Even though at first I wanted to update my blog every week, I now decided to do it each time I finish an important process, to share what I have learnt, what problems I encountered and how I solved them. I feel like this approach fits me better and I will find more things to mention here.
In the beginning I thought I wasn’t going to enjoy this project as much as I did. I’ve always had an irrational fear of modeling hard surface assets, since I am not a huge 3Ds Max fan and I used to rely solely on ZBrush when creating the high poly version of the character. For the entire summer I tried to dig deeper into ZBrush’s features, learn new “tricks”, work faster. The most useful tool for me that I started using only recently is be ZModeller. Before knowing anything about it, whenever I had topology issues, hidden polygons, holes or thin meshes I had to use a combination of ZRemesher, DynaMesh, colse holes, weld points, extract, inflate, or other ZBrush features to work my way around the problem.
However, all the new knowledge didn't help me much when it came to hard surface modelling. This time, I knew I had to raise my game at strict modelling if I wanted to finish this project on time. I started by organising my workspace and shortcuts so I wouldn't spend most of the time looking for menus, commands scrolling down, an so on. Despite trying my best to work as much (and efficient!) as I can, some personal issues made me take a couple of days off.
Still, by the end of the third week I had started working on the high poly sculpt. Again, I went on to tidy up my workspace and edit the interface to fit the purpose of this project. A great advantage when using this workflow (3Ds Max -> Zbrush) was that I was able to keep subdivision levels until a late stage of the sculpture and I could easily make adjustments to the main shapes anytime. I think shape design is exactly what makes the concept look good and I wanted to make sure I capture everything as accurately as I could. However, there were parts of the armour where I couldn't clearly distinguish the exact forms - the gauntlet and the chain mail. For the chain mail, I made a regular chain mail shirt using NanoMesh, with round chain links, but it looked very different to all the other pieces and it didn't fit in at all. So, I made my research on various types of chain mail and tried to come up with something that would feel natural and match the rest of the armour. I collected more reference pictures and made this second reference sheet:
Screenshots of the first chain mail and gauntlet versions:
I also had to worry about mobility. The chain links, elongated as they are, wouldn't allow the knight to freely bend his arms. As a potential solving to this, I tried adding a velvet shirt underneath the chain mail. Here are sone renders of the final sculpture:
I quite liked the result, so I kept it like that, without worrying too much about how I was going to make a velvet material in Substance/Marmoset. Looking back, I think it wasn't a great idea. In order to convey the look of the material I had to actually set the cloth's metallic value to 1, so it could get dark reflections from the background. Additionally, I found out that if I change the Diffusion to Microfiber in Marmoset, I get a similar result to increasing the sheen value in Vray. The downside of this is that I could not add subsurface scattering, but I accepted the compromise just to make the velvet look a bit better.
Probably the most essential lesson this project has taught me is how important file backup is. After finishing the sculpt, I decimated it, but forgot to export the previous, non-decimated model. So, when I got to bake my mesh maps, most parts of the armour had serious baking issues on the edges. I had to log in on all the computers I used to see if I could get an autoback and luckily I did.
The most enjoyable part of the project was painting the textures. I got to explore a lot, compare the results and ask myself questions, such as "which parts of the armour are more exposed than the others?" or "Is the armour supposed to have any damage at all? what if it is new?"
I started with a base metal on the armour, with the roughness value set to 0.5, then I added a bunch of layers with colour and Roughness variations. In the original concept the knight looks quite clean, with minimal surface damage to his armour. Still, I added some superficial scratches and small black spots, as in the third reference sheet. I liked the look, but I wanted to push it further. I feel like the extra damage adds a lot to the story behind a character. I gradually increased the colour variation, making the cavities a darker shade of gray/black and decreasing the metallic value. Then started to add more and more height variation as if corrosion started to affect the metal. However, if I were to start over, I would definitely create all these surface details in ZBrush from the beginning, since it's easier to control their shape and depth than in Substance.
The next step, rigging the character, might be my least favourite thing...ever. He looks a bit stiff right now, but I didn't worry too much about it. I can't imagine how much more comfortable one can stand while wearing an armour like this. If I have enough time in the winter break, i will address this issue, too.
These are my final shots. Overall, I am happy with the outcome. There were issues along the way, but i learnt a lot by trying to solve them. One thing that I will definitely change during the winter break is the length of the cape - I will make it shorter, to ease the movement.
No comments:
Post a Comment