Faculty of Arts Department of Game Design

(1)

Achieving Visual Diversity

A Study within a Modular 3D Environment

Samuel Dangoor Ehnberg

Faculty of Arts

(2)

Abstract

The digital game industry is evolving fast and games are increasing in scale, both regarding the size of the development teams, as well as the size of the digital worlds provided for the players.

Much time and effort is put on creating these worlds, and there are multiple techniques in use with the purpose of the creation of visually appealing games. One of the issues that follows the design of these vast fictional landscapes, is the risk of the world looking repetitive, thus unappealing for the player that spends several hours in that world.

This thesis covers the approach in the creation of a 3D environment by using different techniques with the goal of avoiding visual repetition within a game environment created as a practice based fundament for testing, and comparing the different approaches that a 3D artist can take when constructing a game environment. In the thesis, examination is made of how the different techniques contribute to visual diversity, trying to draw a conclusion on what role they play within a modular 3D environment. The 3D environment was used to conduct a comparative case study. The environment was displayed for the participants where their opinion on how the different visual elements helped to avoid visual repetition was analyzed. The results from the survey proved that the different techniques applied to the elements, was perceived according to the theory applied to the construction of the 3D scene.

Keywords: 3D, modelling, procedural textures, visual fatigue, ferry, mesh, object, construction Abstrakt

Den digitala spelindustrin utvecklas snabbt och spel växer i skala, både vad gäller storleken på industrin, liksom storleken på de digitala världar som spelarna förses med. Mycket tid och ansträngning läggs på att skapa dessa världar, och det finns flera tekniker i bruk med syftet att skapa visuellt tilltalande spel. En av de problem som följer utformningen av dessa stora fiktiva landskap, är risken för att världen ser repetetiv ut, och är därför bristfällig för spelare som tillbringar flera timmar i den världen.

Denna avhandling täcker tillvägagångssättet i skapandet av en 3D-miljö med hjälp av olika tekniker med målet att undvika visuell upprepning inom en spelmiljö, som skapats som grund för testning och jämförande av de olika metoder som en 3D-artist kan applicera vid bygge av en spelmiljö. I arbetet granskas hur olika tekniker bidrar till visuell mångfald, med avsikten att dra en slutsats om vilken roll dessa spelar inom en modulär 3D-miljö. 3D miljön användes för att sammanställa en jämförande undersökning. Miljön betraktades av

undersökningsdeltagarna, där deras åsikter kring hur de olika tekniker påverkade den visuella enformigheten inom 3d miljön analyserades. Resultaten från undersökningen visade att de olika tekniker som hade tillämpats på de grafiska elementen uppfattades överenstämmande enligt teorin som var tillämpad på konstruktionen av 3D miljön från början.

(3)

(4)

List of figures

Figure 1:Picture from reception part of the scene (Dangoor Ehnberg, 2017). ... 4

Figure 2:Coulage from survey, (Dangoor Ehnberg, 2017). ... 10

Figure 3:White-boxing layout (Dangoor Ehnberg, 2017)... 13

Figure 4:Overview picture of wall from scene (Dangoor Ehnberg, 2017). ... 15

Figure 5:Figure 5: Modular parts used to build the wall ... 15

Figure 6:Different objects from scene using the same material ... 17

Figure 7:Roof example from scene (Dangoor Ehnberg, 2017). ... 18

Figure 8:Reference picture from authentic ferry, taken by author (Dangoor Ehnberg, 2017). ... 19

Figure 9:Picture displaying each tile used to build the roof (Dangoor Ehnberg, 2017). . 19

Figure 10: RGB mask used in material Unreal Engine 4... 20

Figure 11:The blended materials on the carpet material (Dangoor Ehnberg, 2017). ... 21

Figure 12: Survey results ... 23

Figure 13:Results on comparing pictures from scene (Dangoor Ehnberg, 2017)... 24

Figure 14:Larger detail opinion graph (Dangoor Ehnberg, 2017). ... 24

(6)

1 Introduction

I have always been very interested in 3D art within games and other media. For me 3D is a tool for expression and the creation of dynamic game experiences. Since I started the Game Design and Graphics education at Uppsala University my interest in 3D environmental design has grown rapidly. My interest in games in general has been extensive ever since I was little.

When I learned how to construct 3D, since my area of interest was in environments, I wanted to create worlds for games that would leave an impression similar to my gaming experiences.

This thesis presents the process of creating a dynamic and realistic 3D environment using different techniques to avoid repetitiveness and achieve realism. The project that the thesis revolves around is a 3D constructed environment based on a real-life ferry that runs between the cities of Visby and Nynäshamn in Sweden.

This ferry was used as a source of inspiration when it came to the actual construction of the scene. The fundamental hypothesis of this thesis is how a 3D artist can avoid repetitiveness using a constructing principle called modular environment creation, often used within games requiring many different assets throughout a large area.

One of the main issues that game studios today struggle with is to avoid visual fatigue in their fictive game environments. One could think of the game as a puzzle, requiring many pieces to be fully complete and can be assembled in different ways that affect the computational

performance. However, this thesis does not focus on performance issues, but rather aesthetic appearance. The thesis examines how a 3D artist can avoid repetitiveness using as few puzzle pieces (assets) as possible.

The assets in themselves were constructed in a way so that the designer easily could change its appearance with the least amount of resources as possible. This was of high importance, since the survey included in the work was based on how the different techniques were perceived by the audience participating in it. By the parameterisation of the assets, many different scenes could be assembled quickly, which was required for the survey, since the participants could compare many different scenes.

Certain concepts that are used in this thesis have been defined in a glossary of terms at the end of the thesis.

(7)

2 Background

This thesis is based on the design of a 3D environment originating from a real-life venue.

However, the purpose is not for the created environment to resemble the actual location that the scene was inspired by, but rather work as a basis of structure in a fictional universe. The scenario is based on a pandemic post-apocalyptic theme. A virus has harvested almost every living life on earth and only a few have survived the lethal virus. The last survivors are scattered across the globe, most of them on remote locations far from tightly inhabited areas where the virus was most concentrated. One of those locations is the Gotland ferry that floats around in unknown waters. Ever since the outbreak the boat has functioned as a well-guarded quarantine that protects its citizens from the disease. The boat has functioned as an optimal zone for safety, since it is far away from all possible biological threats, and after society broke down, threats from raiders looking for slaves and food.

The protection from the virus did not hold for long. Now the disease has reached the ferry and exterminated its inhabitants. All that is left is a deserted ferry that floats around in the vast oceans of earth. Some things are still functioning on the boat, things like the electrical generators driven by the floating motion of the ferry, which have kept the ferry lit even though no one is left to maintain the systems.

When the virus struck the ferry, chaos broke out. Desperately trying to inhibit its rampage, the military that stayed on the boat tried to shoot the people that were infected when they

desperately ran towards them for help. This chaotic scene has left permanent marks on the boat with dents on the walls and debris all over the ferry's interior. The only traces of the people who lived there are found on the floor and the walls in form of blood stains.

The thought of the story behind the scenario was that it should follow a somewhat realistic setting and not be exaggerated. The main reason for this was that the setting should match the graphical theme of the scene that was realistic. It was also thought that a post-apocalyptic setting would facilitate the creation process, since it often includes much destruction that can be used to camouflage repetitiveness and construction errors in a subtle way. The post- apocalyptic destruction came in this case in forms of debris piles and chaotic texture pattern that proved to be effective in its ability to avoid repetitiveness. The fact that the theme was post-apocalyptic, brought construction as well as story-telling benefits expressed trhough the environment.

(8)

3 Previous Work

This section mentions some previous works on the subject matter of 3D environments, and are considered relevant for the thesis. The previous work has a common ground of research. In

“Implementing a Modern Workflow in a Realistic Environment Design” Medina at al. (2014) have written about modularity within 3D environments and its technical aspects. Within this study, Medina at al. examines the different technical aspects of building a modular environment.

Their focus is on the technical aspects of how one can resemble a authentic environment, and build it using modular assets within a game engine while comparing the different artifacts to find out how they affect the appearance of the scene.

Similar to “Implementing a Modern Workflow in a Realistic Environment Design “is

“Generating Compelling Procedural 3D Environments and Landscape” where Blomqvist at al.

(2016) examines how a world can be auto generated using algorithms rather than constructing the assets themselves. The mentioned theses revolve around modular environment creation and its benefits. They do not specifically bring up the subject of avoiding visual repetition, which is something that this thesis revolves around. Since the subject seems to be lacking in previous documentation, it becomes more relevant for this study to bring up the issue of visual repetition specifically within modular environments.

(9)

4 Purpose

The goal of this thesis is to give the reader a clear perception of how visual fatigue can be avoided using different modelling techniques, such as modular level design, placement, and procedural materials, among other methods described in the text. The thesis covers the whole process from the idea to the final results, and how the expectations of the outcome covers the final results. The principle question for this thesis has been what techniques a 3D artist can use in order to avoid visual fatigue and make the scene as dynamic as possible.

The thesis examines the different aspects of what it takes to create a dynamic, realistic and adjustable 3D environment. In order to provide for a good foundation for the thesis, a survey was launched, with results that seem to indicate to support the hypothesis of the paper. The main purpose of this was to avoid subjectivity in the theorization process and to give a clear picture of how the scene was perceived by an independent audience. The survey participants provided their opinions on different pictures displaying the different parts of the scene where different decisions had been made, and how those decisions were perceived by the audience.

These opinions were later on gathered and compared to the thoughts and decisions made by the designer during the process.

Some chapters describe the technical process of constructing the different components used within the environment, while some chapters discuss the practical solutions implemented for functionality, all with the main goal of achieving visual diversity.

Figure 1:Picture from reception part of the scene (Dangoor Ehnberg, 2017).

(10)

5 Methods and Materials

To fulfil the purpose of the study a practice-based approach was used, which is to create 3D artefacts, that would build the environment. These artefacts would later on be used as

comparative base in a survey.

5.1 Pipeline

The main pipeline for assembling the scene had to follow a strict structure in order to facilitate the construction process of the environment. The main pipeline for the whole construction process was as follows:

1. Planning what part of the boat that was supposed to be built in the 3D scene, keeping in mind what benefits and cons each part of the boat would entail if they were chosen as the target for building the scene. Since the time span was quite limited, the most efficient and least time consuming plan had to be followed in order to be able to assemble the project. But the part chosen also had to contain sufficient aesthetic interest for the purpose of the project.

2. Gathering as much reference material as possible from various sources. Sources such as pictures taken by the author and evacuation plans from the ferry company's website that gave an overhead picture of the structure. And from this decide the scale of the props based on the estimates from the reference material.

3. Starting to white-box the scene. Determining the placement of each object in relation to each other. Starting to plan the level in detail and perceive how the objects relate to each other and how the space in the environment appears to the spectator.

4. Starting to plan how the different meshes were supposed to be constructed in order to be able to be assembled correctly, avoiding gaps and construction errors. Building each mesh in 3ds Max and later exporting them to the engine to see if they fit. Measuring each mesh in order to be able to snap to the grid in the engine that facilitates the assembling process.

Without any doubt one of the most time consuming parts of the project.

5. Building each mesh in detail still having the modular layout in mind, and smaller details on the meshes that physically stand out from the basic mesh.

6. Starting to create basic materials for the scene in order to evaluate the different textures color composition and decide the way the materials should be designed, in order to facilitate short and long term design decisions about the scene.

7. Making the materials procedural so that they can be manipulated later on using instances and parameters that affect the materials appearance in real time without having to

destructively change the basic texture.

8. Assembling all the components included in the scene, such as final mesh, material, and lighting results.

Gathering external opinions based on a survey that to examine questions related to the purpose of the thesis. The survey aims at getting the participants perception on how repetitive patterns appear to them, and what tools they think contributes the most in order to achieve graphical diversity within the scene. The questions in the survey follows a structure of having two

(11)

parallel pictures next to each other, captured from the same position within the scene but with different appearances. One picture contains variation in texture while the other contains variation within object placement and mesh design. The purpose of this questionnaire design is to understand the participants opinion on which of the texture variation or mesh designs contributes the most to diversity within the scene.

The target audience for the survey was from a specific gaming forum found on

GOG.com, a game retailer site with a community for gamers. This particular audience was chosen since they were estimated to have experience within open world games in general and thereby being able to provide with their opinions on the scene, with their other gaming experiences as comparison and fundament for their thoughts. A thread was started on the forum explaining the work behind the designers scene, and asking them to fill out the survey.

The survey questions do not direct the participants to a forced answer, but rather try to find out which of the pictures they think is the most diverse without telling them what techniques that were applied to the picture. Since the designer knows the difference in methods between the pictures, the survey would not benefit from highlighting to the participants the technique in use, but would rather risk to nudge the participants to give a forced answer, since they would understand the structure and purpose of the questions. The answers have to be as honest and natural as possible, avoiding leading questions. The main purpose of the survey was to find out which factors, consisting of placement, texture or sculpting, that contributes the most in the avoidance of visual fatigue (Burges and Purkeypile, 2017).

The pictures used for the survey was as follows:

(12)

(13)

(14)

(15)

Figure 2:Coulage from survey, (Dangoor Ehnberg, 2017).

5.2 Modular creation

Modular 3D environment building is based on a principle where the various meshes are assembled in an engine rather than constructing larger pieces in the main 3D software. One of

(16)

character's size does not fit within the game asset, a door for example. Instead of making the adjustment in the 3D creation software, one can simply move the object that is blocking the character in the engine. Of course some assets in the game can be static and not movable in the engine and therefore have to be adjusted in an external program, but the point is that there is in general faster to make adjustments in a modular environment, since the whole principle of the technique revolves around a large amount of separate and movable pieces (Mader, 2005)

5.3 Software Autodesk 3ds Max

3ds Max is a 3D modelling software developed by Autodesk. The software was in this project used to construct the 3D meshes in the scene. The program was used to create the basic mesh and then refine them to the final result that would be implemented into the game engine.

Before exporting the mesh to the game engine, all the final construction had to be performed:

to look for construction errors, limit the amount of polygons, make all the measurements fit according to planning, UV-map the mesh, and assign the correct material IDs so that the different materials in use could effectively be mapped to the surface of the mesh.

Allegorithmic Substance Designer

Substance designer is a software used to develop adjustable and realistic materials using information nodes that in a non-destructive way affects the material appearance. The different nodes contain parameters that can be exposed, thus are visible in other software, enabling the designer to adjust the appearance in external software, in this case, exposing the parameters to Unreal Engine 4.

Allegorithmic Substance Painter

Substance Painter is unlike Substance Designer, a software where the designer paints textures directly on the mesh, resulting in bitmaps applied to meshes. This is due to information degradation, a destructive way to create the materials, since the artist has to switch between software applications in order to make adjustments.

Unreal Engine 4

Unreal Engine 4 is a game engine software, the final media platform of building a game, or in this case just a game environment. The software was used to assemble the whole project and put all the objects in place in order to create an appealing scene. The software contains various functions that facilitate the process of creating a game, such as a grid based snapping function that was vital for the modular technique to work.

(17)

6 Design process

6.1 Preparation of scene

Since the work on the ferry was inspired by the real life Swedish Gotland Ferry that runs between the cities Nynäshamn and Visby, preparation had to be done in order to achieve some basic similarities when it comes to scale, overall color composition, object density and

placement. The first step was to gather as much information as possible on the actual ferry to facilitate the white-boxing process that was soon about to begin.

Thus, the first step in the preparation was to gather reference photos from inside the ship's cafeteria, which was the part chosen as the modeling subject. Since the author often traveled by the ferry, this was not an issue. When taking the reference photos of the interior, the author had to keep different perceptions in mind: overall layout, placement along with the mesh, and the texture details. Some photos were taken trying to capture the bigger perspective while some photos were aimed at the finer details. All of this data was later used to facilitate the modeling and texturing of the ferry.

When using the photos as the reference material, the author soon realized that a more general reference was needed in order to get the overlaying placement of the objects within the scene.

This was performed through gathering an evacuation plan which displayed the different parts of the boat from a top view perspective available at the ferry company's website (see

Appendix). On this map quite many assumptions had to be made due to the absence of information on specific parts that was needed in order to get a good overview of how those would relate to the ferry's size. The only measurement information that was published was the length and height of the boat. By using these values, the author could get an idea on what scale the actual objects inside the boat had in relation to the overview.

On the map where the emergency escape plan was depicted, notes were made in order to determine the scale of the objects, measured by the metric system. Since the size of the ferry is known, conclusions could be drawn on an object size based on how it related to the scale of the whole ferry. The planning can appear as insignificant and overestimated but proved to facilitate the perception of the layout in an effective way.

Next it was time to place actual objects within the scene based on the planning. When starting to place the objects, not much attention was drawn to detail since that would be the next step in the construction process. Now it was time to white-box the scene.

6.2 White-boxing

White-boxing was used in order to structure the overall construction of the cafeteria interior.

Based on the assumptions from the top view escape plan, the white-boxing components had to

(18)

all the exact measurements before the detailed assembly began. If the measurements are unknown when the detailed creation of the mesh starts, it can result in construction errors, leading to unnecessary iterations (Casen, 2016).

In order to avoid repetitiveness in a level, one has to not only give attention to the smaller details in the scene, but also larger parts of the level. The cafeteria part, as shown in Figure 3, contains a repetitive and organized layout from start. One of the reasons why a post-

apocalyptic theme was chosen was to be able to disrupt and camouflage the repetitiveness using destruction. The destruction was composed by different deformed and destroyed objects that were in the scene. Objects such as destroyed chairs and withered concrete objects. These meshes were made as separate pieces like the rest of the 3D objects following the modular principle of making multiple objects that can create larger constructions when put together.

The destroyed parts were also included in the white-box prototyping of the scene. These parts were used to see where the different repetitive patterns could be camouflaged and at some points avoided. This contributed to the planning of the layout since it provided ideas on how and where the destruction could be used. In this scene, destruction allowed the building of the scene to be more flexible than what it would have been without it. Not only was destruction used to camouflage parts of the map for aesthetic reasons, but it was also used in order to cover different construction flaws that comes with modular environment creation, in this case, gaps between objects. This could be regarded as one of the flaws with modular assets, since larger objects that are fully assembled in a 3D software do not contain gaps to the same extent as the modular assets. These gaps were concealed using separate meshes created for that purpose, but they could also be avoided based on how the materials are created and how the light in the scene is mapped onto the object.

Figure 3:White-boxing layout (Dangoor Ehnberg, 2017).

(19)

6.3 Window wall

The white-box planning of the layout facilitated the start of the more detailed mesh

construction. The scene was now ready to be assembled since the main measurements were done. The main wall that ran across the far side of the boat was the first subject to be constructed. The reason for this was that this particular construction would work as an indication on how the modularity of the scene would be presented. It was a great example of how you could assemble a larger object using smaller pieces.

The wall design incorporated many different construction principles that were perfect for the purpose of mapping each part out, and from that it was easy for the designer to determine how each mesh of the unison wall would be constructed. The reference wall in its whole actually looked like as if it was constructed using various smaller pieces. The two main components that defined the wall was a window next to a pillar part. Between these parts there was a seam that indicated that they were apart constructed.

Figure 4:Overview picture of wall from scene (Dangoor Ehnberg, 2017).

(20)

In the measurement process, these two objects worked as benchmark for measuring the scene.

They were placed next to each other and by knowing their measurements the amount of pieces required could be estimated. The required width amount for the wall had to be three meters while the pillars in between each window were one meter in width. When assembled

accordingly to the layout, it resulted in a total length of 69 meters. This proved to match the estimated length of the authentic cafeteria that was 70 meters, only leaving a one meter margin. It was of much importance that the main measurements of the width and depth of the boat was more accurate than the other scale values within the scene. The main reason for this was that the outer parts, such as the long and short wall would act as a guideline for the overall size of the smaller decorative objects, such as tables and chairs.

Each part of the wall had a determined pivot point that was supposed to act as an anchor point so that the mesh would snap accordingly to the scenes grid-based coordinate system. This coordinate system was used in order to make all the independent assets able to be placed next to each other in an effective way in the 3D space. Without such an alignment mechanism, it would have taken significantly much more time to place the objects on the intended positions.

Since the pivot was placed at the exact center of the object, it related perfectly to the rest of the environment (Mader, 2005)

After the mesh was done, different material IDs were applied onto the objects. By using these material IDs different textures can easily be assigned to specific parts of the object. The alternative would be to adapt the texture manually onto a rendered UWV map. This way of creating textures is, in general case compared to material IDs, not very efficient. But since this whole project revolved around parameterisation and the ability to adjust all the different objects in retrospect, painting directly on the UWV was not an option. Instead, different material IDs were assigned to specific polygon areas on the mesh.

By using this technique, the appearance of the different meshes in the scene could be changed either by changing the material itself or by applying a new material onto the object. In order to save as much computational power as possible, instances were used to keep the

computational power budget at a valid level. An example of this was the metal material that was frequently in use on many different objects. The main light metal material was built on simple principles. The diffuse color on the metal was grey, which meant that it easily could be adjusted within Unreal Engine since it had no color value, but only light information. In doing so a separate parameter has to be declared within Unreal to be able to adjust the color.

(21)

Figure 6:Different objects from scene using the same material

One of the most dominant materials in use was the plaster wall material. A total amount of seven objects in the incorporated this material. It was therefore important that it could change appearance so that the spectator would not notice that most of the objects in fact shared the same material and looked repetitive. The way that it was done was through exposing as many parameters as possible that could change the appearance of the multiple objects.

When constructing the material, the main parameters that were exposed controlled the amount of the destruction on the material. If the parameter was set to zero, it meant that the material would have a clean surface. When increasing the value the destruction would become more apparent.

6.4 Roof

Apart from the floor that was built using one separate asset, the roof had to be constructed in several pieces in order to fit the layout of the cafeteria´s roof structure. Since the roof in its whole corresponded to the size of the floor it was not an issue to estimate the size of the final assembled roof. The problem was in knowing how to construct each part of the roof since it was according to a quite irregular layout, meaning that the different parts of the roof, would have differing scale values depending on the location that it was in. The X and Y axis of the space within the cafeteria contained uneven structures that had to be kept in mind when building each part of the roof.

(22)

Figure 7:Roof example from scene (Dangoor Ehnberg, 2017).

Since the roof like all the other assets had to be properly aligned, it was of much importance that the different square sided pieces that the roof consisted of, was constructed so that they could be seamlessly assembled. Unlike the wall assets that only had to be assembled and matched along the X axis, the roof had to be put together from all sides of the mesh. Meaning that there could not be a single gap in any direction, with the exception of certain areas where destruction was needed. With that in mind it was very beneficial that the roof was built using smaller pieces and not one large piece like the floor, so that the creator easily could remove parts of the roof in order to resemble destruction, leaving room for irregularities. (Burges and Purkeypile, 2017)

On the surface of the authentic roof, there was a panel structure. It was decided that the fictional roof also should contain this detail. This could either be constructed in 3D or it could be done using materials. If the panel structure would be constructed using polygons it would take up much time and effort in order to make details of the roof pieces fit. Because of this, the designer decided to create the panel using a procedural material made in Substance Designer. Why the two options were kept in mind from the beginning was because each alternative brought advantages and disadvantages. The advantage of constructing the panel from the mesh itself would have been that it brings more depth to the surface from multiple angles. It would be possible to see the actual depth of the gaps in between the panels unlike a texture, with the exception of applying tessellation to a texture, and in that case it would have meant that the texture actually contained 3D information. But the author made the assessment that tessellation and 3D information would be unnecessary since the spectator would only see the roof from a bottom perspective (Jim Watson, 2017).

(23)

Figure 8:Reference picture from authentic ferry, taken by author (Dangoor Ehnberg, 2017).

The advantage of only using texture to resemble the panel was that it would requires less computational power for the engine and would increase the potential flexibility in retrospect if any adjustments had to be made. It would also mean that the various parts of the roof could be tiled together more easily. If the panel would have been constructed in 3D, it would mean that all the gaps in between the planar surfaces would have to be perfectly tiled together causing many issues to deal with, which the time frame of the project did not allow.

The texture developed in Substance Designer, was designed using simple rectangular

geometry. Underneath the rectangles, a ventilation net was added as shown in Figure 7. Since all the textures created within Substance Designer are tileable it supports workflow and makes the process of creating textures effective. The combined results were a simple tileable plane created within 3ds Max on which the texture was applied. The author now had to make sure that the mesh had a correct UWV map that would present the texture in a correct way.

(24)

6.5 Floor

The floor was constructed using a single large mesh that covered the whole cafeteria floor.

Unlike the roof that was constructed using multiple smaller pieces, the floor only used one single piece. The main reason for this was that the designer thought that it would be beneficial if the floor followed a basic construction so that it would be easier to place objects on top of it and not having to keep construction factors in mind. The variation of the floor would be represented through blending different materials on top of it using RGB masks, and also by objects placed on it. If the floor would consist of tiles that had different appearance properties such as height information and not just a flat surface, it would bring additional factors that determined how the objects could be placed. By keeping the floor simple in its shape, it proved to simplify the placement process since the designer could build the scene in a more flexible way, without having to adapt the objects on top to any particular shapes.

In order to achieve variation to the floor's appearance, focus was on making the texture as diverse as possible. This was done by using three different materials that were supposed to be blended in a proper way. The first and main texture resembled a carpet structure that the authentic cafeteria had on its floor. The material was constructed in Substance Designer using rectangular and circular shapes that were tied together in order to create the pattern

resembling the actual carpet.

The second material in use was a blood texture. The reason was to be able to tell where the passengers had fought during the last virus outbreak on the ferry. And thus, telling a story through the material by placing it on certain locations and by that tell the spectator where the passengers had been during the last seconds of their lives.

Figure 10: RGB mask used in material Unreal Engine 4

(25)

Figure 11:The blended materials on the carpet material (Dangoor Ehnberg, 2017).

The last material was supposed to represent the surface that hid underneath the carpet itself.

Many attempts were made in order to find what type of material that fulfilled the purpose. At first the creator tried to figure out what kind of material that looked most appealing and realistic. Many attempts using different kinds of concrete materials were made since concrete was a material that was used quite frequently on other objects. The problem with using concrete on the floor was that it was hard to convey that it actually was underneath the torn carpet. This was because the carpet had a quite dark color value, while the concrete contained a very bright value. A consequence of this was that the concrete always looked like it was on top if the material when it was supposed to be portrayed as if it were peaking up from the gaps in the carpet.

Attempts using other materials were made, and finally a suitable material was found. The material that proved to be the best for the purpose was a rusty metal surface. The material contained values that contrasted from the carpet material in a good way, making the border between the carpet and metal more distinct. It was also easier to convey the differences in height between the materials since the metal was darker. Thus, it could give the illusion of absorbing less light than the carpet, and from that give the impression of that it was

underneath the carpet.

The way that these materials were blended was by using a red, green and blue mask. Each of these colors represented a channel which contained information about how the different materials were placed. These channels worked as masks for the materials, telling the textures how they should be distributed on top of the carpet material. This proved to be a very

(26)

7 Results

7.1 General strategies for avoiding visual fatigue

When building the scene, the author kept three strategies as a base for avoiding visual

repetitiveness within the scene. It was thought that by using placement, texture variation, and mesh difference as main factors, it would facilitate the process of making the scene as diverse as possible. However, it was important not to make too many major changes to the 3D objects within the scene since the mesh library was supposed to be kept within a small frame,

meaning that the amount of modular assets would benefit from being as limited as possible.

There are different reasons for this. The first reason would be that it becomes easier for the designer of the game to design it since there are fewer pieces to keep track of. This does not only apply for the project in question, but also for larger scale projects within general game development. (Burgess, 2013). But if the asset library is kept limited in size it can also be very important to make the different assets as functional as possible, meaning that the different sides of the mesh can be used for different purposes, only depending on how the mesh is placed allowing the differently designed parts to be shown to the spectator, depending on its position and orientation. In the current project, this was not frequently used since the assets were based on authentic objects where all sides were exposed. A great example is when rotational variation is used is within organic asset designs, such as rocks and trees. Since they contain irregular patterns they can often be put together, forming a bigger representation of the object. A rock wall for example could be based on many smaller cylindrical shaped rocks that are placed on and next to each other forming a bigger construction. Since the shape is cylindrical, the creator can apply variation to the part that faces the spectator. In the project in question, focus was more on practicalities such as simple rotational diversity and difference where variation laid in the texture and not the mesh itself.

One of the objects that depended heavily on rotation in order to be varied was the pillars. The pillars were one of the most dominant mesh objects within the scene that could be seen from almost every angle. It was thereby of much importance that they appeared as different from one another. Only two sets of pillar meshes were made since it was thought that the rotational placement in itself would give enough variation, and would give equivalent results as if they were designed in multiple different ways. Since the pillars had a round shape they could have a 360 degree.

Unlike most of the other assets within the scene the creator could take use of the round shape of the pillar to create variation. The other assets often contained a square shape, which forced them to be designed in a way so that variation could be achieved through other factors than rotational placement. For these objects, such as the tables, variation was achieved through texture differences.

Another example of objects that used placement in order to create variety was the chairs. The chairs were designed as independent objects that could be scattered around the whole

cafeteria. Its function could be compared to the debris within the scene that was used to create disorder and break repetitive environmental patterns. Unlike the pillars, the chairs did not have to be attached to any certain object, but could in practice be placed at any location and still be convincing from a narrative and logical perspective. The chairs proved to be one of the major variety factors for the project. They were designed in three different versions. One unbroken version, where all the parts of the chair were intact, one version where the seat was removed, and one were almost all the parts except the legs were removed. The reason for why the chairs were designed differently was because of the large number of chairs that proved to

(27)

be used in the scene. It was also to convince the viewer about the chaos that had caused the apocalyptic situation, and thus tell a story using the assets just like the earlier mentioned blood texture.

On the objects that were attached and limited in their shape, focus was on making the texture diverse. In order to achieve enough texture variation, it was important to make the material as diverse and adjustable as possible. An example would be the earlier mentioned floor that contained three completely different materials blended together. The object followed a good principle since one single mesh contained good variation. This was something that the designer strived towards to achieve on the other objects.

7.2 Survey results

To understand how the different techniques are represented and interpreted for someone who has not built the scene, a survey was assembled. The survey contained different pictures depicting the scene where a certain location was captured twice by a camera. It was in the same location in the scene that the picture was taken, but one picture was of the area built with more texture variation, and the other with more shape variation. The purpose of this was to be able to come up with a conclusion on weather shape, texture, or placement contributes the most to variation, and which that avoids visual fatigue most efficiently.

The survey contained five questions with concrete examples from the scene. These questions contained the earlier mentioned pictures that were taken from the same location but built differently. One picture where variation was in textures while the other was more varied in shape. By constructing the two sets in this way the author could easily understand the

participant’s perception and opinion on what visual impact the different techniques had. These questions were followed by more general questions with the goal of finding out what the participant regarded as the most important factor for general visual diversity within the environment, and compare that result with results from the concrete examples to see if the participants could understand the difference between the techniques at all.

(28)

Figure 13:Results on comparing pictures from scene (Dangoor Ehnberg, 2017).

Figure 12 represents the participant opinion on which factor that was most relevant for avoiding visual fatigue. It showed that texture seems like the most significant factor for the participants. Figure 13 is the summarized results from the specific questions (red representing texture as main factor). The result from figure 13 did not match the specific questions from figure 13, on which most of the participant vouched for that shape represented by red were the most important factor for achieving diversity within the scene, even though the results were quite even. Since the pictures were not marked with any indications of weather it consisted of texture or shape variation, the perception was kept honest and personal not directing the player to a forced answer. The outcome of the survey proved to be interesting in regards of indicating the most important factor for diversity within the scene. Since the majority chose the pictures containing variation in shape as the most appealing alternative, while stating that texture is the most important factor it became hard to determine the most significant factor for achieving variation. But since the majority chose shape variation as the most variative option in the pictures, one can say that for this particular scene, mesh construction is indicated to be most important for variation.

Another question that indicated to clash with the opinion that most participants regarded texture as the most important factor, was a later question that asked if repetition is avoided through larger or smaller detail.

As shown in Figure 14, most of the participants considered larger details and shapes as most important. Included in the results were the opinions that that the light differences in the scene contributed much to variation, which one could consider as part of the larger shape variations.

Figure 14:Larger detail opinion graph (Dangoor Ehnberg, 2017).

(29)

8 Discussion and Conclusion

The principle question for this thesis has been what techniques a 3D artist can use in order to avoid visual fatigue and make the scene as dynamic as possible. Focus has also been on how the designer can facilitate the process by using different techniques that makes it easier to make direct changes without having to switch between software, using modular assets. The survey worked as an indication on whether the principles behind the techniques gave matching impressions on the spectator as they were supposed to from the beginning. The survey indicated that the building technique applied to the scene worked out, since variation was achieved through both texture and shape differences.

The process of building the ferry did not only revolve around achieving diversity within the scene, but also to resemble the authentic ferry, which was a time-consuming process itself.

For future similar projects, planning is likely the key in order to be able to create a real life inspired game level. The designer has to keep many factors in mind when it comes to the planning of the scene, since building a modular level requires detailed arrangement on all levels, weather its detail or larger contexts.

So how does one achieve visual diversity within a scene? What I learned from the process of this environment creation and the included survey, was that using as many building factors as possible is beneficial for avoiding repetitiveness. In several previous projects I have

participated in, I did not take use of modular environment building, nor procedural textures.

When comparing the results and working process of this project to others, it really shows that not only workflow but also graphical quality is increased when using these types of

techniques. I would say that one of the biggest reasons why the working process was effective within the project was mainly due to the possibility to iterate the scene effectively, since it was built on modular assets and procedural textures, which allowed effective iteration, avoiding processes that forced the designer to switch between software in order to make changes. In addition to this, focus could be on detailed construction work that was needed to avoid visual repetition. I would consider modular environment creation as a tool that supports workflow but must be seen as an approach, and not as a guarantee when it comes to the quality of the design in the game.

I would consider parameterization and modularity as part of the main factors to avoid visual fatigue, since they facilitate iteration. Iteration in this project proved to be the main factor for assembling the scene and achieving visual diversity. In terms of knowing what building factor that helps the scene to appear varied I would say, based on the process, that texture variation is the most important factor in order to avoid visual fatigue. Of course, the other factors play a big part in the process, but texture variation seems, according to my experience to have the most significant impact on the result of the scene. However, the potential lies in

(30)

(31)

References

Amato, G. (2016). Creating A Basic Blend Material In Unreal Engine 4. [online]

Available at: https://www.gametextures.com/creating-a-basic-blend-material-in-unreal- engine-4/. [Accessed: 8th May 2017].

Burgess, J. (2013). Skyrim's Modular Level Design. [online] Available at:

http://blog.joelburgess.com/2013/04/skyrims-modular-level-design-gdc-2013.html.

[Accessed: 5th may 2017].

Casen, S. (2016). White Boxing Your Game. [online] Available at

:http://www.gamasutra.com/blogs/SaraCasen/20160713/276970/White_Boxing_Your_Game.

p hp.; [Accessed: 10th April 2017].

GDC. (2017). The modular leveldesign of Fallout 4- Joel Burges& Nate Purkeypile.

[Online Video]. 14 April 2016. Available from:

https://www.youtube.com/watch?v=QBAM27YbKZg. [Accessed: 15 May 2017].

GDC. (2014). How We Used Iterative Level Design to Ship Skyrim and Fallout 3.

[Online Video]. 22 January 2016. Available from:

https://www.youtube.com/watch?v=PhW8CY8XkFg. [Accessed: 2 May 2017].

Gnomon. (2016). 3D Environment Art for Video Games: Artist Panel. [Online Video]. 24 May 2016. Available from: https://www.gnomon.edu/community/events/3d-environment- art-for- video-games-artist-panel. [Accessed: 30 April 2017].

Mader, P. (2005). Creating Modular Game Art For Fast Level Design. [online] Available at:

http://www.gamasutra.com/view/feature/130885/creating_modular_game_art_for_fast_.php.

[Accessed: 23rd may 2017].

Wagner, O. (1988). Modern architecture. United States of America: Getty Center. pp. 86-90.

Watson, J . (2017). Advanced Materials for Modular Environments. . [online] Available at:

https://80.lv/articles/advanced-materials-for-modular-environments/. Last accessed.

[Accessed: 5th may 2017].

Software

Allegorithmic, Substance Designer https://www.allegorithmic.com/

Allegorithmic, Substance Painter https://www.allegorithmic.com/

Autodesk, 3Ds Max https://www.autodesk.se/

(32)

Glossary of terms

Asset

An asset is an in-game object that is used to build the environment. An asset can be a complete object including materials and other properties or either.

Bitmap

Bitmaps are digital images that can be used for the assembly of a material.

Game engine

An engine is within game creating contexts where all the different components of a game are assembled. It is a software platform designed for the creation of games.

Material

A material consists of combined textures. A material is a tool that contains information on the texture appearance. The material in the engine is a combination of different textures that determines factors such as light value, shape, and shadow projection,

Mesh

A mesh is the specification of the shape of a 3D object. A mesh is in other words an assembled 3D object that is put together using polygons.

Modular

Modular assets can quite accurately be described as puzzle pieces. Instead of assembling large objects from an external 3D program, one can construct smaller pieces that can be placed in a larger context in the main game engine. The opposite of this would be to create large objects that are completely assembled before implementation in the engine.

Procedural texture

A procedural texture is a texture that can be manipulated from non-native software, often by exposing different parameters that affect the value of certain components in the texture, such as light value, height information and color value.

Texture

The texture is the surface appearance of an object. It can be described as the paint on a sculpture. The shape of the object is the “statue” while the texture covers the outer surface of the object.

Visual fatigue

In terms of 3D modeling, visual fatigue is used for describing when objects within a 3D context are designed in a way that makes the environment repetitive. This applies to all the factors that define a 3D object, such as texture, shape, and placement. It can either be larger or smaller factors that contribute to the repetitive patterns within the game. In several cases during the thesis the word “repetitiveness” has been used as another word for visual fatigue.

White-boxing

Much like sketching an image, white-boxing could be explained as its equivalent for 3D. It is a technique that makes use of basic 3D models in order to map out a game level. It can be used for both game design and for planning the layout of the game level.

(33)

(34)

Appendix

Survey

(35)

(36)

(37)

(38)