Rookie Awards 2024 - Open for Entries!
Ilias Morscher - The Potion Shop Lady - Rookies 2024
Share  

Ilias Morscher - The Potion Shop Lady - Rookies 2024

by ThePolygator on 1 May 2024 for Rookie Awards 2024

Hey there, I'm Ilias Morscher, a student with a focus on character art. In this post, I'll walk you through the breakdown of my recently completed potion shop character and share insights into how I achieved the final product.

31 673 7
Round of applause for our sponsors

INTRODUCTION

Quickly about myself, I'm Ilias Morscher, currently a student at PIXL Visn Media Arts Academy in Cologne, where I am focusing on character art. With 15 months of study behind me and just 3 more months to go until graduation, I'm eagerly anticipating finding a job in the field. This project represents my first demo, recently completed as part of a larger collaboration with my colleague and good friend, Felix Riehle, who handled the environment for the character. 

IN POSE TURNTABLE

STILL RENDERS

First off, I'd like to share a few close-up shots before diving into the breakdown.

REFERENCES

This project was primarily inspired by the incredible work of the talented Rhys Griffiths, who served as the main reference throughout the creation process.

Main Reference for the Outfit

Secondary Outfit References

Head and Body References

I opted not to pursue an exact likeness sculpt because I didn't want to complicate things further. However, I did strive to capture a resemblance without getting too bogged down. Eventually, I reached a point where I was satisfied with the face and decided to move forward. Despite not focusing on a precise likeness, I believe that if you overlook the hair, there's a resemblance to the character in the images.

HEAD SCULPT

The head was the first thing I did; I sculpted it in ZBrush. It's the first time I tried the workflow of sculpting everything except the pores. I used to do them in sculpting too, but I learned from Tom Newbury that this isn't a smart thing to do. It always required using HD geometry and generally had to go way too high with the polycount to get everything crisp. And even then, it still wasn't crisp. Also, there is no flexibility. I know you can use layers and stuff, but I'm not a big fan of it as they sometimes caused displacement/normal baking issues. And now like this, the head is only 2 million polygons. Then, I pulled it into Substance, made the pores, and later applied it in ZBrush again and tweaked some more volumes and wrinkles. This is, to me, the best workflow and so much easier to handle. Below, I'll show how the head looks out of ZBrush without pores/how far I pushed it, and then how it looks with the applied pores created in Substance.

Head Sculpt Turntable

With and without pores applied

I didn't use any head scan data. Instead, I sculpted the head from a sphere and manually retopologized it.

EYE´S BREAKDOWN

I took an unconventional approach to creating the eyes, especially for games where it might not be very common. After watching Tom Newbury's Eye Course 3.0, which is primarily for film, I adapted a similar technique for my eyes, and I'm pleased with the results. Essentially, it involves using a sclera with an opacity/translucency map for the lens. Behind the lens, there's a circle with a hole for the iris texture, and further back, there's a square plane without any specularity and completely black to emulate the pupil. The eyes are complemented by a tearline/meniscus and separate caruncles. I'm not sure how common it is in games to use actual geometry for the iris and details like tearlines, or if it's usually done through shaders, but in this case, it worked well.

Below, you can see the eyes separated to make it easier to understand what's happening.

GARMENT HIGH POLY

Now we come to the part that definitely consumed the most time and was the trickiest of all. I must have restarted the garment about 3 or 4 times in total because I simply wasn't satisfied with it, and I wanted to push it as far as I could. Creating a base in Marvelous Designer was particularly challenging as I had to learn the physical properties of cloth from scratch. Overall, it was a tedious process to art direct the folds and everything else, especially considering I was completely new to this approach. 

Once I had the base, I imported it into ZBrush and tried to add as much detail as possible, along with additional geometry. As mentioned earlier, I referenced various images to inform my design, using a primary reference as well as elements from secondary references to develop my own concept. However, in hindsight, relying solely on my interpretation rather than sticking closely to established references and concepts may not have been the best approach. Some areas ended up overly busy and distracting, while others lacked definition. Moving forward, I'll likely adhere more closely to concrete references and concepts rather than improvising. 

Below, you'll find a short video showcasing the high-poly cloth rendered in ZBrush. It's worth noting that the fabric weaves and details were added for aesthetic purposes; I eventually recreated them in Substance for greater flexibility.

MARVELOUS BASE

Garment Isolated

MAPS BREAKDOWN

I handled all the texturing in Substance Painter. Texturing is something I would have loved to dedicate more time to, but as deadlines approached, I found myself pulled in different directions. Fixing poses, refining the groom, and other tasks took precedence, leaving me with limited time for texturing. Despite the time constraints, I did my best to maximize the quality of the textures. I've come to realize that achieving realistic cloth textures can be particularly challenging compared to skin texturing, which I've done several times before. While I started with skin texturing, which I found comfortable, I quickly learned that cloth texturing presented its own set of complexities.

GROOM

Regarding the groom, it was my first attempt at working with hair cards, and let me tell you, I completely underestimated it. Alongside the garment, it was undoubtedly one of the trickiest parts of the process, one that I ended up remaking many times.

The character went through various stages, from having a ponytail to dreadlocks and more, so I had to do a lot of reworking. Although I'm not 100% satisfied with the outcome, I'm still proud of completing my first groom.

I experimented with different methods, starting with manual soft selection in Maya, which was quite tedious. Then I tried out Blender's hair tools addon, Fibermesh in ZBrush, manual curves in Blender, and geometry nodes in Blender. However, the workflow I ultimately preferred, despite its time-consuming nature, was using GS Curve Tools in Maya, an addon I purchased specifically for creating controlled hair cards.

While the process was time-consuming and challenging at first, I've grown to enjoy working on grooms after getting the hang of it. Admittedly, there's room for improvement in terms of density and optimization, but I was adamant about avoiding visible edges and blocky shapes, which led me to create a slightly denser groom than what might be typical for a game character.

Additionally, I aimed to avoid the appearance of obvious hair cards, although I feel I didn't entirely succeed in that regard; they're still somewhat noticeable. Nevertheless, here's the groom isolated and in wireframe for your review.

For the hair textures, I relied on an invaluable tool called Fibershop. While I've seen methods for achieving similar results in XGen and other programs, they appeared tedious and inflexible. So, I went searching for alternatives and stumbled upon Fibershop. Let me tell you, if you're working with hair cards and need to create textures, Fibershop is the way to go. It made the process so much easier, faster, and more flexible, with plenty of customization options.

The fur on her vest was simply scattered in Blender using some pre-made fur hair cards. It definitely could look better and be more optimized, since the fur contributes significantly to the character's overall polycount. However, like many other aspects, I ran short on time to refine it further.

CLAY RENDERS

WIREFRAME AND POLYCOUNT STATS

Head mesh: 34'000 vertices

Full Outfit including fringe alpha planes (without arms and legs): 49'000 vertices

Groom mesh: 11'000 vertices

Herbs mesh: 14'000 vertices

Legs and arms mesh: 5000 vertices

Guitar mesh: 5000 vertices



I noticed the head appears quite dense. My inspiration came from examining the wireframes of characters like Ellie from "The Last of Us: Part II" ( My favorite game ever) which have a high level of detail. I aimed to achieve a similar quality and prioritized ensuring there were no visible edges on the face, as I believe it's the focal point. Regarding the eyebrows and eyelashes, I may have gone a bit overboard. I used multiple planes for each clump, which might have been excessive. My goal was to create a face with a cinematic quality, steering clear of a more game-like appearance. While this approach might not be standard in the industry, I saw it as an opportunity to experiment with techniques. I welcome any feedback you have.

POSING THE CHARACTER

For the posing, I relied on Accurig, which actually did a surprisingly good job. Nevertheless, I had to make numerous corrections with corrective blendshapes, especially with the cloth. I opted for Blender for this task because I find its simplicity for blendshapes much more appealing than Maya's, and it's incredibly easy to switch tools and compare before-and-after adjustments.

Below, you'll find a few before-and-after blendshape images for reference.

CREATING THE HERBS

This was my first attempt at creating plants entirely from scratch. Initially, I considered using Quixel atlases for them since they offer a wide range of amazing-looking options, which would have made the process quick and likely resulted in better-looking plants. However, I decided to challenge myself and create my own from scratch.

Initially, I tried modeling the leaves and then texturing them by hand in Substance, but this turned out to be time-consuming and the results didn't come close to the quality of Quixel's options. So, the next day, I headed to the forest in the morning and gathered some nice-looking plants and herbs. I then photographed them at home using my camera, with my drawing sketchbook serving as a white background to facilitate easier cutting out of the images.

Here are a few of the images I captured

Cutting out the images was the most time-consuming part, taking me about 6 hours in total. However, I enjoyed the process, simply listening to music while meticulously cutting out the plants.

Afterward, I created modular kits using these images. I placed the assembled texture from Affinity onto a plane and used the knife tool to cut them out. Then, I employed soft selection to give them a more organic appearance and duplicated them to ensure they looked good from all angles and not just like a flat plane.

Following that, I crafted the stems of the plants using cylinders and soft selection. Then, I utilized surface snapping with my modular kit to attach the plants, flowers, and leaves onto the stems, thereby creating finished plants.

Afterward, I moved on to the final step, which involved attaching the plants onto the belt with strings (once again utilizing Blender's curve tool for the strings) and tucking them into her waistband. While the plants may not hold up under close inspection, I'm satisfied with the overall appearance and believe it works well for a game character. The entire process, from gathering plants to attaching them to the belt, took about one and a half days.

GEORGIAN GUITAR (PANDURI)

Just a quick note about the panduri (Georgian guitar): I whipped up this asset in a rush, with only a single day from start to finish. So, yeah, it's not exactly complete, especially in terms of texturing. However, I felt it was still worth including because it serves its purpose adequately. Just wanted to let you know that I'm aware of its shortcomings and would have loved to refine it given more time.

Panduri Still Renders

Panduri Maps

CONCLUSION

While I wish I could have made several adjustments and had more time for refinement, I'm overall pleased with the outcome. This project marks my first experience creating a game character with grooming, cloth, and other elements. There were many firsts and challenges along the way, but after about 2.5 months of work, I can confidently say I'm proud of what I've achieved. I'm eagerly anticipating my next creature project, armed with the knowledge and skills I've gained from this one. Despite the intense time pressure and numerous setbacks, particularly with the garment and groom, I'm excited about the prospect of applying what I've learned to future projects. I'm open to any feedback you may have, and I hope you enjoyed seeing my work. Thank you for taking the time to read through my post.

Feel free to high five this project, it would mean a lot! And if you're interested in exploring more of my work, consider to follow and connect with me on LinkedIn. Thank you once again for your support!


Comments (7)