Most of this section will be a repeat of task 3 – problem solving – because I forgot about the dev. log. 🙂
Week 7 – Concept
By the time the practical portion of Unit 9-10 rolled around, I had already begun creating the wooden planks material as I had bought a tutorial series to improve and expand my skill set. As I came to end of this series, I decided it would be a waste to not use it within this unit. Therefore my goal for week 7 was to come up with a concept to suit the material I’d worked so hard on, as well as finish said material.
I spent the majority of the first 3 weeks experimenting with different concepts, which included a few western/sci-fi crossover ideas, some of which can still be viewed during the research phase or on my artstation portfolio. E.g. weaponry, vehicles, and interior environments. However, after watching too many terrible crossover films, I decided to move away from the unpopular mix-up. Finally, I moved back to the basic and boring: a saloon interior. I’d hoped I could eventually be able to implement both my exterior from the 3D classes and the results from this project, for my portfolio. This decision was also influenced by a material I had already started prior to the unit.
This material took me a few weeks to complete to a standard I was happy with, with the help of Jacob Norris’ video tutorial series on Gumroad. The videos provided were more like overviews of the sections he’d created inside the graph, and were far from linear; this made the tutorial series difficult to follow and so I spent the majority of my time experimenting with new nodes and techniques. The graph is far from organised and is quite resource-heavy, meaning it takes my PC a while to render the final result with my remaining memory in the extreme negatives.
Week 8 – Materials
Naturally, after finishing the initial wooden planks material for the floorboards and creating a concept I was eventually happy with, I planned to create the next procedural material for the scene. And so my goal for week 8 was to create a new material that would serve as the design for the walls, as well as produce a low poly wall module that matched the material.
Now using this material, I was able to create the 19th century wallpaper and wooden paneling that can be seen repeated across the walls of the saloon. This file was split up into a selection of graphs that represented different elements of the material, which were then finally masked and blended together to produce the finished render below. I’m extra proud of this one because I didn’t use a single tutorial in the process, and is absolutely the best thing I’ve produced in the last few years. It’s also original – I’ve not seen another material quite like it, so far.
From this point, I went back to modelling the walls and floors around the material so that I wasn’t using up any pointless tessellation in UE. This process was really easy and satisfying. The only issue I ran into was forgetting to make sure the floorboards were tile-able, but this only meant going back and adjusting the height of the planks that met the edges of the plane. This is the first time I’ve used a non-linear pipeline in a project, and although it took some time to adjust to, it made the process a lot cleaner and produced a more professional result. Although these modeled tiles work great for the isolated diorama, I should consider how I’m going to add variation to the wallpaper damage for my complete environment in the future. The material may have complete damage scale and control via exposed parameters but because the texture maps take up the entire 1-t0-1 UV space, variation would be down to UV layout per modular piece. Alternatively, I could add a grid of vertexes to the flat plane of the wall so damage could be painted instead. Or lastly, I’d simply create multiple instances of the same material and interchange between two random seeds. Unfortunately, the new Substance plugin for UE doesn’t support multiple instances per import, yet.
Half Term & Week 8 – 3D Assets & Textures
My half term was carefully divided between creating the first small 3D assets for the scene, and Destiny 2 (about 15:85). My plan for this week involved bashing out the cowboy’s hat, a column design I was happy with, and some windows/door experiments.
Now that the procedural materials are finished, and I had some clean, low-poly meshes to match, I was able to move onto the smaller, more unique assets. I began with a classic cowboy hat, and used a new modelling technique to create it – one I’d never used before. This process involved beginning with an extremely low poly mesh, that would never be considered an acceptable product, and turbsmooth-ing it by an iteration of 1, then adjusting the points in the low-poly to manipulate the new object. This idea negated the need to learn the basics of ZBrush, for now. I then cleaned up the topology where needed and created a new copy that would then be turbosmooth-ed again. This high poly mesh would then be baked down onto the mid-poly in Substance Painter for a smoother AO and normal.
The columns were more simple than the hat, probably because I didn’t create a high poly for baking, which I wish I had done. I’m not very happy with the result. This could be because of the texturing or simple because I was unable to produce a high poly that looked right. Sorting out the smoothing groups of the an object like this for turbosmooth-ing was just too tedious. After putting everything together, I also found that the smooth midsection would fit the scene better if it were wider.
After browsing some screenshots of Red Dead Redemption 2 and some assets on Sketchfab, I went ahead and created some variations for the windows and doors. This was discussed in more detail during task 3 – problem solving. Although I was happy to use door design #2, I felt the window frames were too thick and difficult to texture, so I experimented with sinking the frame into the wall to make it less noticeable and adding some extrusions/intrusions to create a more seamless asset.
Week 9 – More 3D Assets & Textures
It should be mentioned again that this pipeline is far from linear, and involved a lot of going back and forth between modelling, texturing, and engine work. As a continuation from week 8, my plan for week 9 consisted of modelling, texturing, and importing more small assets, including light fixtures, a ram’s skull, and other filler items.
The light fixture was pretty simple to make. Which is great because I had to do it twice after a flash drive meltdown. I used a similar method to the process used to create the cowboy hat, which involved starting with a ridiculously low poly mesh that’s turbosmooth-ed once to create an acceptable product. The mid-poly is smoothed once , then collapsed and edited to make sure the topology flows correctly and appropriately. This is then smoothed once more for the high poly mesh, which is then baked down in Substance Painter. After texturing the model to a satisfactory quality, I realised I had missed the bulb, and had to go back into my chosen modelling package to add the missing element. As Painter refused to accept my new base mesh replacement, I needed to export the textures I had made and import them into a new file. After that, I felt the flair of the glass rim was too extreme and once again went back to 3D Studio Max to make some changes. Because I made sure to preserve my UVs carefully, I didn’t need to re-texture the new iteration. I should also note that I had issues exporting texture maps outside of the standard PBR configuration (AO & opacity) at college, as a custom preset – this could be resolved by using a different pre-existing config, or simply exporting at home. Without troubleshooting the issue, it could have been caused by the chosen file path for exportation, texture map names/titles, or even the file path of the preset (on college networks).
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. At this stage, 2 weeks remained before the final deadline. I feel this is an appropriate amount of time to prepare for uploading to Sketchfab/Artstation, as well as writing the final evaluation.
Week 10 – More 3D Assets & Textures
My plan for week 10 involves filling out any missing written work, including the problem solving, peer feedback, and development log. All of which didn’t take any time at all, so I resumed with the practical work. Nathan suggested that I fill out the empty spaces in the middle with some furniture. This week didn’t have much of a plan, which is probably why I struggled with the end of this unit most.
Although everything on this blog (bar the final presentation) is complete to a satisfactory standard, I still have yet to texture the window and ceiling light. I think I’ve just forgotten how large and important this stage of development is, and under-estimated how much time it would take me to bring everything together; I’ll make sure to consider this for the next 2 units. For now, I’ve begun to light the scene using point lights and IES profiles – this ensures the form the light takes is accurate to the shape of the source and its mesh.
At this point, it was suggested to me (by Nathan) that I should fill the scene further with smaller, more intricate props. And so I jumped back into 3dsmax and produced a few more unique assets. As you can see below, this included playing cards, a switchblade, and decals. All cards use the same material but their UVs are placed differently to create new cards. Both hands have been customized and a card back design was added to the stack. My new plan involves adding an ashtray, cigarettes, and a subtle, ambient smoke particle effect to fill the room.
There have been numerous scenarios where I have felt the need to reach out to the UE and games dev. communities on Reddit, for answers to issues and errors I’ve encountered throughout the process. This has proven to be extremely helpful and quick system that has solved multiple hurdles along the way, including opacity map resolution issues (below), and feedback on new particle systems.
As of 3 days post deadline, I think I’m done with the practical side of units 9-10. I was unsuccessful when it came to creating a rolling smoke particle system that gathers on the ceiling, but I’m satisfied with the final product. I felt if I had created an exact plan for what I wanted to produce, I’d have gotten to the result a lot sooner. However, I also think this winding tangent has helped me make a better product overall, so it’s a careful balance in the long-run. As you may notice from the slideshow below, the wall-mounted light’s textures are still displaying incorrectly, but I’ve yet to find a solution – for now, I’ll choose angles that hide them from view.
Level 3 Year 2 