Design Final Major Project (DM3103)
Roman Fort Diorama
Project Synopsis
The final major project will aim to bring Roman history to life through a Diorama, set within a Roman fort. The project will consist of both world-building and 3d modelling. Throughout my time at University, I have typically left my 3d assets with substance and never brought them into the engine. This is why I have decided to create Diorama. My hope for this project is to expand my current skill set in modelling while also gaining fresh approaches to game engine development. The project will consist of the following tasks: sketching and blockouts, modelling development, textured final renders and finally world building. The end product will consist of high-resolution screenshots that offer different perspectives through the models and the environment.
I will create 4 models for this project, one hero asset and three secondary assets that will act as supporting models to add some depth to the scene. The hero asset will be the Roman fort itself, and the secondary assets will be a farmhouse and stables. I will be downloading other assets like foliage, and they will be referenced and credited. All references will be added at the end.
The 4 models I will be modelling and developing are as follows:
Hero Asset (Roman Fortress)
Secondary asset 01 (Farmhouse)
Secondary asset 02 (Stables)
Secondary asset 03 (Barracks)
Project Schedule
My Trello board outlines when I plan to begin each stage of the project to help me keep a clear and concise plan, which will allow me to keep track of my progress. The first step to make the Trello board was to understand what tasks would be needed while also thinking about how much time would be required for each task. The sketching and blockout stages would be relatively straightforward, as I knew once I identified the right reference images that I felt suited my vision, I could begin sketching through Figma. The Modelling development section is where I expect more time to be used. I mentioned below that the time committed to the project will be determined by the level of detail. The texturing process should only remain one week, as I believe that my sketches will have a clear art direction, which will allow me to begin texturing without having any delays. The game engine work will very likely be two weeks, perhaps more. I made it clear than challenge here would be dealing with UE5 optimisation and other technical issues. If needed, I will either keep a week free to catch up if necessary, or I will simply move to Unity or UE4, which will still allow me to finish of the project.
Figure 1
Digital Sketch (Hero Asset)
The first task was to begin designing my early sketches and outlining my art style. The first task was to choose a reference image that I felt would represent my vision for the project. The reason I opted for this design is simply that it embodies a Roman-style structure, and as a fort, it feels and looks different to a regular medieval fort that is more common in history, and this project would be very different to past historical structures that I have made.
I began with a basic outline of the structure. This was quite a challenge due to the amount of very solid straight lines, and using the mouse became a problem. I quickly found that holding shift while sketching helped keep everything organised and clean.
The annotations I have created outline my design and technical decisions when it comes to my modelling and texturing. The fort path will be developed with the game engine, while the rest of the annotations will be a part of my modelling workflow.
Figure 2
Figure 2.1: Roman forts in Britannia (HeritageDaily, 2018).
Secondary Assets 02 & 3 (Roman house and Stables)
The second and third sketches focus on the secondary assets. The Roman farmhouse and stables will be placed within the inner walls of the castle. The art style for this asset will be focused on different wood designs, which will be different from the castle and will offer some realistic depth to my scene.
The sketching process began by creating a thicker line to gain an outline of the model. This made it easier to avoid drawing outside and making the image look less messy and consistent. The biggest challenge I found within this task was keeping the lines straight, which, like the fort, became slightly harder; however, I was able to fix and overcome this by simply holding shift, which allowed me to fix and overcome any in even parts.
While developing the sketch for the stables, I realised that the model would be quite simple and lacked visual depth. I decided to implement significant changes that still included elements of the original design while introducing new creative decisions. I will explore this task further within the modelling development stage of the project.
Figure 2.2
Figure 2.3: Villa Ventorum at The Newt exterior (Exploring Building History, 2022).
Secondary asset 04 (Roman Barracks)
The final model design required was the Roman barracks. This is a key asset for a castle or a fort, and I was able to find a reference image that suited my design. The sketch was not quite as good as the rest, and this was due to my time management; I was struggling on what asset I should design, and I came across this design through a modelling website.
The key skill I anticipate is creating the tiles, this is a design choice I have never conducted before, and I feel that it will be a good challenge to overcome as its detail that I don’t always think of adding. My hope here is that I can learn and develop this skill due to the importance of visual visibility.
Figure 2.4
Figure 2.5: Roman barracks reconstruction (MyMiniFactory, 2022).
Top Down Map (Blockout)
The pre-production consists of a digital top-down map and a technical interpretation. The benefit of creating both is to gain digital annotations while also giving a technical viewpoint from which the model will be created. The structure I decided on had two main gate entrances that serve as typical pathways into the fortification. A Roman fortress would typically have multiple entrances, depending on its size and position. (Historic England, 2018). The Sketch also includes the 3 secondary assets, which will be the main assets in the scene, but I may adapt their placement, as I may reuse these assets to add more depth to the diorama.
Figure 2.6
Figure 2.7
Every event we host is designed with intention, from the atmosphere we create to the way each session flows.
Figure 3
Roman Towers Stage 01
The first step in creating the towers was to focus on the scale and measurements. This will allow me to focus on consistency and reusability, which are key concepts of the project that match the walls and towers, so the assets look and feel equal. The main task here was to build the tower’s roof. I added in 1 swift loop to create a suitable edge, and then extruded the edge upwards. The angled edges were created using the bevelled tool. This creates the inverted edges within the inside of the asset for the roof.
Figure 3.2
Challenges and errors (Problem)
The only issue and error I encountered while modelling the walls and towers for the castle, the process was simply creating a cube that fitted cleanly between the towers and main gate. I repeated the process for creating the walls; however, once I cloned them, I made the mistake of cloning them as instances and not as copies. This caused the asset to multiply during the attachment process, therefore creating overlapping assets and poor typology. This did happen to the windows, as I did not realise at the time until I cloned each of the assets.
Figure 3.3
Figure 3.5
Windows (Before refinements)
Once the cuts were made, I could begin adding key detail points. The first task involved using a new skill to create a brick asset that could sit on the face to add some visual detail. The process to make the bricks was to select polygonal modelling, then I chose the brick pattern. Once inserted, I could chamfer the edges and extrude the cubes out, which created the brick detail. The other method, which I typically have used, was to add swift loops across the face and extrude each of the created polys. The challenges with using this method are that you never see clear gaps between the bricks. I believe the polygonal modelling method works for this specific task. The archway and windows were added after the bricks had been created in order to manage typology and to avoid overlapping layers.
Figure 3.1
Roman Towers Stage 02
The process for adding detail involved adding the arched windows and the bricks to the towers, the only difference being the addition of new polygonal modelling. The bricks were made again to fit the specific shapes and dimensions, which, compared to the walls and main gate was very different. Once one of the towers was completed, I could clone it and quickly place it in the specific locations.
Figure 3.3
Challenges and errors (Fixed)
The method I used to fix this was to select the element tab from within the assets properties panel, and this allowed me to select the specific asset, and then I could delete the asset that was causing the overlapped asset. Had I not fixed this, the texturing process would have encountered inconsistent layers and would have the work look lower quality. This fix demonstrated my ability to fix mistakes and maintain the project’s timescale
Figure 3.4
Texture test
To understand this challenge further, I decided to conduct a texture test. I imported the model into substance and the results were that some parts of the model textured fine and some parts did not. The polygon selection tool was not accurately applying textures to assets like windows and doors, which resulted in the model being of a lower quality and lacking key details.
To further this, I tested the model into ue5, and the textures did not reflect the export from Substance. This result convinced me that the bricks were causing more issues than they were worth, and I decided to delete them and focus on improving other key assets to ensure the final model looked as good as it could. Figure (3.5) shows the substance texture, and Figure (3.6) shows the UE5 export.
Figure 3.6
Structural Refinements
While I did decide to remove the bricks, I did want to make some key refinements to minor assets like windows, doors and the barriers and finally tiles . The changes I made to the windows were to add wooden planks to the windows, which added a more realistic appearance. The windows were simply flat surfaces, and this lack of detail would not represent the vision I had for the project. The doors needed improvement as they had uneven surfaces and visual detail that I do not believe was a suitable design. Both the windows and doors did have problems with the polygonal selection, so I knew this was something that needed to be fixed. The wall barricades contain gaps between each of the assets, which was a mistake I made while using the array tool. The changes for this are to add spikes, which I feel would make the fort look more secure and more historically realistic. The tiles were not originally added, mainly due to the already high poly count due to the bricks. Since their removal, I decided to add them to the model. The tiles were created by bringing in a cylinder and cutting it in half to create the shape, i then used the shell modifier to create the thickness. this then was placed on the roof where I could duplicate using the array tool along the roofs.
The changes for the windows are kept in (Figure 3.7) and (Figure 3.8).
The Door improvements are situated within (Figure 3.9) and (Figure 4)
The wall barricades improvements are within (Figure 4.1) and (Figure 4.2).
The creation of the tiles are located within (Figure 4.3) and (Figure 4.4)
Figure 4.3
Windows (After refinements)
Figure 3.8
Doors (After refinements)
Figure 4
Wall barricades (After refinements)
Figure 4.2
Tiled Roof (After refinements)
Figure 4.4
Figure 3.7
Doors (Before Refinements)
Figure 3.9
Wall barricades (Before refinements)
Figure 4.1
Tiled Roof (Before refinements)
Roman Fort (Modelling Stage 01)
Roman gate (Stage 01)
The first stage of developing the main gates was to add a cube and begin by scaling it to a realistic height and width. To make it accurate and realistic, I brought in a cube to act as the tower, which allowed me to make a judgment of the size of the assets. The second objective was to create cuts into the wall that could be suitable for archway structures. One issue I encountered was that the pro-Boolean tool created an inaccurate cut using the archway, which led me to a basic insert using a cube. This method created equal cuts, and while it was not the specific arch insert, I was still able to utilise a archway shape to create the detailed viewpoint through the fort.
Roman Gate (Stage 02)
Roman Farmhouse (Secondary Asset 01)
House structure
The base structure of the house had two technical directions, One method was to create a pyramid and munipulate the edges to create the firm base, however I found this method to damage the typlogy and I did not feel it represented my original design. The other design choice is extruding the main face and adding in the edges to make each side of the model can be cleanly extruded which created the cube angled faces for the roof.
Figure 4.3
House Windows
The windows were developed by using two cubes; the design began by adding a swift loop to each of the corners. I was then able to extrude the shape inwards, which creates the gap to add the window frames. I could have undertaken this by using the original asset to create the frames, but that would have encountered issues that could cause issues with UVs. I also felt this looked much closer to the reference image meaning it was more visually appealing.
Figure 4.5
Main door and lower walls
The door was designed using a cube shape, which was then extruded to make the cut in the wall, I then cloned the asset which allowed me to begin adding the key details. The first was 4 small cubes that were attached to the cube to act as anchor between door and the hinges. The second shape was acting as the ropes, I completed this by adding a set of torus shapes, while applying the array too keep accurate spacing along the door. The lower walls followed the same workflow to add the bricks to the lower area of then model.
Figure 4.4
Roman stables (Secondary Asset 02)
Roof frames
The original design of the stables did not have enough visual appeal, and I decided to make some changes to the structure. The first task was to create the frames for the roof. I began by adding a cube and then creating an angled asset that would connect downwards to create a suitable frame. To avoid the asset looking consistent, I rotated the bottom faces to sit on each edge; the other reason was that the cube was not positioned properly and it made the model look uneven.
Figure 4.6
Roof completion
Once the roof planks and frames were in place, I was able to clone the wooden planks to fit each of the gaps. I angled each of the faces and fit them into place, which were accurate due to the accurate spacing using the array tool. One important factor during this process was remembering to clone as a copy and not as an instance. This was a problem I faced during the modelling process of the fort . This demonstrated a clear response to how I deal with the challenges within a modelling practice.
Figure 4.8
Roof wooden planks
The roof frames were cloned using the array tool. This was done to create equal spacing between the assets to ensure that the roof was equal. The design of the wooden planks was undertaken by using a cube to create the plank, which was placed at the same angle at the frame. I was then able to use the array tool to create the roof in order to maintain reusability and allow for accurate spacing. This process allowed the model to look accurate and even between both the frames and the planks.
Figure 4.7
Barn development
The final task was to create the main structure. This was completed by adding 4 support beams that were placed across both sides. This process was made much easier due to the array tool, as I can cleanly align each of the main support beams. The fence was created by extruding the face of the cube inwards and then deleting each of the faces. This was done to make the model feel free and visual rather than closed in and empty of detail.
Figure 4.9
Roman Barracks (Secondary asset 3)
Barracks Structure
The Roman barracks development began by creating a slanted roof that could then create a different feeling compared to the simple flat roof design. This design choice was influenced in this way because of its visual detail and the opportunity to learn a new skill of modelling tiles, which are a crucial aspect of the reference sketches.
Figure 5
Barracks Structural details
The pillars acted to support the lower roof; this was a design that could have been completed using Roman pillars or basic cubed boxes. Although the reference image showed simpler support beams, I decided to create a more authentic asset that reflected the Roman era. The door proved to be slightly harder to do due to the angle of the reference image, which meant I could not see details like the handle as easily. To fix this, I added a torus shape to act as the handle, which felt more appropriate for an ancient/medieval rather than a modern structure.
Figure 5.1
Roof Tiles
The process for creating the roofs was to follow the same procedure as the tiles within the Roman fort. The tiles were taken from the Roman fort I designed, and this greatly improved efficiency while developing this model, as I was able to avoid the repetitive process of recreating the asset. The only challenge with this was to scale and resize the asset down to fit the barracks. Once the tile was scaled and placed began to use the array tool to efficiently duplicate the asset along both sections. This process not only supported the reusability of assets while also maintaining historical realism.
Figure 5.2
Roman Fort (Final Textured renders)
Roman Farmhouse (Final Textured Renders)
Roman Stables (Final Textured Renders)
Roman Barracks (Final Textured Renders)
SketchFab Design documentation
Environment Development Process
Stage 01: Downloaded Assets
The first area I will discuss is the downloaded assets. The assets I chose were to offer increased depth and realism to the world I want to create. The first asset was a forest material, which was chosen due to its contrast between dark and light colours and its connection to the grass foliage asset, which was unique in its ability to blend in with the forest asset. The 3rd asset I added was a tree asset. I had different models in mind; however, the one I looked at felt slightly too cartoonish, which would not have suited the art direction for my diorama. This encouraged me to choose the tree below, which I felt suited the project better. The final asset I downloaded is the stone pathway. While looking at Quixel mega scans, I found a Japanese-style stone pathway that, while not Roman in origin, did resemble the bricked Roman road, fitting both my scene and artistic vision.
Figure X: Forest floor asset (Quixel Megascans, 2024).
Figure X: High poly tree model (Next Spring, 2025).
Figure X: Grass 3D model (Ogussstavo, 2025).
Figure X: Japanese park stone floor (Quixel Megascans, 2024).
Environment Development Process
Stage 02: Landscape development
The vision for the diorama was to create a set of the mountains and hills that surround the Roman settlement, displaying depth and congestion.
The first step was to add the landscape material and begin attaching the textures within the material graph. Since this texture was downloaded from the UE5 marketplace, I understood that I only needed 4 specific textures to create the material: Base material, Roughness, AO (Ambient Occlusion) and finally Normal maps. These were combined within the material graph to produce surface detail with realistic lighting and depth across the landscape.
Once the terrain was properly sculpted, I could then add specific effects to make the hills look more realistic. This task was conducted using the erosion tool, which was a tool that I was not familiar with; however, this became very crucial for creating damaged and eroded areas in the hills. Alongside this, I utilised the flatten tool to create circular flat faces that could be eroded into the mountain; the smooth tool proved useful for making the hills more bumpy than sharp, which resulted in the hills feeling more realistic and appropriate for my scene.
The construction of the landscape began by sculpting the terrain upwards to create hills that resembled a semi-circle shape that surrounds the scene enough, while also offering a clear and direct path for a player. To understand the scale of the scene, I decided to import the Roman fort into the scene, since this was the hero asset of the Diorama.
Environment Development Process
Stage 03: Asset Texture Maps
Environment Development Process
Stage 04: Fence development
During my development of the scene, I decided to 3d model a wooden fence that would be placed alongside the stone pathway. This was not originally planned; however, I felt this would enhance my project massively and give greater depth to the approach to the castle.
Due to the time constraints and lateness in the project, I decided to go with a basic wooden design. I began by creating two cubes, which both had chamfered curved edges, which could then be cloned and attached.
The 2 cubes going across were connected using the bridge tool; this method was used due to its accuracy in connecting separate models together.
Fence Final Render (Substance Painter)
Fence Final Render (Unreal Engine 5)
The final step of developing the model was to use the array tool to quickly duplicate the asset and create a suitable fence. I decided that it would be more efficient to make the fence in full within 3ds Max, as I knew that I could quickly clone and scale the asset if necessary.
The fence also contained an arch, which was to allow the stakeholder to move in other directions. This design was designed by extruding the top face of the fence, which could be connected using the bridge tool.
