developingskills1and2

[developingskills1and2]

Practice 1: Developing Skills

Developing 2 (Conclusion)

Image 1 (Final dungeon diorama)

This second developing project has further deepened my love for stylization of 3D environment art and has helped me in learning industry techniques that are employed in creating stylized assets. I am now fully confident that I can use ZBrush to its maximum efficiency to create stylized assets which are game ready.

The entire assignment was really fun to work on. I first started with my research online on what type of diorama would I be working on and as I was scouring through the internet, I came across this beautiful dungeon concept by Tobias Koepp. I then set to create a low poly scene of the dungeon diorama in Maya and used ZBrush to create high poly versions of the said assets created. I then baked the normals onto the low poly meshes and textured them in a stylized format. Finally, I concluded by rebuilding by dungeon diorama in Marmoset and rendering it.

I am fully certain I can improve upon certain aspects of the scene when it comes to making the assets more stylized, texturing them properly and lighting the scene properly. I am content with what the scene turned out to be. I would really like to thank Alex Jerjomin and Harry Biggs in making me realize what can be done better to improve the scene and Neil Gallagher in aiding me with certain tips and tricks to make my scene stand out more professionally.

I do believe that the scene for me is not personally complete and portfolio ready as I would love to port this entire scene to Unreal Engine 5 as it gives me more flexibility to work on the materials and the lighting system. But as of now, I am happy with how the assignment turned out to be.

Figure 2 (Dungeon diorama by Thomas Hamilton, (Hamilton, 2024))

Figure 3 (Environment diorama by Ross Garrard, (Garrard, 2018))

REFERENCES

Hamilton, T., 2024. ArtStation. [Online] Available at: https://www.artstation.com/artwork/Xg50Ll [Accessed 09 January 2025].

Garrard, R., 2018. ArtStation. [Online] Available at: https://www.artstation.com/artwork/moqZ1 [Accessed 09 January 2025].

[developingskills1and2]

Practice 1: Developing Skills

Developing 2 (Mistakes)

During the entirety of this assignment, after much inspection and reflection, I could point out certain mistakes I did while working. Some of those mistakes were simple to fix whereas others were a bit tough to remedy given the remaining deadline duration.

Some of those mistakes were also kindly pointed when reviewed by my lecturer Neil and Wayne and also highlighted by Alex Jerjomin and Harry Biggs.

Lighting was primarily one of the weakest aspect in the initial render of the dungeon diorama. As Harry pointed out, the lighting really felt flat in the initial render which I instantly set out to improve. These 2 images below show the stark difference between the improvement of the visual quality just by improving the render.

Figure 1 (Flat lit initial render)

Figure 2 (Final lit render)

Even though I have only tweaked a bit of lighting values and deleted a lot of unnecessary lights in the scene, it immediately provides a crisp contrast and the ambience of an actual dungeon. Also, I believe if I were to recreate this scene in Unreal Engine, I would be able to perfect the lighting as Marmoset 4 has a lot of limitations when it comes to lights. You cannot have more than 3 lights for your viewer export file and the cast shadows from the directional lights do not get rendered.

Neil pointed out the flat textures of the floor tiles implying that the grunge detail of the tiles were very lackluster. I improved upon this by increasing the opacity mask of the grunge details for the floor tiles inside the Substance Painter file. Since, the textures are directly linked to the original directory, it gets updated immediately in Marmoset.

Figure 3 (Low grunge details on the floor tiles)

Figure 4 (Final grunge details on the floor tiles)

Initially, I wanted to render the scene with ray traced enabled but it introduced a lot of noise and blurriness into the scene. That decreased the stylization of the dungeon and I later rendered the diorama without ray traced enabled.

Figure 5 (Raytrace enabled that introduces noise and blurriness)

Figure 6 (Raytrace disabled that eliminated noise and blurriness)

The lower lights of the vents on the floor emitted lights of high brightness that caused the light to bleed on the pillars, doors and the walls. I fixed the light bleeding by changing the lower directional lights to a spot light, tweaking its cone angles and changing the value of its brightness so that it doesn't bleed into the rest of the assets.

Figure 7 (Light bleeding in pillars, doors and walls)

Figure 8 (Light bleeding eliminated)

I would also like to add that I have finished rectifying what mistakes I could find for this assignment. I will still work afterwards on finalizing this dungeon scene to submit it on my portfolio.

[developingskills1and2]

Practice 1: Developing Skills

Developing 2 (Reviews)

As with the first developing assignment, Alex Jerjomin and Harry Biggs immensely helped me during the entire assignment. Alex being a specialist in stylized assets himself, gave me constructive feedbacks as to how I can use certain techniques to improve the visual stylization of assets.

Figure 1 (Reviews by Alex Jerjomin)

The thing that stood out the most about his feedbacks were to go easy with the highlights for the curvature maps of the assets. That would smoothen out the stylization.

He also told me to research about how cracks occur in similar real life stone structures and take into account how heavy the asset is as that plays into how deep the cracks are made. Increasing the stylization according to him can also be done by drawing a noise mask on the assets or even adding bloom to it.

Harry critiqued my work on the textures I created and the initial lighting of the dungeon diorama. I screenshared my dungeon scene in Marmoset to him and he spent a lot of time going over all of the details in the scene.

Figure 2 (Reviews by Harry Biggs)

He made it very clear that the framing of the scene through the lens of a camera is one of the most crucial aspects of 3D for the audience. Since, my dungeon initially was framed incorrectly that included a lot of empty space, he told me to make full use of the 'Safe Limits' option in Marmoset that provides limit guidelines which can be used to properly scale a render frame for the scene.

Next, he went over my lighting as he believed the scene looked more flatter and the settings for the lights could be tweaked to achieve some great results. The one thing he said that stood out to me the most regarding lights was a properly lit scene with less lights is always better than an unproperly lit scene with more lights.

Finally, he also believed I could later in my own free time, tweak the textures of the floor tiles a bit as it looked more similar to the pillar textures and that eliminated the contrast of the scene. Nevertheless, both Alex and Harry were really impressed by my craft on low poly models and baking the normal textures correctly as it is a very crucial aspect and a founding stone of any 3D artifacts.

I also posted my renders on the discord server 'The DiNusty Empire' where the Principal 3D Environment Artist of Compulsion Games James Arkwright went over what could be done better to enhance the scene.

Figure 3 (Reviews by James Arkwright)

He noted out that since the entire shape of the scene resembles a box, he would like for me to explore the variety in shape of the walls and the floors. He also pointed out the significance of populating the scene with small props relevant to the dungeon context and tweaking the existing assets to differentiate the hierarchy for the focal point. Lastly, he mentioned the importance of thickness in the scene as the floors look very thin and increasing the length of the carpet to add more dimension to the diorama. I will be keeping this in mind when I further develop it for my portfolio.

I am really grateful for them to have overseen my assignment, provide me constructive criticism and feedback and praise where necessary. I believe that under their guidance, I have improved more as a 3D stylized environment artist.

[developingskills1and2]

Practice 1: Developing Skills

Developing 2 (Week 6)

We approach at the very last final step of the assignment which is rendering a screenshot of the stylized dungeon diorama. There are a lot of ways you can go about doing this in Marmoset and it is really convenient and flexible for the users.

You set the correct render settings on the left outliner when you click on the 'Render Settings' tab. I also tweak the post process effects of the scene by a very little as the entire scene gets affected by it. For instance, when toning down the exposure, if you try decreasing it by a great value, the entire scene becomes super dark which messes up with the lighting you set in the first place.

Figure 1 (Tweaking the post process effects)

From the render settings, I introduce a bit of bloom so that the lights emitting on top of the vents look more bright to the user. A couple of different settings like exposure, highlights and contrast is tweaked to a very small degree.

I also introduce little gradient vignette to the corners of my scene, trying to redirect the attention of the user completely to the diorama.

You can render an image by clicking on the 'Render Image' button inside the render settings. Be sure to set the correct resolution when exporting a render image to your desired location.

Figure 2 (Render settings in Marmoset)

However, there is a limitation in Marmoset 4 where you can only render out 3 lights at once and the cast shadows by directional lights are not displayed properly. That contradicts the entire lighting setup in our scene as we have more than 5 lights that each have an important role to light the assets.

Thus, I decided to render the viewport image of the diorama scene that can be accessed by clicking the 'Render' tab on the top bar and clicking on 'Viewport'. This allows us to capture the entire lights in the scene while providing us a final render.

Figure 3 (A render of the dungeon diorama)

I then set about taking different viewport renders while tweaking the lights in the scene, so that I can grab the best render among the handful renders. This also allows you to examine areas in the scene which is not being lit properly and allows for a room of inspection and tinkering about.

With this, the entire assignment now comes to an end having achieved a final viewport render of the stylized dungeon diorama scene in Marmoset.

[developingskills1and2]

Practice 1: Developing Skills

Developing 2 (Week 6)

I begin by adding lights to my scene. Marmoset provides you with 3 sets of default lights, a spot light where the light will be emitted in a spot, a directional light where the light will be emitted in the direction the light is facing and an omnidirectional light that emits light around its set radius.

Figure 1 (Adding lights to the scene)

I introduce tint to some of the lights in my scene. The vents emit an orange tint and the omni light emits a blue ambient light for the entire scene.

You should make sure to spend a lot of time in this phase trying to get the lighting correct. You don't want the lighting to be too dull or too bright as that degrades the quality of the final render. For that, it is important to set the value of the lights appropriately.

Figure 2 (Tweaking the lights #1)

I also decided to set my background to a darker shade so that the entire emphasis leads to the dungeon and no where else. You can easily do this by clicking on the 'Sky' option on your outliner. This is the sky lighting for your scene where you can decrease the background brightness, set a tint or even assign an HDRI to the scene.

Figure 3 (Tweaking the lights #2)

Here, I am just tweaking lights constantly until I get a result that I am satisfied with. At the same time, I am changing my render settings to ensure I get a good render quality. I set my samples at different resolution and also enable ray tracing and advanced light bounces so that the lights being emitted act more realistic to real life and the calculation for the bounces are calculated very precisely leaving little to no room for light bleeding.

[developingskills1and2]

Practice 1: Developing Skills

Developing 2 (Week 6)

We finally arrive at the last phase of this entire assignment, lighting and rendering the scene. For that, I import the obj file of the entire dungeon scene into Marmoset and place it on the viewport.

Figure 1 (Importing the diorama scene in Marmoset)

Since, the diorama also includes two vents on the floor, I add a couple of lighting to make sure it looks illuminated as it draws the attention of the viewer towards it.

Also, just so I can see the textures properly, I set a couple of omni lights to light up the scene. Then, I go about creating materials for each of the assets present in the scene. This is a very simple procedure to do so.

On the right hand side, if you are using the default viewport, you can see a tab called 'Materials'. From there, simply click on the plus icon and you add a new material. If you click on this new material, the properties panel below will display a couple of different options with slots where you can plug in all of you materials.

Figure 2 (Creating materials for all of the assets)

I plug in my albedo, which is a fancy way of saying the color map, normal, ambient occlusion and metallic map into its corresponding slots. To be honest, you only need to plug in your albedo, normal and ambient occlusion map since that is what we are mostly working with but I also tweaked the roughness of those assets and exported an entire set of maps along with it. That is why, just to make sure all the materials, although some being empty, work properly, I assign all the maps to its slots.

Once you create a material in Marmoset, you just simply drag the material onto the asset you want it to be assigned to. This is where combining the modular assets come in hand as instead of assigning each of the assets with the material continuously, you can assign the material to a combined modular mesh and all of the assets combined with it will share the maps and you can see the result instantaneously, saving you a lot of time which is more efficient and convenient.

I create materials for all of my assets and start assigning textures to it.

Figure 3 (Assigning materials to the assets)

With this, we complete dressing up the dungeon diorama scene and now we are ready to move onto the lighting phase. This is completely subjective and you get to be creative as to how you want to light up the scene.

Figure 4 (Final textured stylized dungeon diorama in Marmoset)

[developingskills1and2]

Practice 1: Developing Skills

Developing 2 (Week 6)

Before we proceed into our final phase of dressing up our scene, we can do it in two different methods. The first method is to individually import everything into Marmoset, rebuild the scene and assign textures to it. The second method would be to combine modular assets in Maya, export the scene as a whole and import the entire dungeon scene directly into Marmoset where you will then assign textures.

I will be using the second method as it will be more convenient for me to create materials for combined and modular assets and then export the entire scene. That way, when I assign the materials to the combined mesh, it will assign materials to all of the assets together.

This post will be very short as I will just be creating materials for the assets in Maya with only the normal maps assigned.

Figure 1 (Creating a material for the assets)

Once I set the material with its normal maps, I plug in the materials to their corresponding assets in the dungeon scene.

Figure 2 (Assigning materials for each of the assets)

This is also a good way for users to check if the normals are correctly being applied to their corresponding assets. However, Maya is known to display a glitchy viewport when it comes to displaying textures with normal maps. It may look something like this image attached below.

Figure 3 (Glitchy viewport in Maya)

Figure 4 (Combining modular assets)

The image attached shows the normals of the assets being either displayed in pure white or a shade of gray. This is completely normal and users shouldn't be scared if their normals maps are being displayed incorrectly. If you use the same maps and assign it inside Marmoset, it will display correctly.

I ran into an error where I miscalculated the document size in ZBrush while trying to export the tiles as a map after which I overrode its save file. That way, I had no way of going back and create the tiling textures for the scene. This is why I chose the alternative of using the actual tile objects, decimating their resolution in ZBrush that allowed me to export a very low sized high poly model and use those models to get the baked maps.

I finish assigning all the materials correctly to their assets. After this, I set out to combine all the modular assets and the pieces that will end up using the same texture maps.

Figure 4 (Final dungeon scene)

This officially concludes the last step before moving onto the final phase of dressing up the scene completely in Marmoset and lighting it up. The entire scene is selected and exported so that it can be imported as an obj file in Marmoset.

Figure 5 (Exporting the final dungeon scene)

[developingskills1and2]

Practice 1: Developing Skills

Developing 2 (Week 5)

We are almost done with texturing most of the assets of the dungeon diorama. I am left with texturing the door structure, the only piece remaining. Same as before, I import the door structure into a new Painter file and set about texturing it.

Figure 1 (Texturing the door #1)

Sometimes, when you want to texture specific faces or polygons, you can press the fill option on the left toolbar or press 4 on your keyboard to access the fill option. This allows you to cycle through options you can use to fill; it could be faces or polygons or a brush. Some of the cracks in the door structure were harder to reach, so I used the polygon fill and selected the face option to fill in the base color for the faces of the stone structures.

Figure 2 (Texturing the door #2)

I start painting light shadows in the cracks I sculpted with the orb crack brush in ZBrush. I also darken the curvature of the cracks with the filter 'Levels' that allows me to control the values of the black and white channel of the curvature maps. I use both of them in combination to add a painted shadow effect to the stone structures.

Figure 3 (Texturing the door #3)

Since, I didn't want the door structure to be the same color, I used black masks on different color variations and also used the polygon fill option to fill different polygons with different color variations.

Figure 4 (Texturing the door #4)

Figure 5 (Texturing the door #5)

With this, I finish sculpting all of the assets needed to finally dress up our scene in Marmoset where we can add lighting and render out scene screenshots.

[developingskills1and2]

Practice 1: Developing Skills

Developing 2 (Week 5)

After the completion of the texturing for the wooden tiles, I move onto the remaining assets for our scene. These blog posts won't carry the entire detail for the remaining assets as they are exactly the same process but do contain the progress for the remaining assets.

Figure 1 (Texturing the wooden beam #1)

I import the wooden beam into a new Substance Painter file and start texturing it. Before that, I plug in the normal, ambient occlusion and the curvature maps of the mesh before proceeding any further.

Figure 2 (Texturing the wooden beam #2)

Here, I have created all of the base colors along with its grunge variations. Now, I create a separate layer to paint occlusion, shadows and edge highlights to the asset. I then use the slider under the 'Normal' tab to control its opacity.

Figure 3 (Texturing the wooden beam #3)

After adding a small tinge of dark and light gradient to the beams, I was also finishing with its texturing.

Figure 4 (Texturing the pillar #1)

As the ornamental and the base pieces of the pillar are separate, it is very easy for me to texture those pieces individually. I plug in the corresponding maps, add the base color and its variation and add filters like blur and levels to adjust the color texture.

However, for the pillar, I decide to add emphasis to the edge highlights and metal shine for the bolts sculpted. I use a blank layer and paint those details onto the asset so that it is painted rather than the objects itself holding the properties I want to be displayed.

Figure 5 (Painting the metal shine)

Figure 6 (Texturing the pillar #2)

Figure 7 (Texturing the pillar #3)

After I finish painting the textures for the pillar base and its ornamental pieces, I create a new Marmoset file and import the low poly model and assign it those exported textures just to see if it works perfectly without any errors.

Figure 8 (Testing the assets in Marmoset)

After ensuring that the model works perfectly with the textures we created in Marmoset, we go back to finishing texturing the remaining assets.

Figure 9 (Texturing the barrel #1)

Since, the cap of the barrel already has data sculpted into it, it wouldn't require any additional height texturing for its lines.

Figure 10 (Texturing the barrel #2)

Figure 11 (Texturing the barrel #3)

While texturing and adding grunge maps, I make sure to not overdo it so as to overshadow the base color of the barrel rings. That would make the texture very noisy and break the stylization as we want the assets to look painted, not sprinkled.

[developingskills1and2]

Practice 1: Developing Skills

Developing 2 (Week 5)

After the completion of sculpting the high poly models and baking the mesh maps, we move on to the texturing phase. I start by importing low poly models into Substance Painter.

Figure 1 (Importing low poly tiles into Substance Painter)

I also plug the already baked normal and the ambient occlusion maps in the texture set settings. Now, we are ready to start painting the tile.

First of all, we start with a base color of a wooden texture. Keep in mind that the choices of color is fully up to the creator but it should be kept within natural standards.

Figure 2 (Painting the tiles #1)

I then add a couple of variations of the grunge colors onto the base color of the tiles. You can easily do this by adding a filter and choosing a grunge map of your liking. I then top it off by using 'Slope Blur' as a blur filter on top of the grunge map.

You can think of Substance Painter as the painterly Adobe Photoshop. So, the hierarchy of layers matter here and the operations will be carried out accordingly.

Figure 2 (Painting the tiles #2)

Figure 3 (Painting the tiles #3)

I forgot to add this to a couple of earlier posts but, it would be more helpful if you also bake curvature maps of the meshes. That way, you can manipulate and tweak the edge highlights if you want. Of course, you can always paint it too but I like to use that as bit of flexibility where if the edge highlights of a mesh is, say too bright, I can just tone it down. I personally use a blend of both but it is completely up to the user's preference.

Figure 4 (Painting the tiles #4)

In the same way we tweaked the edge highlights, I also use the curvature map to tweak the ambient occlusion. Here, I use a black base color layer and add a black mask to it. This allows me to paint directly on top of the other stacked layers. I can then tweak the properties of the black mask to make it more dark or light or even change its base color.

Figure 5 (Painting the tiles #5)

Finally, I add a bit of gradient of both light and dark with the help of sphere projection. It allows me to control the hardness of the gradient as also position the gradient box as desired.

Figure 6 (Final tiles painted)

With this, we are done sculpting our stylized wooden tiles. I will be using the same procedure to texture the rest of the assets for our scene.

[developingskills1and2]

Practice 1: Developing Skills

Developing 2 (Week 4)

After baking the main pillar mesh, I set out to bake the remaining maps for the remaining objects in the diorama scene. You can normally bake the maps and start painting on it, but I decided to go one step further and bake those maps and save it to a different directory so that I can use those and paint later on.

For certain high poly models, since they have a very high poly data, I use the decimate plugin to reduce their points and reduce their size so that computer doesn't overheat and the software don't crash completely.

Figure 1 (Baking the door maps)

You can tweak the max frontal distance to get better bakes. Also, you could increase the secondary rays to get a better occlusion lighting for the baked maps.

Figure 2 (Baking the arrow piece)

Figure 3 (Baking the barrel piece)

Figure 4 (Baking the wooden beams)

Figure 5 (Baking the tiles)

After baking all of the maps, we are done with the sculpting phase. We are closer to completion and now we move to the texturing phase.

[developingskills1and2]

Practice 1: Developing Skills

Developing 2 (Week 4)

Now, we arrive at the sculpting phase of the tiles. This is going to be a bit tricky as we are supposed to create tile sculpts for all 3 of them. The tiles are imported into a new ZBrush scene.

Figure 1 (Low poly tiles)

After this, the basic technique of splitting all tiles, adding dynamesh and merging them and assigning new polygroups is carried out. This allows us to free sculpt on any of those individual tiles at will while isolating the specified one.

Figure 2 (Tiles ready for sculpting)

At this point, you should probably isolate the tile you want to sculpt so that your sculpt doesn't bleed onto the other tiles. I use the trim dynamic brush to trim the edges of those tiles while hiding the plane behind the tiles.

Figure 3 (Trimming the edges)

After finishing trimming the edges, I use the orb crack brush to introduce the cracks. Again, I make it so that not every tile has cracks, leaving some intact.

Figure 4 (Cracked tiles)

After concluding with the floor tiles, I continue doing the same with the wall tiles.

Figure 5 (Wall cracked tiles)

Figure 6 (Wall wooden tiles)

After this, I start exporting the high poly models into Maya scene where I try bake normals onto the low poly models. You can use a variety of methods and techniques to bake normal and ambient occlusion. I will be showing you two different methods you can use to bake normals.

The first way is to use xNormals, a baking software. The second way is to import both, your low poly and high poly models onto Substance Painter and bake the normal and the occlusion inside it.

First, I tried baking the normal and occlusion maps inside xNormals. I export the high poly models into a new Maya scene with the low poly model.

Figure 7 (Ready for export)

Since, I don't want the designs in the ornamental pieces of the pillar to bake onto the base of the pillar, I separate those two pieces differently while merging everything else together.

Figure 8 (Low poly model preparation)

To bake the maps properly, you need to make sure that the low poly model has its normal softened with the UV edges inside the editor hardened. This makes it sure that the model bakes properly. But, I would like to argue this is a bit older technique but it does help you during the baking process.

You export the hardened normal mesh and then work on creating a cage file. This cage file makes sure that the normals don't bleed onto the other UV maps. This is very easy to accomplish by creating a duplicate of the said low poly mesh and scaling the normals of the vertices a bit bigger.

Figure 9 (Scaling the normals)

The normal scaled mesh would act as the cage mesh. After exporting the high poly, low poly with the normals softened and the cage mesh, we plug it into xNormals. The UI is fairly simple and easy to understand even for beginners. You plug each of the mesh into its corresponding slot.

Figure 10 (High poly mesh import)

Figure 11 (Low poly mesh import)

To add the low poly cage mesh, you right click on the low poly mesh and click on 'Browse external cage file'. This opens a directory box where you can locate your cage file and add it in.

Figure 12 (Setting the maps)

You should also spend a bit of time tinkering with the settings. I set my maps to a 2K resolution, however that comes with a cost of size. I set the edge padding to 4, bucket size to 32 and the anti-aliasing to 4x. Then, you should click on the maps you want to bake. In our case, it is the normal and the ambient occlusion maps. After that, you would want to click on 'Generate maps'.

Figure 12 (Normal map generation)

Figure 13 (Ambient occlusion map generation)

If you see the normal map attached above, you can see that certain maps have dark purple and a bright cyan color in it. This is a very bad baked map. It means somewhere during the preparation phase, you have committed an error leading to bad baking.

I didn't figure it at the time for the cause of bad baking but after some snooping around, I found out that since the high poly mesh, the optimized low poly mesh and the cage mesh weren't sharing the same placement in the Maya scene during export, this caused the maps to bake incorrectly.

However, this was just an introductory lesson into baking with xNormals. I normally bake maps with Substance Painter and that's what we will do here on out.

I imported the low poly mesh into Substance Painter. If you click on the 'Mesh' tab, you can bake your maps inside Painter. From there, you click on the document icon besides the empty 'High Definition Meshes' option and add your high poly mesh.

Figure 14 (Baking your mesh maps)

Figure 15 (Baked maps onto the low poly mesh)

After this, it's just a simple matter of baking in Painter and exporting files either in a png or a targa extension.

[developingskills1and2]

Practice 1: Developing Skills

Developing 2 (Week 4)

After sculpting 2 of the main pieces of the diorama scene, I set out on sculpting the remaining pieces which are really similar so I won't comment much on it. However, I believe the barrel is a bit complex so I might comment on it a little bit.

As default, I import the low poly obj export of barrel into ZBrush. Keep in mind that I split all of the parts beforehand and delete all but one wood plank since they will share the same UV resources.

Figure 1 (Importing the low poly barrel)

Using the same techniques, I split the mesh into parts and add dynamesh with a resolution of 600 and merge them together. After that, I assign autogroups to each of them.

Now we are ready to work on the sculpting of the barrel.

Figure 2 (Sculpting the wooden plank)

Using a combination of brushes like orb crack, trim dynamic, smooth brushes and so on, I create patterns, cracks and engravings on the plank to make it appear stylized. The stylization on wood plank is something one must research beforehand as they seem to follow a certain set of rules. You must never go too overboard with it as it is easy to break the immersion aspect of the stylization. With that being said, proper research is necessary.

Figure 3 (Creating engraving on the wooden cap)

The wooden caps of barrels normally have straight chipped dips and at first, I wanted to do that in Maya itself during the modeling phase. However, I decided to sculpt it as it would be more convenient for me to do everything all at once rather do one at a time and then end up forgetting it later on.

Figure 4 (Final sculpt of the low poly barrel)

People might be confused when I say the final sculpt of the low poly model, it is just to make it easy to understand. When you sculpt a low poly, it automatically converts to a high poly model so I believe the right terminology should be the final high poly sculpt of the model.

I continue my sculpt for the stone blocks and the wooden beams for the dungeon scene.

Figure 5 (Sculpting the stone block)

Figure 6 (Sculpting cracks in the blocks)

Very important as a creator when you are sculpting stylized assets is to make sure that the small details like the size of brushes and its intensity value is appropriately sized relative to the assets. Since, the stone blocks are really small compared to the dungeon diorama, I use the relative appropriate sizes of the brush when sculpting the asset.

Figure 7 (Sculpting wooden beams)

The wooden beams follow the same principles. The main difference between sculpting a stone asset from a wooden asset is probably toning down the value of smooth trim borders in the wooden asset as it looks more organic and natural. However, don't hesitate to add in your own creativity.

[developingskills1and2]

Practice 1: Developing Skills

Developing 2 (Week 3)

After finishing with the sculpt of the pillar, I import the low poly obj export file of the door and import it into ZBrush. Since, I have deleted some of the faces as the duplicates would share the same UV space when I bring it back into Maya, I have an incomplete door to sculpt but it is totally fine. It is more efficient and saves a lot of time sculpting the door.

Figure 1 (Importing the low poly door)

Same as for the pillar, I split the door into parts, add dynamesh with a resolution of 600 and merge them together and then assign different polygroups to each of the pieces.

Figure 2 (Different polygroups of the dynamesh door structure)

After the individual polygroups have been established, I just the trim dynamic brush and the trim smooth border brush to sculpt the edges of the stone structures.

Figure 3 (Sculpting stone structures)

Figure 4 (Sculpting wooden plank doors)

After I conclude sculpting the stone blocks, I move on to the planks. No need to fret as it is the same procedure where we trim the borders of the planks.

I then use the smooth the trimmed edges of the stone blocks of the door so that it looks more polished. To be completely honest, it's artistic choice as to how much stylization you want in an asset, but since I was working with low poly models, I wanted them to look stylized which retaining a polished visual.

Figure 5 (Polishing the stone blocks)

With the help of the orb crack brush, I start isolating those stone pieces and introduce cracks and chippings onto it. This adds wear and tear to the stylized stone blocks.

Figure 6 (Orb crack brush)

Since, we are dealing with wooden planks as doors, you don't need to add much edge polish as wooden structures and textures rot away naturally with edges jagged and imperfect.

Imperfection is the virtual perfection, but sometimes people take this too seriously and go bonkers when adding creative details on assets. This in turn breaks the balance of visual and quality of the asset. I use the same orb crack brush and add engravings onto the wood.

Figure 7 (Wood chippings)

After sculpting all of the stone structures, the arrows, bolts, the wooden doors and the doorknob, we complete our stylized sculpt. Don't worry about the missing pieces as we will duplicate and reattach them to its original missing place.

However, since we will be rotating the pieces by a complete 180, make sure that the detail varies in the stone structures so as to make it more organic.

Figure 8 (Final sculpt of the door)

After finishing the sculpt of the door, I again open the pillar document as I wanted to add designs in the upper and lower ornamental pieces. I decided on adding a skull and a group of 3 circles on the pieces respectively.

To create a skull on the upper ornamental piece, I use the mask tool and draw a freehand skull. Since it is a mask, it would be very easy for us to inflate the mask and get a skull sculpt after which we can use the smooth brush to smooth out the jagged details.

Figure 9 (Creating a skull mask)

Figure 10 (Inflating the mask)

I use the same process and create a trio circle design on the lower ornamental piece.

Figure 11 (Low ornamental piece)

For the remaining slab piece, I add a couple of metallic bolt sculpts and then I am done with the entire pillar piece as well.

Figure 12 (Final pillar sculpt)

[developingskills1and2]

Practice 1: Developing Skills

Developing 2 (Week 3)

I first decided to work on the sculpting for the pillar model, seeing there are many individual pieces I needed to work on. I first import the obj export file of the pillar into ZBrush. One thing to note is that before exporting the pillar, make sure to separate all the pieces in your Maya scene so that you can isolate those pieces in ZBrush.

Figure 1 (Low poly pillar in ZBrush)

In the 'Subtool' tab, you can click the 'Split into Parts' option that allows us to split the pillar into small individual pieces. This only works if the model was beforehand separated before importing it into ZBrush.

Before sculpting those pieces, we need to make sure that we have enough resolution or data for those pieces to work with. Since, we are working with low poly models, without the help of enough data in those models, the sculpt doesn't work.

There are a lot of ways you can go about increasing the resolution. What I normally do is enable the 'Dynamesh' option with a resolution of 600. This increases the resolution of, say the upper slab of the pillar, and we can start sculpting it.

Figure 2 (Introducing dynamesh to the upper slab)

After that, I then enable the dynamesh for each split parts with the same resolution value and then I merge all of those pieces together. You really don't need to merge it but that is normally how I have always sculpted pieces in ZBrush.

Within the 'Zplugin' tab, I open the 'SubTool Master' and click on 'Merge'. While merging it, I make sure that I click on 'Preserve existing Polygroups' and 'Merge and delete extra SubTools'. This helps preserve the existing polygroups and merges it without any duplicates to eliminate further confusion.

Figure 3 (Merging the dynameshed parts)

We are now left with a single polygroup pillar with a lot of resolution to work with. Since, it is more convenient to work with multiple polygroups so that you can isolate it, we enable the 'Auto Groups' option in the 'Polygroups' tab.

Figure 4 (Auto Groups for the pillar)

Before sculpting it, if we have less vertices to work with, you can always increase its subdivision level. This introduces more data and resolution you can use to sculpt on but you should be very careful of the total points of the model. If it exceeds millions, then you could run into software lag or crashes and that could ruin your progress.

Now, I first isolate the upper slab of the pillar and use the default trim dynamic brush to trim the edges of the slabs. This gives us a more stylized look and I trim all of the edges of the slabs. This is a very common approach whenever you want to sculpt a stylized asset.

Figure 5 (Trim Dynamic Brush)

You should be careful not to increase the intensity or the scale of the brush too much as you would need to undo and redo the entire thing from scratch. There is a trick you can employ which is enabling the 'StoreMT' inside the 'Morph Target' option. This stores the previous states of the assets and if you end up sculpting incorrectly, you can enable the Morph brush and it reverts the sculpted piece into its original state.

Using the same technique, I work on the upper ornamental piece of the pillar with a trim dynamic brush.

Figure 6 (Sculpting the upper ornamental of the pillar)

Since, I wanted to add bolts to one of the slabs, I mask the inner faces and deflate it so that I can add data. I then use a smooth brush and smoothen out the rough deflated edges.

Figure 7 (Deflating the masked inner faces)

I then continue sculpting the remaining pieces of the pillar using the same techniques as before, trimming the edges and smoothing out the jagged random data.

Figure 8 (Continuing the sculpting process)

Once I finish with the first sculpting pass, I move on to the second pass. This is where I normally use the trim smooth border brush to smoothen out the trimmed edges. It makes the assets more polished not grungy.

Figure 9 (Trim smooth border brush)

After I finish smoothing out the trimmed edges of all of the split parts of the pillar, I move on to the third sculpting pass which normally includes adding details like cracks or split edges and so. There is a free stylized pack called 'Orb Brushes' and I use some of those brushes for it.

Figure 10 (Orb crack brush)

While creating cracks with the orb crack brush, be mindful of the size and the intensity of the brush as too much intensity could ruin the stylization of the asset. Lastly, I use the orb smooth edge brush to smooth out the cracks I created and I am done with the sculpted pillar.

[developingskills1and2]

Practice 1: Developing Skills

Developing 2 (Week 3)

I realized that we still had 3 more assets to build for our scene. That would be the floor tiles and 2 different variations of the wall tiles that would populate the scene. This provides rich flavor to the scene.

I create a new scene in Maya where I start working on the floor tiles. I decided to create tiles based on planes that I could then duplicate easily on the floor. This is a bit tricky as you need to make sure that it tiles properly on both the horizontal and the vertical axes.

Figure 1 (Creating different tile blocks)

I start placing those blocks in a creative arrangement in the plane. The only thing I do here is leave some space on either sides and make sure the block extends outside the plane tile on the opposite side so that when we tile them together, it all fits perfectly.

Figure 2 (Placing blocks in the plane #1)

Figure 3 (Placing blocks in the plane #2)

Once we are done placing the blocks on the plane, it looks something like the picture attached below.

Figure 4 (Final floor tiles)

Here, I just tested if the tiles when duplicated together place perfectly and it does. I start randomizing the scale of all of the tiles randomly so it looks more man made than machine made.

Figure 5 (Randomizing the tile scale)

Since the floor tiles are finished, I use the same methods and create a horizontal stone tile and a vertical wood tile for the walls of our diorama scene.

Figure 6 (Final tiles for the floor and the wall)

After all of the assets in our dungeon diorama scene have been UV unwrapped properly, it is time to export those said assets and plug it into ZBrush where we start sculpting those low poly models. This was by far the most fun phase of the entire pipeline as you get to be creative and show off your own designs.

[developingskills1and2]

Practice 1: Developing Skills

Developing 2 (Week 2)

Before unwrapping other models, I decided to see which pieces in an asset itself could be modular. For instance, the door bolts and the arrows are modular of course, but the stone blocks on the door structure is also modular as it is the same pieces duplicated 180. For that, I decided to delete those pieces and only unwrap the original pieces.

This way, when I duplicate it in the end, they share the same UV maps and the same UV space, which helps optimize the scene as well as saves a lot of time when it comes to texturing those said original pieces.

I work on those door pieces in the wireframe mode and delete duplicated faces.

Figure 1 (Wireframe mode)

Figure 2 (Deleting duplicated and/or repetitive faces)

Now, I am left with these door pieces. It saves a lot of time trying to UV unwrap those pieces and cuts the time by half when sculpting the door structure.

Figure 3 (Remaining door structure)

Same as for the pillar, I start cutting seams and unwrapping the UV maps for the door structure. Then, to pack them properly in the UV editor, I use the Layout option.

Figure 4 (UV unwrapping the door #1)

Figure 5 (UV unwrapping the door #2)

Figure 6 (UV unwrapping the door #3)

Figure 7 (UV unwrapping the door #4)

This is the final UV map for the door structure packed properly in the editor. This means that it ready for an export to ZBrush for sculpting.

Figure 8 (Final UV map of the door)

For some of the basic primitives, I only cut little seams and use the automatic wrapping as those elongated beams still retain their cubical properties and don't need much unwrapping.

Figure 9 (Final UV map of the beam)

Figure 10 (Final UV map of the modular small beam)

Figure 11 (UV maps of the assets in the diorama)

Figure 12 (UV maps of the barrel)

All of the assets in the scene have been UV unwrapped properly and are displayed below.

Figure 13 (Final UV maps of the diorama scene)