Introduction
The title of this project as proposed is "Explore the abandoned house!" This project is focused towards creating a small but highly detailed environment, with the key being detail and story. There will be a mixture of unique models with textures, and tileable materials. There will also be a focus on creating a believable scene, with a focus on achieving realism. The reason being that most game companies that I am interested in create games in this style.
The end goal of this project would be a first person playable level from UE4, as well as an asset zoo to showcase and separate the assets to be easily assessed/viewed. There will also be a collection of beauty shots throughout the scene, along with a video fly through. There will be documentation on the process of creating this environment, which is what this blog serves. Both here and on my Artstation blog will record the research, development and reflection of the project. You can find the Artstation blog by clicking here.
The technical limitations will be similar to AAA games of the PS4 and Xbox One. There will be a Texel Density of 1024 per square meter. There is no tri budget, however the tri's used should be sensible and smart. If there is a need for an asset to have a high amount of tri's there must be a good reason for being so (such as additional loops for vertex painting, or mid-poly assets. The playable level should run at a minimum of 30fps, the reason being that this will be more of a portfolio piece to show of my own artistic capabilities.
Inspiration
With the previous projects I have worked on I found that working on something smaller lets me concentrate more on the details. So when finding inspiration it didn't take long, since I had a game I have previously played in mind. The game is Metro Exodus. Now something that I haven't necessarily done before is work from someone else's concept. I've either made my own concepts or based projects off of real life locations. I really wanted to challenge my production and planning skills so I went out of my comfort zone and decided to base this project from something that I loved playing.
Now I want to specify that this project is my own re-creation/fan art of a concept piece from Metro Exodus. All the assets within the environment will be of my own creation, the only commonality will be a similar composition. There will be many changes or different directions this project will go down, as I will rely on real life reference images and my own artistic decisions. I have also reached out to 4A Games Malta from their contact page but to this date I have received no response.
This concept art piece was created by Adrzej Dybowski for 4A Games, you can find the this piece here.
Planning
The first task I did was breaking down the scene and all the assets within it.
I wrote down all the assets and colour coded some of the assets with how I may approach with creating them. I know that certain parts of this scene can be created with simple geometry and a nice tiling material, other parts would need a little more detail.
Afterwards I created a spreadsheet on all the assets and putting down all the dates of each week and figuring out what must be done. Throughout this project I will refer back to this with whats been completed, what else is left and when things should be done. Also if I have a good reason I may move or shift certain parts of the planning in the future for if I come against any unforeseen circumstances that might affect production.
Set backs
The first two weeks didn't go as planned. As you can see I put that I will work on my portfolio at the beginning of this project, this is because I had nothing new/relevant on their. With industry related events and people visiting the labs in the future, I needed to have something on my portfolio.
The first week I spent mainly on producing good renders from my Arctic Base project, a long with producing and editing a flythrough video.
The second week was similar in that I implemented the feedback I received into my Lower Gun Deck project. I composed a new game level for a better composition with the same assets, and produced a flythrough scene. I also created renders of my materials in Marmoset, and added links to my portfolio and my own branding on top of the images.
And the majority of the third week was creating a new CV and multiple cover letters to apply for jobs. Working on my portfolio could potentially help me with seeking for work early, which is important.
Production
The first thing I did in this project was to gather a collection of references for all the different elements I could find in the concept. And additional real life references that are related in either the location or items that I would like to add for narrative purposes.
This PureRef sheet will be added to during the course of this project, as I will be constantly looking for references of whatever I may be working on at that moment.
I separated the images into different clusters that share similar properties, that being either they are vegetation, or miscellaneous house items. I would also add text if I need to remember the names of certain plants, or links to videos I found. The image is quite large, and I encourage viewing it on a new tab to zoom in.
After gathering all my references I started to block out the scene. I created some simple assets in 3dsmax that would be used as the anchors for getting the sense of scale and space correct. These assets were the oven, the table and the chairs. I tried using some of the simple geometry in Unreal Engine to get some simple measurements, but then I quickly changed them into models in 3dsmax. I couldn't find any information on the measurements of Russian Izba's, but after looking at multiple pictures of people standing in the homes I made the height of the walls roughly 3m tall. With the unreal character is 180cm tall, I want the environment to feel small but not suffocating. Adding height to the interior can give the illusion of more space. I may change this later on however.
After getting in some basic models for the walls and major pieces, I worked on to block out the space. There was quite a bit of going back and forth with creating these assets, as I would have to change the size and shape of elements to better match the concept. However after closer inspection of the concept, I realised that the perspective of the building wasn't reliable. And that I should rely more on my own skill and real life references to get a better sense of scale.
After blocking out the primary and secondary elements I started to block in the tertiary assets. These are smaller models that help add detail but doesn't really change the composition of the environment much. Such as bottles or plates, the details on how the walls were constructed, and decorations such as paintings. With the nature of this environment I try to block out what I can create immediately. However I will have later passes adding more assets that will either tell a narrative or fill in more of the visual space. Something that I need to be careful of is creating too much visual noise. I should have some rest areas in the environment that will then help to draw focus to the more important details. This will also bring a balance to the environment.
Once I felt happy with the amount of assets were in the scene, I then shifted my attention to creating some of the materials that will take up a large visual space. This being the floor and stove. The way that I treat creating materials is similar to the modeling, in that I do it in passes. I feel fairly happy with the textures I have created so far, the clay doesn't look too noisy but you can see distinct features when looking closer. The floor also has the problem of repetition, but this can be sorted with vertex painting different conditions of the floor (broken, old, rotting etc).
After getting feedback from my tutors, I also changed the lighting more accordingly to the concept. My first approach to the lighting was trying to achieve that horizontal slice of light on the floor, however that led the sun position to be dawn/dusk.
If you look at the concept you can see that the outside looks more midday overcast. And some of the feedback I received was to change the focus of the lighting to be on the stove instead of the door. Which makes more sense to highlight the stove as it is essentially the visual anchor of this scene. There will also be a need to darken the lighting in the future so that it more closely resembles the lighting conditions of the concept.
Something that I'm trying to avoid is to spend too much time and attention on one singular part of the environment. And doing the work in passes definitely helps, as I can take a break from one thing and start something else. However I do try to get to certain stages before I can move on to something else. Such as having more appropriate lighting, before creating more materials that will change depending on the lighting.
Another piece of feedback that I received from my tutors was to possibly add a tree, or some sort of object that can take up a large space so as to visually tell the player where the play space ends. And also to explain why the roof of the cabin would be in such a poor condition with so many holes. Another goal was to make this element reinforce the interest/composition of the stove.
As with everything I tend to do, I started with researching what it looks like when a tree falls into a house. As you can see there is a lot of destruction, however something to note is that most of the weaker branches snap from the impact. Which means that only the larger and heavier parts of the tree will make the path of destruction. Leading to something like a vertical slice of destruction. However there are some details to consider, my cabin is created with whole wood logs stacked on each other. Which is a much more sturdier and dense compared to conventional houses seen in the images above.
Looking into the type of tree that would fall onto this home, I researched the more common tree's you can find in the Russian environment, and a type of tree that doesn't have too many thick branches. Birch is one such tree. A lot of my environment is made up of brown wood, and having the white birch will add contrast and break up the colours in the scene.
I took some screenshots of my environment and started to design this element. I originally conceptualized that the tree could be coming from the window to the right of the oven, however this could occlude the light that is highlighting the stove which is something that I didn't want. I then realized that the only place that the tree could come from is the window in the far back.
I had a hard time visualizing the tree from this angle so I decided to design from another angle.
Here I had the idea of possibly having the tree growing through a window. This isn't a great idea as it wouldn't make sense as to why a branch that searches for light would go inside a dark building. The only example that I could find of this is the image below, and you can see that the branch is growing towards the window, where there is light.
At this point I asked for some critique and help from one of my classmates.
My friend Chloe Mitchell kindly drew up this tree from two angles. This was really helpful as it allowed me to springboard into more ideas. On top of her drawing, I added some more details such as the positioning of one of the branches that points at the stove. The branch was below the horizontal ceiling beam, however this wouldn't make sense unless the branch broke through the beam. Instead I change the branch to be above the beam, and I added some more details to the structure of cabin. I also started to figure out how the nature would invade the building.
Working collaboratively is a great way of getting new perspectives on things, and seeking for feedback is something that has always helped me improve in my work.
Before I could conceptualize how the roof would break, I first had to design what the structure of the roof looked like. Again I looked at more real life references, and also some model figurines of some Izbas.
As you can see most of the Izbas have a diagonal roof, with the wooden boards/roof pieces being placed Vertically so as to suggest the water to fall off and away from the house. Another thing that I took note of is that sometimes the construction material changes when creating the triangle wall. The change is from the whole wood log cylinders, to a flat wood boards. This change would allow me to move from a cylindrical model to a flat texture, which also breaks up the repeating shape language.
The two quick sketches I did was to figure out whether to follow the concept or the real life references I found.
The first sketch is similar to the original concept, the entrance's roof is built to align with the larger roof.
The second sketch is something that I conceptualized. I wanted to break up the silhouette by having the two rooftops being at different heights, adding some asymmetry.
I decided to continue with the second sketch as it made more sense to me in that the builders would want to maximize the use of fewer materials.
This is a sketch trying to figure out the destruction of the tree on both the roof tiles and supporting beams. From my research I know that fallen trees tend to leave a clear stamp of destruction, and that the destruction will dampen where the tree is thinner. There would also be more brittle branches appearing all over the roof and surrounding environment.
With the stove taking up a large amount of space in the environment. I had to think efficiently about how I should approach creating the oven. The first idea I had was possibly sculpting the oven to add all the lumps and bumps of the clay coating. However this would mean that the whole mesh would need to be baked. I have a problem with this as it would break my texel density I have set up (1024 per 1 meter² ) since the stove takes up a space of 2 meters².
I could make an exception of the texel density, or I could even break up the stove into separate meshes and bake unique texture sheets.
However the smarter way I could approach this is making the stove a mid poly asset by weighting the vertex normals. This would make the model slightly more denser than a low poly, but this process allows my asset to be textured with tiling materials that adhere to the texel density. This is a much quicker and efficient method than needing to sculpt, bake and also uniquely texture.
There were some difficulties that I faced but I eventually solved them. The first problem was this artifact that I got from exporting the texture from ZBrush. With the export I had this fine edge artifact in the normals that outlined the whole leaf. I think the problem occurred due to not having a basic plane below the leaf, I have still yet to test this theory. I fixed my issue by painting out the artifact in Photoshop.
The other problem I had was that designer had difficulties picking up the veins of the leaf. This would be due to the large changes in height on the leaf. Something that doesn't help as well was that the vein details being quite minor in strength. I wanted the veins to be more subtle the smaller they get, but Substance Designer did not like that at all. Unfortunately I had to poly paint a mask in ZBrush.
Conclusion
After the three weeks of setbacks in the beginning of this project, I am currently behind on what I have planned to have done already. I will need to in the coming weeks work harder to catch up with where I need to be. This will be difficult as I am still learning new techniques and knowledge that I am incorporating into my work. I need to make sure that I am not rushing too much, as it will lower the quality of work in the end. However I know that once I've got my momentum going, I will be able to get back on track.
