Here is the breakdown of the Chasmfiend, a real-time creature made from the concept Chasmfiend - Stormlight Archive by Regin Wellander.
When preparing to make this creature, I had main goals, which was time. I tried to force myself to make this project in a timeline that would reflect what it would be like to make it in a studio setting, so 80 hours for the high poly sculpt, UVs and bake and then 40 for the texturing and shaders. I somewhat managed to fulfill that with a tiny bit of overtime, making it in 98 hours for the high poly/UVs/bake and then completing the texturing by hour 127. Considering this was the first attempt at limiting the time to create a project, and that my initial estimation would be closer to 300 hours… this was a solid attempt!
I had 4 weeks to complete this project, and as such I tried to create a schedule where the high poly would be made over 2 weeks and then a week for the retopology/UVs and another for the texturing. This was also the first instance of me trying to regulate my working hours into something more professional. I rarely got to 8 hours a day, but it is a work in progress, and so I am hopeful to achieve a proper work routine with a bit more effort.
When approaching this project, it was evident that every leg couldn’t be unique, both because of the time restraints and optimization. Kit-bashes are a popular option when creating an environment for games, and as such, because of the way the creature was concepted, it seemed like the best workflow to follow.
I had never followed such a workflow as a character artist, so many things were figured out on the spot. Early on, the modules were chosen, and quick tests confirmed that they would probably work.
My biggest worry early on was that it would be too easy to tell the individual pieces, especially with the texturing and details. In that scenario, I had a plan to use vertex color in unreal with a shader that had some slight color/roughness variation, but in the end I didn’t feel the need for it.
How to place the modules would also be imperative to breaking the repetitiveness they had, especially for the scales. The legs and tail tip scales that just had to be scaled and placed could easily be done in 3ds max, but for the rest of the scales and the pincer legs, I went with Zbrush, so the editing would be easier and the result better.
When sculpting the meshes, I had a big problem with adapting the concept. Something that is beautiful in the concept will not always render well in 3d or be good for gameplay. I also just don’t have the level of skill of the concept artists, so I had to try to find some way of having a good result.
In this case, the solution was nature. This Moroccan crab had similar features and coloring to the Chasmfiend, so I tried to study the shape languages and replicate them.
My biggest struggle for this project was the texturing. I went through a bunch of iterations and in the end I’m not the happiest of the final result. It works, it wouldn’t have gameplay visibility issues, but it did lose a lot of the charm the concept had.
The texturing fell short of what I was hoping to achieve, especially realism wise. But it is also the first sea creature I’d ever made, and this just showed that it is something that I need to work on more during the r&d of projects, as well as flat texturing skills.
Since there were 7 texture sets for this creature, I chose to follow the UDIM workflow in painter. That way, the organic materials that replicate through multiple texture sets wouldn’t need to be copied and pasted constantly. The only fallback I found is the need for renaming the outputted textures at the end, since they were not all the same resolution.
Some values needed specific control in unreal, such as the SSS amount. Since it was important for it to be specific, I chose to control it through the emissive pass and export it through my MRA alpha pass.
Something that was also important for this project was getting the scale. At first, I was hoping to achieve the same grandeur as in the concept, but unfortunately it wouldn’t be as convincing as reducing the scale and treating it like a Monster Hunter creature. So the scale was reduced and to test it, I interacted with it ingame in UE5. Walked around it, saw the points of interest and how a player would take down this creature.
The goal is a game model, therefore I interacted with it like a gamer would, or as close as possible to it.
For posing the creature, I used the rig made by dear friend Abhinav CHalla.
In the end, what I learned most about this project is listening to feedback when it’s given. Someone gave me the same feedback for a solid 5 iterations, and I didn’t listen to it till dozens of hours had been spent on the texturing… and it’s fine this time since this was on my own time, but I now know this is something I need to focus on in my next projects.
I am immensely grateful for all the help that was given while making this project, even though sometimes it had to be bashed into my head a little before I understood it. I am proud I managed to accomplish this, and I am hoping this impresses the people it needs to so I can get a job I really want.