I created a schedule in Google Calendar. This is an overview of my week, it's not entirely precise. I usually don't use Calander to plan out my day, it's mostly weekly reminders of other stuff. I prefer following deadlines or milestones, as I feel like there will always be some changes if I were to create a weekly schedule. I use Trello checklist for all my module work and I write down my daily schedule on paper as I focus better.

I will take breaks in-between and do other activities to get some rest from work. Due to the pandemic, I feel like I will have more time working on the project. The blue section is this module, I expect to spend at least 20-25 hours a week on GAMA3000. There might be more time to work on the project since GAMA3005 relates to this project. I do work overtime sometimes, which is during my free time. I don't mind, as I enjoy working on this project and do focus better on the evenings sometimes.

1. Issue crash in all software - Too high poly

When attempting to bake the object, the loading time would take extremely long time. Before that I had issues exporting as an OBJ in Zbrush and Maya due too high poly count. I realised I had some unnecessary meshes I subdivided, these meshes wasn't sculpted at all. So, I reduced the meshes to subdivision level to 1 and exported four subtools as OBJ to prevent the crashing. I basically divided the Dance Dance machine into four parts to minimize the memory and faster loading time. I already came across a lot of crashing when importing those four OBJ files to Maya. I kept reducing the polygons by using the reduce tool in Maya 50% repeatedly without losing any details. I deleted a couple of the high poly meshes I didn't sculpt, I replace the high poly models with the low poly models. It also had better results as it had smooth edges and didn’t look any different.

It was quite a long process to fix it, as the high poly models would slow down my process and loading time. At this point I learnt not subdivide meshes I won’t sculpt and at least keep the subdivisions low at all time, as one mesh can take up to 500 000 polygons. This was also a larger asset with various subtools, it's my first time working with a larger asset so it was expected I would come across some issues. This is something to keep in mind when I develop the arcade machines later.

I decided to do some reading about baking in Substance Painter as I want to get a better understanding of the different settings in the baking window. Since I don't know much about the frontal distance and rear distance. I have the platform on the Dance Dance Machine I want to bake into the low poly but couldn't get those details appearing. Knowing the value and the purpose of the settings I don't have to spend as much time re-baking all the time.

I also looked at David Masana's understanding and process of baking in Substance Painter and tried his baking settings. The frontal distance is the external cage, meaning how far I want it to be baked. Whilst, rear distance is the internal cage, this decides where the ray will stop. Ray is the distance between the frontal distance and rear distance. [3] [5] In this case, I would want the rear distance to have a slightly higher value to get the high poly details on the platform to be baked onto the low poly platform.

Alphas

I will be using alphas to add further details, I will create my own alphas in Photoshop and use the alphas in Substance Painter. Alpha map is a black and white image which is stamped as a height map. The black and white will add shadows and sort of create an illusion of the details, but the surface is actually all flat. This will be used on smaller details that doesn't require any modelling since it will take up too much poly count. And most of the time, the players won't even notice the smaller details. Alphas is going to be the key to get the final and realistic results as the reference images, I will be using alpha bolts, circle dots for the speakers, different shapes of geometries etc to get the final results.

Sculpting - Zbrush crashed way too many times since I had so many subtools I subdivided. I should have ONLY subdvided subtools I sculpted. The smaller details should have been saved for texturing, to add details using alphas, the reason is because smaller details require more polygons for sculpting. I felt like I wasted a lot of time sculpting since there weren't a lot of difference after baking or the silhouette. Therefore, for the next asset, I will only sculpt the sides of the arcade cabinet since its larger details. The gasmask is organic, so that will be sculpted to add further details.

Baking - The high poly model caused a lot of issues, importing and exporting in Zbrush and Maya was a struggle. I had to export the subtools individually in Zbrush and reduce the polycount in Maya, I replaced some of the high poly with the low poly meshes as it looked better and lowered the poly count. I baked it in Substance Painter, the mesh ended up with bleeding issues, I had to remove those manually which was way too time consuming. Next time, I will use Marmoset for baking as I can visually see the changes I make on the rear/frontal distance. This will save a lot of time since I can skip the manual adjustment.

Texturing – Texturing is one of my weaker areas I wanted to improve on. I have learnt a lot using Substance Painter, using different layers, smart materials etc. I had a good plan, using Color ID, allowing me to work more efficiently. I really enjoyed texturing and will keep the same process here for the other assets. I'm very satisfied with the results as I focused on the details a lot and this is something I will showcase in my portfolio.

Keeping track of my schedule, I'm currently working on the 2/8 priority asset which are the arcade cabinets, pinball machines and gasmask.

Basing it on my current process for the Arcade Cabinet - Donkey Kong, the sculpting and baking took around 1 week. Due to my analysis of the Dance Machine, I didn't sculpt everything for the arcade cabinet which saved a lot of time. The other parts will be textured, as its smaller details which wouldn't require sculpting as it would be way too high poly.

The UV was already done in December, allowing me to jump straight into the rest of the pipeline. I might need more time to work on the assets, as I only have three weeks to finish the rest of the 6 priority assets. I'm not worried about it, I will spend more time on this module during progress week. The Dance machine took a lot of time to create since I had a lot of issues and adapt to.

The pipeline for the rest of the high priority assets will be: Modelling (Maya) > Sculpting (Zbrush) > Bake (Marmoset) > Texturing (Substance Painter). It's similar to the dance machine besides baking will be done in Marmoset instead of Substance Painter. The Arcade Cabinets section will consist of all arcade cabinets since it's the same process for all of them.

Why am I using Marmoset for baking instead of Substance Painter? [17]

The reason I will be using Marmoset for baking is because it previews the changes I do with the rear distance. The application also allows you to do some offset and skew painting to make any changes and bakes very quickly. My experience baking in Substance Painter for the Dance Machine was very time consuming, I ended up with a lot bleeding which I had to remove manually. Hopefully, using Marmoset will help me skip that step.

I haven't done any baking in Marmoset befure before, although I have seen the process Chris has showed me and it seemed simple. I have used Marmoset for rendering, so learning to bake in Marmoset shouldn't be an issue.

Formative Review - 8/3/2021

For my formative review, I have been getting some useful information from Chris. I have organised my folders and renamed my meshes and materials. Other than that, I had some issues with my texel density for my modular assets Chris went through.

Grids/units

I didn't use the Maya grids, this could cause me problems when I work in Unreal Engine, I had uneven measurements. The reason I should work with grids and units is because it will make my workflow easier and cause less issues. Especially, when I work in the game industry, as it should be easier for the artists to understand the assets and snap to grids without a problem.

Since I'm going for a first person in this scene, I need to think about the texel density for my assets. Having the wrong texel density can make my textures blurry. I did have a look into texel density in GAMA3013, but I didn't have a very deep understanding of how texel density worked. Besides that, I should keep the texel density consistent for all UV shells.

While writing this, I have already finished 50% of the high priority assets, I didn't really think about the texel density but used a 4K texture. But I made sure the texel density would stay consistent, and eventually scale up smaller UV shells to prevent blurriness. I wish read more about texel density as I have finished some assets already. However, I will be more careful now when working with assets.

What I learnt and could have done better

Baking - I learnt that, it's not necessary to bake all my high priority assets and if the silhouette doesn't changes, it's not really necessary to bake it. I can use alphas instead, which will save me some time. Since the camera is far away from the assets, the details won't be obvious. But as artists, we tend to be very specific with the details even though the players won't pay attention to it.

Smart materials - repetitive assets - The pipeline was repetitive, as it's similar models but different shapes. To make my process quicker, I created my own smart materials by selecting the folder in my layers window. Creating smart materials saves the materials and layers I have been working on. As example, when I texture my next arcade cabinet I can import the saved smart material which will transfer the previous layers I have worked on my previous arcade cabinet. This saved me a lot of time, and allows me to do some custom changes to prevent the same pattern on the textures. However, I wish I knew about it earlier, this was something I started doing towards the end.

Behind schedule

As I mentioned, I'm behind schedule. The reason is due to bad planning when it came to the asset pipelines. I shouldn't have created all high poly for the arcade machines at the same time, I realised it wasn't necessary. Instead, I should have started with the first asset and tried the pipeline, from there I could have reflected my process. And asked myself, if following all those steps for my pipeline would be necessary. This is what I did when I was working on the Donkey Kong arcade cabinet. This is something I will keep in mind for future projects, although I'm pleased with the final meshes and renders.

The machine cover will be in the far back room, which won't make it very visible in the scene. It will mostly be the silhouette showing. I also found Marvelous Designer a great software for cloths and fabrics, and I will use this in the future. It's not often studious require Marvelous Designer from environment artist, but this could be a bonus or skill to add for the resume.

The material appears less glossy in the scene due to the shadows, but there will also be another machine cover in the arcade room. This will show more details due to the broken ceiling. I also saved the texturing I did in Substance Painter as a smart material which can be applied on the next machine cover. I do think the machine cover came out well, I could have added more folds but considering it will be further away, I didn't want to spend too much time on every single detail.

Quixel & Gumroad

Quixel and Gumroad offer free assets you can download to add into your scene. This method can be useful if I have limitations with time and to speed up my project. I haven't used this method before, I didn’t feel like it was necessary to use it in my other project and wasn't familiar with these websites. These assets were planned into my project later to fill out my scene and add those smaller details. As example the trash bag, having a bin without a trash bag would look weird. So, adding those details brings the asset together. And perhaps, creating my own trash bag would have taken me a couple of days as it might require sculpting to get those details.

Since I'm aiming for a realistic environment, I would have to be careful with the assets I download. It can ruin the realism if the asset is stylised as example. The assets work really well with my scene, it helps me fill in the empty areas. I used 4 free assets downloaded from gumroad and quixel, I did change the light switch and plug sockets which is shown on the low priority google slide above.

I have looked up whether game studios use free/purchased assets and apparently it is very common. Considering the assets needs to work with each other first, but it happens the assets needs some adjustment to make them work properly. As example the light switch and plug socket where I had to adjust the textures and UV. But there is also a budget for it as they tend to buy purchased assets. Overall, I personally find this very efficient and helpful. I would only use this for smaller assets and not hero assets.

The vertex paint was also mentioned in GAMA3005, since I created materials for the modular assets. To prevent the repetitiveness of the textures, I use vertex paint to add some further details and get the worn out walls, floor and ceiling. I have used these three tutorials above and master materials to create my own 3 blend vertex paint and a height lerp. The height lerp will be useful for the wooden floor and dirt, I want to make the dirt sink into the gaps between the planks. This makes it look a lot more natural.

I have been experimenting with the vertex paint before, so I knew about the basics. Other than that, I haven't looked into the graph editor until now. This is because I wanted to include some custom changes, the three tutorials have got some useful tools I want to put into one master material such as height lerp, parameters and connecting the material attributes. I used the Tech Art Aid (left) tutorial as my main master material, this included 3 blends and heightlerp. The two other tutorials had the parameters and material attributes I wanted to add to the main master material. Since I have used Substance Designer, master materials and material instances I found it easy to understand the nodes and connect them.

How is it used?

Besides preventing the plaster and wood floor looking repetitive, I used this method since it's used in the game industry for the same reason. Vertex paint is normally done inside the game engine by using the RGB channels. Each channel has its own material, by using the nodes in the material editor it allows me to add further options. As an example a parameter for the blend, I can adjust the value depending on how much I want materials to blend with each other, giving a softer or sharper blend. Besides that, it's a much faster method to do things inside the engine such as vertex paint. It's useful on modular assets since it contains tileable textures, it's a way better option than texturing a wall one by one in Substance Painter.

The vertex paint and height blend master material I have created can also be used as a material instance which can be re-used on multiple assets. It is a very efficient and quick method. I could also use the same master material for my future projects, meaning I won't have to create a new master material from scratch.

Whilst adjusting the lighting with the feedback I have received, I looked into 80lvl articles, videos, conferences and artstation to get a better idea of lighting. There were a couple of things on the feedback I was unsure about, such as how to use the direct and indirect lighting to draw attention to details. Besides that, I had to look into dynamic and baked lighting, and when do I use moveable, static and stationary lighting?

The videos and articles are used to help my lighting settings and improve the quality.

Is my scene static or dynamic? [28]

Considering my scene is to showcase my work, I can go with static lighting. My scene isn't playable and won't require any movement by the character as example. I explain it a bit more below.

Baked lighting vs Dynamic lighting? [28]

Baked Lighting is a cheaper option since it requires rebake each time changes are made. Baked lighting require static or stationary lighting, static lighting is lights that never change whilst stationary lighting has an animation with it such as intensity. Stationary and static can't be moved. Considering I 6 point lights for the string lights, I can't set them as stationary since maximum 4 stationary light can overlap.

Baked Lighting This is more demanding in memory but gives much better results. Baking also needs lightmaps, it is basically stored for the lighting level. The UVs cant be overlapped and must be unique, if the UVs overlap the stored shadows and light might be incorrect. The light map resolution can be adjusted, a higher light map resolution is slower to generate but gives better shadows.

Dynamic Lighting is real time, meaning it updates whenever any changes are being made. Dynamic lighting updates all the time which needs to be set as movable. In this case, I have flickering light in my scene which is dynamic lighting. In games, if it's an open world it's most likely the game will be dynamic since the player is constantly moving and might require building stuff such as Fortnite.

What is the difference between direct and indirect lighting?

Direct lighting comes from specific object or to focus on any specific areas, such as a screen. Spotlights, point lights and directional lights are a few examples that are used in my scene.

Indirect lighting focuses on larger areas giving an even smooth light. The indirect light tends to bounce on walls which gives and even smooth lighting and shadow.

How can I improve my lighting quality? [28]

I am aware of the Lighting Build Qualities as it has different options such as preview and production. Production will be my final bake as that's the best lighting I will get.

Besides that, the World Settings has a Lightmass Settings which got a couple of options to improve the lighting quality.

• Static Lighting Level Scale - Reducing the value gives higher value but increase noise, it bakes a smaller portion of the world.

• Num Indirect Lighting Bounces - How many times the light is going to bounce around, normally 3-5 is a good number. Going too high can get too expensive.

• Indirect Lighting Smoothness - Smooths out the shadow, too high value can cause bleeding and incorrect shadow. 0.7 and 1 is great value.

• Indirect Lighting Quality - Building time, by selecting preview or production the building time can be longer.

To prevent noise appearing, the Static Lighting Level Scale x Indirect Lighting Quality = 1.0. By multiplying those two and get the 1.0 results, it will reduce the noises.

• Volume Light Samples - Small spheres placed in the world and calculates the indirect lighting during the lighting build. If it's dynamic lighting, the Volumetric Light samples uses a interpolated/insert (mathematic term) the each pixel being shaded, this gives a pre-calculated indirect lighting.

What is a Light Portal? [28]

What the Light Portal does is that it makes the Lightmass come with more light rays around the area giving better light and shadows. I added a light portal for the broken ceiling since it is an open area, it is also good to use with a skylight which I have in my scene.

To briefly summarise this project, I really feel it has been a success as I have been able to achieve my goals through research, feedback, experimenting and learning lots of new stuff. I did have my weaknesses but managed to work on them during this journey. There is still a lot to improve on such as lighting, composition, materials and more. However, I can definitely say that I can take everything I have learnt from this module into my next project.