For this project, since I escaped the soft fleshy bits of character sculpting, a lot of my references are just trying to understand the shape language that Horizon Forbidden West uses for their warriors, in this case the Tenakth Marshalls.
(I dont have the images at the time of writing this blog post but I also had both the Horizon Forbidden West and Zero Dawn art books physically infront of me that I was referencing; I will add them ASAP.)
Clean versus annotated breakdown
I was able to find a free model of Aloy in one of the Tenakth armor sets to break apart and see how they do everything, however I will mention it is the wrong clan since the designers of this game are incredibly cool and made each subsect of the Tenakth -- Desert, Sky, and Lowland -- all have different designs in their specific armor types
This is the Aloy Tenakth Recon Outfit, extracted by Tosyk, my third favourite set in the game. There is a pretty important note that "if I have this model, surely someone must have extracted Kotallo's model, especially since he's one of the most beloved characters in this specific game." I thought so too; it turns out extracting models from HFW is incredibly difficult.
One aspect I immediately learned from this model is that there's actually an entire texture map that is just an atlas of rope. That is so incredibly cool and isn't appreciated enough. I also stole Aloys' face for a mannequin to make sure proportions are correct.
Also, Aloy has baby hairs on her face. I love these artists so much.
Back on topic, working on this project is something I've wanted to do for an extremely long time. I played HFW right when I got into the major, and every couple of years I end up right back at it's development process. In 2022, when I got my Blizzard scholarship and got access to the GDC Vault I watched the Guerrilla Games presentations over and over and over again. Specifically these three come to mind a lot while working on this project.
Although im not happy with the general blockout, I am really excited on the experimentation for the hair cards. In Common Art I've gotten the opportunity to work with and learn new workflows for creating afro-textured hair for one of our main characters. I believe I can apply some of the same logic to here. Although this is most likely just utilizing curves rather than the volume-to-curve pipeline we're using in CA
CA pipeline uses a volume -> proxy hair cards or mesh -> curves -> final cards pipline which allows us to keep everything editable while visualising the end product.
For Kotallo's headpiece it would likely just be focusing on the curves
permission to create her was obtained before the project began :)
Schedule
[05/19] BLOCKOUT [✔] -- Overall blockout complete; some secondary forms emerge, but the overall flow of the piece is present. Polypaint used to differentiate material types
[05/26] SCULPTING [✔] -- Detail sculpting completed. All objects have secondary and tertiary details present.
[06/02] RETOPO & UVS [✔] -- Retopologizing the model down to a goal of 30k tris, following reference from Overwatch 1's character models (Source).
[06/09] TEXTURING & RENDERING [✔] -- Use generator and mask-based texturing workflows rather than hand painting due to time constraints. Render in Marmoset Toolbag.
Shortfalls
Oh boy, due to a tremendous amount of life issues, this project did not go well. I genuinely believe that my absolute love of the character and consistent Dungeons and Dragons nights—where we actually play with this character—are the only reason I pushed through and finished her.
That being said, there are many, many, many things I already plan on going back to fix and tweak, one of the biggest being the faceting of the basic mesh. It's especially noticeable in the lit versions of the renders, but some areas of the hair and sleeves have significant faceting due to hardened edges.
Another fix to make is in regard to the gauntlets, they're just... bad. overall. I ran out of time making the gorget so the gauntlets never really made it out of the blockout stage. even in the final version, they kind of just have some gold thrown on there to make them seem like they belong.
A little bit silly of a fix that was intended for optimization was the hair tubes being uncapped at the end and entirely open. This was the first time I took a sculpt like this through my process, so to avoid hurting myself with around 30 pointed hair tubes, I ended up clipping the end off. The intention was to go back later and merge the verts together so the UVs would pinch closed.
If I were to get a grade for this project, Id honestly expect something like a low 80. Although it is done, it did not meet all of the requirements that I started with, including polycount or even the original texturing method. There were too many corners cut to justify a higher grade; however, I know I am an extremely harsh critic of my own abilities.
Extra Renders
At some point early in development, I got the idea to render the final version in Unreal Engine. However, due to time constraints and my general dislike of Unreal's rendering setup compared to Marmoset, I went back to what I knew best. However I did make some small things to... encourage me, including a plug-and-play master material for the maps so I didnt have to keep setting them up individually for all my 3D projects
And adapting a Regradable VFX for a sunspot to create an evil darklord power for Ireena to wield. for some reason.
Very basic rig for posing
Process Shots:
Comparing the different stages of texturing, I'm someone of a snob when it comes to hand-painting software, and I really don't like any of them, so I used a combination of Substance Painter, Clip Studio Paint, and 3DCoat to achieve the final product.
(1) Substance Painter was used to lay down the initial flats, using a smart material called Simple Diffuse to streamline some of the tedious processes of laying down flats, such as setting up a light and applying ambient occlusion & curvature.
(2) Then I transitioned to Clip Studio Paint, using their rather unknown and old software, Clip Studio Modeler, to outline Ireena's face and apply gradient maps. Substance Painter and 3D Coat both have their own quirks that make it difficult to become comfortable in for me, so I used CSP to align levels and 'start fresh' when blending modes were behaving oddly or I needed to focus on small render areas.
(3) Majority of the painting was done in 3DCoat since it was a program I wanted to learn and eventually became comfortable to work in. As soon as I realized I could lock meshes and the rough way the hierarchy worked, it became a lot easier.
I transitioned away from the Overwatch style of rendering in favor of my own personal style, mixed with a cheeky bit of Arcane, due to my friend group having a big focus on 2D art recently. Two of my friends are Concept Artists who have been working through commissions, as such leading to many discussions on art practices, one of them being a notion that 3D artists don't use gradient maps. As an example,
This is one of my close friends Shayna Chevallier/Usagi.Bunbun who heavily utilizes gradient maps in her work to create unique colour combinations.
For the shirt and the gold I used a combination of two different maps, one affecting the shadows and one affecting the highlights; that way the colours could be unique between them,
This sprint, I mainly focused on retopologizing and UV-ing the Barista character. It currently has 132k tris, including the proxy head.
References:
Perforce:
String
The current project im working on is regarding making a proper string model to be used in conjunction with the rig. Using the notes from Jocelyn as follows, I made a quick mockup and test of the string ball, as seen below.
Since I'm following a single illustrative interpretation of this character, a lot of the focus was put into shape definition and anatomical structuring, particularly on the glabella (skin between eyebrows) and quadratus labii inferioris (muscles on the corners of the mouth)
permission to create her was obtained before the project began :)
Shortfalls:
Schedule
[05/19] BLOCKOUT [✔] -- Overall blockout complete; some secondary forms emerge, but the overall flow of the piece is present. Polypaint used to differentiate material types
[05/26] SCULPTING [✔] -- Detail sculpting completed. All objects have secondary and tertiary details present.
[06/02] RETOPO & UVS [X] -- Retopologizing the model down to a goal of 30k tris, following reference from Overwatch 1's character models (Source).
[06/09] TEXTURING & RENDERING -- Use generator and mask-based texturing workflows rather than handpainting due to time constraints. Render in Marmoset Toolbag.
Since I'm following a single illustrative interpretation of this character, a lot of the focus was put into shape definition and anatomical structuring, particularly on the glabella (skin between eyebrows) and quadratus labii inferioris (muscles on the corners of the mouth)
permission to create her was obtained before the project began :)
Shortfalls:
For this week, the main goal was to get all of the sculpting finalized to continue onto the retopologizing in the later weeks. However, I didn't get all the sculpting done, specifically on the gauntlets. The gauntlets use the same design as the gorget, previously described as the 'breastplate' or 'chestpiece' in previous weeks, which ended up being an extremely tedious process of breaking Maya.
The process I used for making the gorget involved modeling the key shapes onto a plane, focusing on the base glass, shapes then expanding outwards to create the frame. Originally, this was going to be achieved with alphas; however, due to the complexity of the gothic design and thistalk at the 2023 Zbrush summit from Blizzard's Overwatch team, I decided to continue the design in Maya
The gauntlets bring their own challenges in the form of their complexity compared to the gorget; the shapes have very little breathing room against the other stained glass pieces, requiring them to be a tight fit.
This method led to its own downfalls, mainly there being an issue with the inner corner of the shape frames, the yellow polygroup, not going far enough in the geometry, causing the glass shape underneath to be fully seen
Closeup of the gorget
Same image above with polygroups
Another main issue I had was in regard to matching the reference image, because of the nature of posing and stylization of the drawing. Hegval notes the style is reminiscent of Hades, a lot of the anatomy works but is seen at a much wider focal length, bringing many of the details much closer together than I'm able to achieve inside of ZBrush. This could be a case of 'illustrator anatomy' versus realistic anatomy; however, I attempted to follow the reference as close as possible to the sections I could.
A common question I got when making this sculpt was regarding the use of orthographic versus perspective view mode. I feel somehow the issue above is related to the nature at which Ireena is posed; however, I don't know of a specific fix for this.
Orthographic view of Ireena versus Perspective
Resources:
I mentioned these resources last week, however I rewatched them over the course of working on this project and I really think they significantly improved the workflow and direction of this project. Being able to peel back the overall design concepts, both in a professional and casual setting, such as TBSkyen for League, and understand how and why characters are designed a certain way adds a whole new level to the creation process.
It is mentioned in one of the videos above, but Renaud Galand's gumroad has a fantastic free hair brush and matcap/material library that has fit in perfectly with my workflows!
[05/19] BLOCKOUT [✔] -- Overall blockout complete; some secondary forms emerge, but the overall flow of the piece is present. Polypaint used to differentiate material types
[05/26] SCULPTING [✔] -- Detail sculpting completed. All objects have secondary and tertiary details present.
[06/02] RETOPO & UVS -- Retopologizing the model down to a goal of 30k tris, following reference from Overwatch 1's character models (Source).
[06/09] TEXTURING & RENDERING -- Use generator and mask-based texturing workflows rather than handpainting due to time constraints. Render in Marmoset Toolbag.
*During the retopology phase I'll be looking into different methods to work through the gauntlets and potentially redoing the gorget. In the meantime, they will proceed as normal until an alternative can be found.
