portfolio (I don’t know for you guys but I do that quite often) I felt like it was no longer on par with my skills. I didn’t feel like removing it for some reason, mainly because I kinda liked the concept. It was a 3rd-year school exercise I had to do with Photoshop and Quixel. 

The whole idea behind it was to make something that is not typical of video game chests and use the mesh, I had to be a bit more creative. I remember browsing Pinterest and coming across that reference that was the basic inspiration, but I didn’t dig into further research. I recall being quite happy with the result at the time (little did I know).

So here I was, locked down in my student flat. Although my graduation project was doing fine, I wanted to do something personal in my free time and I thought it was the perfect time to start redoing the chest. I thought of starting it multiple times before and never found the motivation to do it. 

When I started, I knew what my main goals were going to be:

• To do something realistic, but not only render-wise. I wanted to be able to show the chest’s story just by looking at it. I know I tend to over damage my props, because they can easily look good. This time I wanted something that definitely looked old and worn but still kept in good condition and not ruined.
• Stay true to the original one I made and keep the spirit with a similar color theme and patterns. I liked the original one and it was very inspirational to look at those beautiful chests when I did it. It was tempting to go for something different while I was looking at other references but I wanted to be able to do a comparison in the end. 
  

I started to look at real old chests and wooden objects to see the real damage and marks of their use. I try to add a bit of backstory when I work because it helps to know what to do, and I really pictured an object that was precious to its owner but made a few generations ago. I also looked a bit at this kind of chest history and found out they’re called dower chests used by single women to store items that’ll be useful in their future marriage. Most of those items were clothing and handmade embroidery, for instance. It was a really precious and personal object and I really wanted to show that in my work. 

The most useful references I found were on the Metropolitan Museum website, by far. It was also the perfect look for me, as they are old objects that are restored to be in good condition. I often struggle to find good references of used objects that are not visibly destroyed or damaged. They have some incredible high-resolution pictures of many things they exhibit and it was exactly what I needed. I was able to zoom very closely and see how the object reacted to time, how its story could be seen just by looking at it. For instance, where it has been repainted, some of the drawer handles have slipped and you can see their old placement, why does the paint react this way with time, how the keyhole area was more damaged because it’s an area of contact, etc. I could also see the areas that experienced the most damage, such as the edges, the bottom of the chest, or the front part of the lid. It was also when I realized that I didn’t think of many things when I made the first chest. For instance, the chest is made to be next to a wall and is, therefore, not painted on its back. The shape is also made to avoid every detail in the area to make it as flat as possible. I could go on for hours but I think you get the point.

Modeling and Sculpting

The first step was rethinking the shape of the chest based on the references I had. I wanted to have something more realistic and historically accurate. I work with Blender after having switched from 3ds Max. During this step, I also try to think of the parts that are going to overlap on the UVs, such as the sides. I didn’t overlap many parts because when you see duplication it kind of breaks the magic. On many of the references I had, I could see a mix of bare material and paint as well, which is something I wanted to add to my chest. At this point, I don’t exactly know which parts were going to be painted or not but I had an overall idea of what I was going to do after. 

I first blockout the shape in ZBrush and model small details like hinges in Blender. I then add edges to control the SubDivision and DynaMesh everything until I have a good base to start sculpting. I never merge parts that are supposed to be different wood planks or pieces, and I really pay attention to that kind of detail as I think it’s making a huge difference in the final feeling of the prop.

I sculpted all the details that I knew I was not going to obtain without a high poly (edge damage, holes, deep wood pattern, etc.). I made most of it by hand, mixed with some brushes and alpha for the wood. I spent quite a lot of time sculpting, and I think it is by far the longest part. I would not advise doing this in production (especially for a prop that is not that important). Here, it was mostly because of my personal interest and because I really wanted to translate the references in the 3D model.

My personal preference when it comes to sculpting is to work on a mesh with already quite a high resolution, starting with the Trim Border Smooth brush to damage the edges, and then switching to the Clay Build to add damage and texture to the model. In the end, I tend to add a bit of texture with alphas if needed, use the flatten brush to fade it, and find a good balance regarding the result I am aiming for. 

I really wanted to add a ton of wood detail and texture when I started sculpting, but I realized it wasn't showing much on my references. I decided to keep it subtle on some parts and to really try to stick as much as possible to reality. 

Once I was happy with the overall sculpt, I went back to Blender to make a low poly of each part. I knew I wanted to keep extra polygons for some details such as edges that make the whole silhouette interesting. I didn’t want to go too low poly and have a cube, so I kept some details in the geometry. It’s not a very exciting part and I try to keep it simple. Mostly I used simple geometry as a base that I extruded and divided depending on the high poly silhouette. In the end, I use the skin wrap modifier to make sure it sticks to the high and eventually modify the topology a bit if it gives weird results. 

For the UVs, I kept some parts overlapping, as I said, and you can see it on the Substance Painter viewport (I deleted them because it was more practical for the texturing). I try to find a good balance between what can be seen and what cannot. For instance, the sides are duplicated but not the drawers even though they are similar, I think it can be obvious they are mirrored as they are seen next to each other every time. Some legs and edge parts are duplicated as well for the same reasons. 

Overlapping UVs are highlighted in orange:

In my final UVs, I decided to save some space by having the bottom and the back of the chest in a way smaller resolution, as they are the less interesting. I use UV PackMaster, a Blender plugin for packing. I simply don’t know how to do without it, it’s very useful to pack effectively and keep overlapping parts (and it has a ton of handy options).

Texturing

Substance Painter is not the best baker in the world but it does the job and I don’t think something fancier would have made a huge difference. I did some back and forth to adjust some smoothing groups and my UVs during the baking, as I saw some mistakes I didn’t think of before. It’s also my personal favorite part as it is 100% rewarding without doing anything but waiting. 

During the texturing process, I always keep my references on another screen, as it is easy to do things you think are right but not quite real. 

Also, as you can see, I textured the hinges and metallic parts on separate folders, as they need different layers with different parameters.

First, I place my base materials, such as the painted wood and the bare wood. The material I use for painted wood is basically a solid color and Roughness with the wood Normal applied. It’s also an occasion to try out different colors. Keep in mind that I had to duplicate most of my layers to apply the right wood pattern direction on the mesh, as my UV islands are turned to gain space. I also had a scanned material of old paint on my PC that I used on top of the painted parts to add more details. The scan was very noisy and using it everywhere didn’t look right, but I blended it with the original material made in SD and it worked pretty well. 

Once I’m happy with the base, I try to work from the biggest to the smallest features. I start at a distance and move incrementally closer. This is, in my opinion, a good way to stay aware of the overall look and not lose yourself in the details. I manually painted some color variations with a new layer in multiply blending. This is a very simple layer that allows me to add low frequencies with darker areas. 

On top of that, I add a lighter layer for edge variation, where paint tends to fade. It is also a good way to add volume and silhouette to something, even though it is not really accurate. I was a bit heavy-handed with lighter edges and I think it could have been more subtle. When I texture in SP, most of the time, I use a smart mask that I’ll modify after with a paint level in the mask selection. I don’t really like using smart masks that much as it tends to look very “substance painter-y”. I tried to do the same technique for the parts where the paint comes off to let the wood appear, but it was not working. I thought painting everything would be easier, as I could apply damage depending on the sculpt where I already had those details. 

The last two layers are smart masks with dirt on cavities that I made using the same technique as for the edges. I added later some layers for leaks and time damage using alphas and other masks, it’s mostly visible on the renders of the back of the chest.

The pattern is applied through different layers in a separate folder. I made different layers depending on the color and the pattern position to be able to move them and tweak some parameters (such as the Color or the Roughness). I wanted to apply a nice bump that was not on my references but I thought would look nice on the final asset – so I just made another layer with a blurred mask and applied a different Height that is going to be added to the Normal later. I also added an extra detail on the Roughness to make the patterns pop a bit more and I made them less rough than the rest of the paint (even though it’s not very accurate either). 

Ornaments and Patterns

One of the first things I did, once I had my chest baked, was to try different things regarding the pattern. I knew I wanted to keep the same color theme as the old one, and I had a ton of references from existing chests. Certain patterns came up more often, such as flowers and vegetation in general, but also horses, birds, and characters. I am also an embroidery enthusiast, and I do my own patterns when I want to start an embroidery project. I really wanted to use those references I had and not stick to the historical chests, because it was the occasion to try something a bit different and be more creative. What I wanted to avoid was mixing existing patterns together to create something non-original, and I tried to find different sources of inspiration as much as possible. I spent quite a lot of time on Pinterest and Instagram, looking at very various types of patterns, and after a few tries, I decided to keep it only floral. 

The most difficult part was the top of the chest, as it is a large area and it was hard to find the right balance of details. In the end, I found some cushions embroidered in a nordic style and the circular pattern was really what was working the best.

At first, I thought that I could do the pattern by directly painting in SP with a layer, but the result was too messy and I was not comfortable drawing like this. I ended up opening the UV map in Photoshop and making tests there. I could then import the selection in SP and make a new layer, using the pattern made in Photoshop as a mask. It’s clearly not the cleanest workflow as I had to import different selections for each color, plus one with all the colors together – but it was the easiest for me at the moment. In Photoshop it was also easier to do some tests with duplicated parts, mirrors, etc. which is something I don’t think you can do in SP.

Rendering and Lighting

To be very honest here I made my renders nearly one year after I finished the chest. The main reason is that I forgot about it after it was finished and I was focusing on other personal projects. I would have normally made my renders in Marmoset Toolbag, but I didn’t have it on my work PC and I already made a few renders in Blender, so it was the occasion to use it once more. The Cycles rendering is quite useful and can yield very good renders with little effort.

I am not super comfortable with lighting in general, I just try to do my best with simple lights. I just wanted a simple “studio” lighting with a soft render, something quite realistic as this is what I was aiming for in the beginning. The chest has a feminine floral look and I didn’t want something overly dramatic or aggressive. Plus, many of my references had similar lighting, as they are from museums and the photography is made to display the object itself. I made a few tries to add a backlight and generate some nice Rim Light on the border of the chest – I think it’s a nice way to show the silhouette of your subject. The chest is quite squarish though, and it’s not the best prop for that kind of lighting so I kept it subtle. My main idea when moving the light was to show some details of the Paint and the Normal. 

I always render with a transparent background and a plane that is used as a shadow catcher. Once I have different renders, I switch to Photoshop to see which ones are the best in my opinion and add the final background. I like using a simple radial gradient whose value depends on what I want to render. My Blender renders are always too bland in my opinion and lack contrast, so I apply an exposure modifier and tweak the saturation a bit. Generally, I don’t use Photoshop a lot in the end, it’s really just a finishing touch tool. 

While sharing my first renders with friends for feedback, one suggested adding fabrics in the chest. I thought it was a good idea, as it completely fits the historical purpose of the chest.  At this point, I was about to finish the project and frankly was too lazy to spend hours doing fabric materials and patterns. Plus, the chest is not made to be opened, so that means that I’ll have to hide the few artifacts that are going to appear if I move the lid. 

For the fabrics, I made a box that is roughly the shape of the chest with some depth and thickness that I’ll use as an avatar in Marvelous Designer. I simply made some fabric squares that I put on the top of the avatar and let them fall, moving and folding them to have some nice folds. For the fabric material, I browsed Megascans looking for something quite sober that will not add too many details and make something visually messy. I happened to come across the embroidered fabric with floral patterns, which was more than I could ask for. I just changed the color of the embroidered leaves and flowers to have something in the same color palette as the chest. The other fabric material is simply something plain and white that doesn’t shadow the chest. 

Marvelous unwraps the clothing automatically, so it was very little effort for me to export them to Blender and apply Megascans materials. The main issue was some errors in the chest texture because it's not meant to be opened, such as the baked AO. The chest doesn’t have any thickness inside either, so I had to use a solidify modifier not to make it too obvious (you can see on the edges that the texture is stretched and doesn’t have any details). Hopefully, I didn’t need to open it a lot to make the fabrics fit, and the shadows inside the chest were eventually hiding the lack of textures.

What I Could Have Done Better

Since this whole project was about progression I think it might be worth it to point out things that I could have done better. Especially because I’m writing this now, one year after I finished it. Maybe I’ll find motivation one day to do a third one, who knows?

One of the things that I wanted to do, but didn’t investigate back then, was how to represent the painted aspect of the pattern. On my references I could see brush strokes and damage over time, depending on the thickness of the paint. I think I would need to rework the pattern on Substance Designer or rework my Normal in Substance Painter a bit more. Overall, I was just too excited to start making the pattern and finally texturing the chest after hours of sculpting to seriously consider it. 

Also, in the end, I have the feeling that the pattern is way too regular and doesn’t have this handmade feel. This is mainly because I didn’t take the time to do every part manually and I duplicated most of it, as it is quite symmetrical. I was a bit messy with this part and you can even see it on the end result. For some reason, some parts of the pattern are not well aligned, mainly because I had to replace everything manually from Photoshop and I didn’t see it. This is definitely something I would do differently today. 

I mentioned multiple times the importance of the past and story in your art, and I think I didn’t go far enough with this. I wish I had some leaks and dirt marks that could show some accident, or add some handprints in the Roughness. I mentioned one of the handles not being below its original placement and I didn’t do it in the end on the end result. I think this is mainly because I didn't dare to damage the chest by adding some marks that I would like to remove in real life.

Also, I thought about it at the beginning, but I didn’t make the interior of the chest. My decision was driven by the fact I’ll have to add this part in the final texture – resulting in less place for the exterior. Nevertheless, I think I could have removed the bottom of the chest that is completely useless and never seen, and use overlapping parts for the interior. While digging into my files, I also found great references with drawings and notes on the open lid, which is something that completely adds more depth and story. It would also have been a good opportunity to add some personal objects inside that goes with the purpose of the chest. 

Challenges

It’s hard to know how long I spent on this chest as, as I mentioned previously, I made the renders nearly one year after it was finished. I was also working during the day while I was doing the chest in the evening, so it was not like I had all the time in the world to dedicate to it. I think I spent 2 or 3 months working from time to time with some breaks in between. The most important part is not to lose motivation and keep regularity. It’s pretty hard to start personal projects after a whole day of work but in the end, it’s worth it! 

Also, I’m trying more and more not to look at digital work when I’m looking for references. There’s so much to learn from other mediums and the world around us is an endless source of inspiration. Looking at real objects that went through different eras and have a visible history is really what makes them unique. Being able to look at something and guessing a bit of what it went through is what makes a world credible and rich. 

Observation is a powerful tool that is crucial when making assets. Having good references cannot be replaced and if you neglect this part you’ll end up with – at best – a nice prop but completely forgettable. It’s also something that comes with training and – on my part – I cannot wait to see what I’ll be able to do in a few years and more experience. 

Amelie Riviere, 3D Environment Artist

Interview conducted by Theodore Nikitin