OOF
This task is designed to teach artists how blocking out environments in 3D space can be crucial to perspective in digital concept art. The objective is to use a pre-made low-poly tiki mask as a base to create more complex shapes and designs. Similarly, isometric graph paper can also be used to create small scenes with orthographic perspectives. The next task, inserted below, is another exercise to reinforce this method of thinking.
Sci-Fi Crossover – Scrapped
Concept #1
The asset list included in task 2.3 was made up of potential objects that would tie in with the western/sci-fi crossover proposal. However, there were a lot of ideas that needed to be crossed off the list to make the idea viable for this unit, and so I thought it would be appropriate to sketch a scene that contained a large number of these items before simplifying the concept. The quick sketch above depicts a run-down unit interior made up to look like the bedroom of a desperate tenant. The walls and floor of the environment would be made up of rusting panels and lit up displays, and the room would be scattered in scrap material and western-esque junk. Vent slats to the right will let in just enough light to illuminate the dirty space. A campfire could sit in the middle, suggesting a hopeless desire for warmth, and simple game could hang from a rack above. Unfortunately, this scene contains far too much work for what’s meant to be a simple diorama so it will be scrapped, for now.
Concept #1 – Revised
This greyscale concept art took me an entire day to draw. The GIF inserted below depicts the processes in which I drew the best 2D art that’ll ever come from me. I began by combing through tonnes of screenshots from Bethesda’s latest DOOM title and 343 Industries’ Halo, to produce a corridor corner that felt right for the scene I had in mind; I used these references to create a perspective I felt comfortable drawing in. With simplistic, stylized textures in mind (most likely in spec/gloss or metal/gloss) I divided the corridor into modular pieces in preparation for sci-fi-esque paneling – I made sure to add some extruded strips for depth. While browsing concept art and key frame pieces in the past, I found that artists tend to begin with a greyscale blockout that helps define which objects are closer or better lit; this proved extremely helpful.
The balance between the sci-fi and western themes were difficult to maintain but I felt the clutter that’s implied through wanted posters and traditional, wooden containers was almost enough to meet the unit’s criteria. If you open the image above in a new window and look carefully at the posters on the wall, you’ll find that the names of the wanted characters come from a popular anime I’ve never watched.
The numbers noted in the bottom right corner are the brush widths of their respective layers in case I have to return to make changes. The wooden containers in the bottom left are made by taking a screenshot of a Sketchfab model at the right perspective to match the corridor; the fuse box and cables are taken from Borderlands 3; and the futuristic saloon-style doors are ‘borrowed’ from this artist – I plan on changing these designs to fit my own style later on. The shift key is a gift, thank you Adobe.
Western Diorama
Concept
Using the isometric graph paper provided during the task at the beginning of 3.1 – Problem Solving – I created a new design that incorporates a larger, more coherent scale. This new idea would require a more broad list of assets and materials but would better represent my own skillset, and resembles a vertical slice. However, the image depicted above is still vague and would need some more in-depth designs of each asset featured in the original sketch. The second image in the slideshow includes screenshots that I took from Red Dead Redemption 2, that I used as reference for this concept.
As a huge fan of horror games, especially ones involving religious nutbags, I really wanted to create something that looked like Chris Radsby’s submission to Artstation’s ‘Wild West Challenge’. Unfortunately, I couldn’t think of anything that strayed too far from this artist’s original concept, so I decided to create a saloon instead. However, I’ve included this piece in my concept section because I’ll still be drawing inspiration from the assets in the render above; this will include the materials. This low-saturation, dark theme is something I’d like to replicate in my own work.
Blockout
Despite my initial concept being only an isolated corner of a saloon environment, I’ve decided that I now have the time to create a small scene instead. This will consist of a broader range of assets, including a bar counter, large shelving unit behind the bar, doors, and windows. For this blockout, I’m simply re-using the window and door assets from the class saloon exterior. Most entities in this scene are just repeated modules and so it shouldn’t be much more difficult to create this than the initial corner. As you can see, I’m following Red Dead Redemption’s design (second slide), providing me with a metaphorical bar of quality to pursue and compare to later on in development. Although this idea is ambitious, if achieved, would look incredible on my portfolio. In this new plan, I should take extra care in considering the clutter that really makes this environment feel lived in. E.g. paintings, posters, misc. decor., ash trays, glasses, bottles, etc.
As suggested by Caz, during the interim 2 feedback stage, I’ve created some more unique ‘blockouts’ with a broader range of variation; these screenshots correspond more accurately with the original plan depicted in the blueprint/sketch, and have been included to produce more of a divide between my peers’ work and my own – I’ve heard a lot of people are creating saloon interiors and cowboy hats. The above slideshow includes different versions of the doors and windows. If we number each design using the first image, from left to right, I plan to use the 1st window design alongside the 2nd door design.
Test Pieces
Hat
Is it really a ‘Wild West’ project if it doesn’t include a cowboy hat? I thought not. As suggested by the array of iterations above, I began with a very low-poly and dodgy-looking base model, which grew progressively more janky as time went on; this model consists of a tall stack of non-destructive modifiers, including a few ‘edit poly’ and ‘turbosmooth’ layers. This method of modelling made it easier for me to create an organic shape without having to learn ZBrush. Considering this vertical-slice style diorama doesn’t have a strict tri-budget, I plan on baking the highest poly version down to the middle range – clearly, the first model isn’t appropriate for any game ever (lol).
Hat_LP 
Hat_Baked 
Stock Smart Mat. & Wireframe 
Stock Smart Material, Rendered
Although the difference is only very slight, I do believe the baked version does look better; the highlight/shadow along the edge is smoother (click to expand image). However, there is an odd artifact at the very tip of the hat. Hopefully this can be adjusted within Substance Painter itself by amending the normal/height information by hand. Before looking at any reference images for ‘cowboy hats’, studying the material, or even thinking about what could be added to create a more unique appearance, I’ve simply dumped the default/stock Substance Painter ‘damaged leather’ smart material onto the model to see the result. (The artifact is still clearly present because I’ve not tried to remove it yet.)
Wooden Columns
Although the scale was a little difficult at first, this design was simple to create and took approximately 30 minutes to complete from start to finish. It began looking more like the vertical pillars of a banister, or newel, and wouldn’t fit the environment I had envisioned. After some research into Georgian interior architecture, I then continued to add features that made it look more sturdy and individual. These will not only be used to support the ceiling within the room, but also implemented to divide up the walls, and add further variation to the modular design. It should be noted that the hooks will be exported separately so that they can be added when needed. The slideshow inserted below is essentially a blockout created to represent how the columns will be used to break up the walls; although the wooden panels have been modelled in the example below, I may toy around with the idea of using Substance Designer entirely.
SPOILER ALERT! None of this process is linear.
Texturing the columns was pretty straightforward, beginning with the stock smart material ‘Wood Walnut’. I then adjusted its base colour and metal/rough to roughly match the other wood assets in the scene (see Back to Modelling for details). It should be noted that the normals in the viewport are always far sharper than they really are when rendered. Lastly, I used the edge wear generator and birch wood stock material to create a worn feel. After toggling the sRGB/colour settings of the metal/rough maps, the column begins to appear as it should. Odd shadows are present because lighting hasn’t been adjusted yet. I’ll also experiment with the desaturation node in the material editor because the base colour is looking a lot more vibrant in-engine.
Windows
I don’t believe there’s much to be said about the windows. The image inserted above shows the 2 designs I ended up experimenting with the most. These ideas were taken from photos of 19th century American architecture so that my scene could be as historically accurate as possible. One opens in a sliding motion while the other doesn’t open at all. Personally, I prefer the former design. It should be noted that I had to split the wall planes using a compound boolean object so that the windows and doors don’t clip into the paneling of the walls.
Using the boolean compound shape feature, I cut a window-sized shape into the wall plane (see ‘back to modelling’ after ‘materials’ for creation of the wall plane) and seperated the UVs into 2 texture sets by rearranging their material IDs. This will allow me to use the original wall module material on the new model without requiring any changes, as seen by the application of the original height map in the viewport. To get this process to work correctly, I must apply a new material or multi/sub material to the object before exporting.
Doors
Similarly, there is not too much to say about the doors either. I experimented with 3 final designs, as shown by the image inserted above. All 3 consist of a different pattern of paneling, and two handle variations, ranging in complexity. Although more complicated patterns can provide more depth to a scene, I personally prefer the simplest on on the far right; and I will probably pair this design with the first, sliding window from the previous section. From here, I can make the scene bigger by repeating the final assets. Afterward, I will experiment with clutter to make the environment feel more lived in.
😡
Creating the Window View
Because I want the scene to look as realistic as I can get it, I should begin to consider how transparent the windows are, and what I’m going to do about the ‘outside’ that can be seen from them. Here I have a few options to take. The first is to make the windows entirely reflective so that the player cannot see outside, giving me a lot less work to do; however, I find this option cheap and unrealistic, especially considering it’s such a small environment to begin with. The next would be to create a small, fake sky box around the outside of the window – essentially a cube with flipped normals and a texture painted across it to give the illusion of an exterior environment; but without proper execution, this method could create an unprofessional result. Thirdly, similar to Insomniac’s Spider-man (PS4), I could study and experiment with cube maps; produced properly, this technique can have some incredible effects that both improve performance, and look stunning. Finally, I could just create a small outdoor environment. Using my knowledge of Unreal Engine, I could create a landscape consisting of painted and animated foliage, as well as a cheap building to block the majority of the view – maybe the log cabin I created for 3D art classes.
Wall-Mounted Light Fixture
This part is pretty simple, really. Which is great because I had to do it twice after a flash drive meltdown. Image 1 depicts my low poly mesh while image 2 shows my high poly mesh. I’ll probably go back and make the flair at the top a little less prominent, but maybe not, we’ll see. Baked down and textured at 2k in Painter using the new ‘Bronze Statue’ smart material and custom mask editor (generator), and the normal map information for this ‘Lava Glass Rock‘ base material on Substance Share, by Marcus Ryell. I slapped a relevant normal-decal on the flat surface of the mount for added detail; I thought anymore than that would be too much and also a waste of time. I was going to upload this asset to Sketchfab as an individual piece because I was proud of the texturing work, and I needed to add to my Unit 11 tally, but it just didn’t look right inside of the browser window.
In addition, I had issues exporting all necessary texture maps for this piece (and Rob told me to include this, lol) – despite all channels options being correct inside the custom present menu, all additional maps on top of standard PBR exports were being cancelled.
IMG #1 – Flair, IMG #2 – Little Flair
Ram Skull
After looking through a collection of reference images and browsing a few photogrammetry scans on Sketchfab, I began to model the skull of a ram native to the US. I did my best to create a mesh with topology that flowed correctly while also being aware of the tri-budget and level of detail required for its placement inside the scene; the back of the skull is made up of a lot less polygons than the front because it will not be seen in any detail. Image 1: low poly mesh w/ wireframe, 2: low poly mesh, 3: high poly mesh.
Table
This table is probably my most successful asset, having only taken 30 minutes to make in total, and being so effective with its baking. I used a smart material I had originally created for a personal project over the summer, which didn’t require any adjusting once applied to the mesh in Painter. Although it could definitely be improved with a different design and less wear on the bottom half, I was honestly surprised how well it turned out in the end. Before I publish these assets on Sketchfab, I’ll edit the underside to seem more realistic in terms of support structures. I only encountered 1 issue during this process with the normals of some faces in the HP catching the light so strangely. This was resolved by simply deleting some faces and re-capping the borders. Some are still present on the underside.
Chair
To help fill out the scene some more, I brought in a simple asset that I created for a personal project over the summer. However, it’s geometry didn’t suit my new style of work and so I made sure to trim in down for the LP and create a copy that I then continued to edit using my new knowledge of this pipeline. “Creating a AAA game asset” by Emiel Sleegers on Artstation Learning has taught me tonnes and I’ve not yet finished chapter 1 of 8. As you can see, I used the same smart material from the table on this chair so that they’d match inside UE. I experimented with adjusting the colour and saturation of the base to add some slight variation between the two assets.
Materials
Wooden Planks
As stated in the proposal’s overview, I plan on creating realistic PBR materials for the western diorama, beginning with some detailed, worn wooden planks, with a dynamic paint mask that can be exposed and adjusted as and when needed. Despite having the height information for the bare planks already done, it’s proving difficult to colour the material so that it retains the details I’ve created. On top of that, the paint mask has been made and the slider works great, but the colour and roughness detail is a slow process. I followed Jacob Norris’ ‘Painted Wood Planks’ tutorial to create this material.
The screenshots above were taken from the final material graph, and each section is labelled according to the content’s purpose. Most materials begin with a its own kind of blockout, using height information to create general shapes and patterns. Above, the first 3 images depict just that process, resulting in a tiled and offset wood plank pattern – variation in height is added using the greyscale flood-fill node (“Tilted Planks”), and further randomised by imposing a warped cells noise over the new height information (“Wood Wobbles”). The last 2 images in the slideshow above show the processes used to create natural-looking wood grain and knots. Each section is then blended into a series of blend nodes in the middle of the graph. Despite being very messy, all required nodes for futher adjustment are available in convenient places for reference later on.
The next three screenshots inserted above show how I masked off the ends of each board (“Create Wood Ends Mask”) and used that mask to create damage (“Wood End Wear”). I used this same mask to place the nails in the appropriate place later on (“Create Nails”) – I found this trick very quick and easy to use, and I believe it could be used for other ideas in the future.
My personal favourite part of the entire graph is the mask I created to represent paint coverage across the set of planks, because the position can be exposed to give artist’s full creative range for how much paint is present. However, the most difficult part of creating this material has been adjusting the structure of the graph to suit this concept. Where I’d normal start a material with the height information, which is then fed through a normal, curvature smooth, and HBAO for all corresponding texture maps, I had to readjust this process to incorporate the new function; the paint coverage mask had to be inserted into each section for each texture map so that I could change the colour/metal/rough according to how much paint was present. For example, because paint sinks into the grain of the wood, the painted parts of the material should appear more smooth, and so the height information should be adjusted in the appropriate areas accordingly. There’s a lot more to talk about for this material, but I only need a pass.
Wallpaper
To reduce the amount of stress placed on my hardware, I’ve decided to divide this material up into several graphs. The first comprises of a bitmap image imported into Designer so that I don’t have to create these complex shapes from scratch. After studying a small range of popular wallpaper designs from the early 20th century (US), I drew a pattern in Photoshop and exported it for use in Designer. The 2D viewport above shows that even the background colour in the wallpaper is filled with subtle noise. From here, I will focus on a damage generator, and a plaster material to place beneath it.
Wall Module
By introducing the wooden planks material to the wall material, I was able to create a wall module, as depicting in my concept sketches with vertical wooden slats. Although it’s far from perfect, I’m happy to use it for this set of units. In the future, I’d like to work a little longer on the differentiation between the plaster and wallpaper, as well as blending the trim with the original wood style. For now, it’s good enough for me. I’ve made sure to expose the histogram range that determines the amount of damage on the walls, so that I can create more variation in my scene. The plan is to create two instances of the material in which one is damaged and the other pristine, so that I can vertex paint the walls to prevent poor tiling.
This material is absolutely the worst organised thing I’ve ever created and I don’t plan on cleaning it up until I decide to either re-use it later on in the year, or sell it online; it’s made up of 6 individual graphs and the “Wood_Planks” graph is the biggest mess you’ll ever not see, because I’m not showing you (jk). “FINAL_Blend” is a graph dedicated to blending all the final pieces together to create the result you see in the slideshow toward the beginning and end of this section: in it, the materials in the previous graphs are transformed, tiled, and masked to fit together; the damage mask generator is also present here, and is exposed for use outside of Designer. Next, “Wood_Planks” is the initial graph created to produce the floorboards. Then “Wallpaper_1” is simply the wallpaper material touched on in the section above. “Wood_Element” 1 & 2 represent the coving and trim between the wallpaper and planks. And finally, “Wall_1” contains the processes used to create the plaster underneath the wallpaper. In addition, the “Resources” folder consists of one bitmap image used for the pattern of the wallpaper, mentioned in the section above. The slideshow below is made up of screenshots of each graph, in order, minus the wooden planks (because it crashes 9/10).
Back to Modelling
Most game dev. pipelines are non-linear, so here we go.
The second image in the slideshow above is a screenshot taken from the games engine, where I’ve applied the wall module material to a subdivided plane and tessellated it further so that the height information is as accurate as possible; this is a terrible idea for the games industry because it’s just so expensive. To reduce the tri count for video-games, I’ve taken the material height map into 3dsmax and have modelled according the the height information, essentially retopologizing the material and negating the need for a height map. The results have been inserted below – the left side is 200 tris, the right is so dense you cannot see the model. This is something I can change later, for beauty renders and portfolio presentation, but the simplified geometry is best for sites such as Sketchfab.
I honestly didn’t expect this method to work but the screenshots below prove that this process is both efficient and effective. These images only contain the new, simplified modules. #1 – The result. #2 – Wireframe view in-engine. #3 – The 2 meshes in 3dsmax.
Avoiding Stress
To avoid the stress of creating a whole environment for the college deadline, I’m going to create a static, singular object that looks more like the standard diorama; this should hopefully reduce the workload as well as make the scene more suitable for platforms such as Sketchfab. However, I will be working on a portion of this environment to the best quality I’m currently capable of for my Artstation portfolio over the course of the year as part of Unit 11. For now, I turn back to my initial design (above). The second image is a screenshot of ‘Summer Break’ by yunova, which was recently featured on the official Sketchfab Instagram account; I’ve included it because I’d like to replicate the simple border around the outside of the corner.
Not Avoiding Stress
The slideshow below depicts the stages of rushing the lighting and smaller assets within the space of 48 hours. Here, I begin to light the scene properly and create smaller assets to fill the table top. More of this process can be viewed in the development log, including bugs and solutions I found along the way. Within this time frame, I managed to add playing cards, a switchblade, poker chips, and water mark decals.
Level 3 Year 2 