The Brief, The Team and The Franchise
Straight off the bat the first major project I will be undertaking this Trimester is the Meet The Team project. In a team of four we must each create a character/creature/vehicle that fits into an existing franchise. These characters have to be animated in a sequence that introduces them and will serve as a title sequence/promotional material for the franchise. Our team elected to make a 3D sequence and have gone with designing our characters to fit the theme and visual style of the existing franchise Bloodborne (the soulsborne franchise). We were all pretty excited for making characters that would fit in the Bloodborne universe, as it is a well developed style that we all think is really neat.
The process for creating these characters is the traditional 3D character pipeline starting at an iterative design process and ending at the final product which is the 3D animated sequence. The individual tasks of each team member is that they are responsible to design, create, rig and animate their character. As a group everyone will give feedback as we progress through the weeks and work on things such as project plans, storyboards, etc to help keep the team on an efficient path/workflow. The trailer for The Old Hunters DLC for Bloodborne served as great inspiration for this project, among other things.
Reference, More Reference, and More Reference
Before I start anything, before concepts and all, I get reference. I love reference, and I am an avid fan of using it obsessively. By using a lot, I take inspiration from different part of each source of reference, and in the end I have an original piece that has drawn inspiration from multiple sources. When I started the iterative design process I had to create concept art for a humanoid, creature and mechanical character. This means that I needed to collected reference for each type of character, and then compile them into an easy to view reference board.
These reference boards were for my first round of concepts, and each time I got closer to my ideal original character I would get more specific references. Alongside these I created a mood board which helped give an overall visual style of what our character theme should fit. I organised the mood board based on colour so that the darker reds/blood was on the right hand side and the left and middle consisted of light/orange colours and the night sky. I used a split complementary colour pallet when compiling the pictures together as it maintains high contrast but at the same time has less tension (“The Ultimate Color Combinations Cheat Sheet”, 2017), providing me with an overall nice aesthetic for the mood board, and helped keep it organised instead of all over the place. With a lot of the references collected I was moving into the bulk of the pre production phase, which is concept art.
I don’t claim to be a great 2D artist, I am very much far from it, however I think everyone has the ability to convey an idea through the use of silhouettes/rough line drawings. In my case this took a few hours to do, even though it looks like a toddler did it. The process can vary greatly depending on each persons skill, however this iterative design is a great way to hone in on an original character that has a cool design. At this point I was really thinking that I was wanting to make a humanoid so with my round two concepts I pushed for body shapes and costumes that would make a good silhouette on a human, rather than continuing to do concepts on creatures and mechanical characters.
Before I added detail I did five basic body shapes so I could see what would look good for my character. I was pretty surprised at this as I didn’t think I would change my mind, however I ended up liking a few different body shapes. This gave me good legroom for doing fairly unique round two concepts, as they all had a different shape. I also made sure to have fairly different clothing for each concept, that way I could pick and choose what I liked by the time I got to my final concept. Round two concepts took a lot longer than I originally anticipated because of a few reasons:
- I can draw moderately okay, however this is only because I spend a long time on each drawing
- Five polished concepts was what we were tasked with, so I didn’t lower quality on any of the drawings, even if I had a favourite already
- The more detail I did early on, the easier it would be for me later
In the end round two concepts set me a bit off schedule, however for once I was actually extremely happy with some drawings I did. It was quickly time to move onto round three though as I needed to try my best to catch up.
I had decided I was going to make a female character so I did some further research into female body shapes (also read as more reference!) and found that the a pear shaped body was what I ideally wanted to use. This helped tremendously as I was also able to find out the best way to dress this body shape in order to make it look good. By adding volume to the top of your body, as well as emphasizing the waist, you are able to create a balanced hourglass look (Staff et al., 2017). This description fit well with my second round two concept, and considering it was already one of my favourites made it an easy choice to base my final concept off of.
I really liked the idea of having a cape and a pirate hat (hunter hat), so I decided to have those as my main features and work around them. Adding a coat was a good supporting feature as it added the volume to the top that the character needed and at the same time kept it tight around the waist. I found adding belts to the character was a good way to keep it looking both unique, and a good way to have any potential loose clothing kept close to the body, like you would if were fighting beasts. This final round of concepts took me quite a while, as I made sure to have a clean model sheet, and multiple versions. One for the body shape, one for the pants, and one with everything. By doing it in stages it made it easier to build parts which might otherwise be hidden by the model sheet with everything in it. The final result was something I was really happy with, as I poured quite a bit of time into concepting, and I felt like it paid off in the end.
Character Creation Research Time
Character creation is a fairly long process, and there are many ways of going about it, although there is definitely a stand out workflow that most professionals seem to take. Thanks to the abundance of software, a few of these early stages in the workflow can be switched around a bit and done in different programs. From what I have found most professionals use a high poly to low poly work flow. This either starts with a base mesh being made in a program like 3ds Max, then being imported into a sculpting program like Mudbox for High Poly details to be added. Or alternatively it can be created in something like ZBrush, where the high poly mesh is either imported in or created from zSpheres/ Dynamesh. These meshes are then retopologised into a lower polygon mesh where the details can be baked onto, rather than using the ridiculously high poly mesh that was originally created.
This is a pretty crude description of it but stands to be the core of the production and modelling phase of a character creation pipeline. There is a lot more creating a character than just this, including things like baking maps, textures, rigging and pre production things like concepts etc. Below is an extremely detailed character pipeline tutorial in ZBrush and Maya, and runs through the process from start to finish. Alongside this I also did some forum crawling through Polycount to find peoples different workflows and how they differ. Some great examples are here, here, and here.
Modelling My Character
For my modelling phase I went with creating a base mesh in 3ds Max and then planned to take it into Mudbox for sculpting. Due to personal issues I fell incredibly far behind, although I will go over that in a post mortem blog post. Using the model sheet I created in the concepting phase, I modeled a medium poly mesh, making sure to add deforming parts to the mesh such as the knees, arms and I even tried the hips (not sure whether it would work well or not). One mistake I made though was forgetting to add deformation topology to the fingers, which will make it incredibly hard to animate later on as it will destroy them.
The problem I usually encounter is that I underestimate how long everything will take, and I am always quick to assume things won’t take that long. Boy am I wrong though. I end up creating something four or five times before I am happy with it, and considering how many parts there are to an organic character, this can be fairly time consuming. An example of this is I assumed the clothes wouldn’t take too long to create, however trying to create the coat look wasn’t as simply as just using the shell modifier, and ended up taking me way longer than it should have. The same went with the hair. I always get caught up on how to create it and make it look good, and usually can’t grasp the concept of it. Even just thinking about it can hurt my head as I can’t imagine how to make it properly. I wasn’t super proud with the character I ended up modelling, but due to the time constraint I ended up creating it was going to have to do.
Unwrapping and Quixel The Life Saver
Unwrapping and texturing was the part of the pipeline where I was able to make up for some lost time. Unwrapping is fairly simple, and because I was fairly far behind I took a very hack and slash method in order to get it done. Convert edges to seams and quick peel. This was more or less done for the whole model, except for the face which was a pelt. Combine this method with the quick and dirty pack together and re scale buttons in the UVW editor, and you have yourself a sub par but workable UV map. If this was for a stylized game character than such a dirty method wouldn’t have worked out alright, although thanks the fact that I would be using Quixel Suite to texture this wasn’t such a problem.
Before importing the model into Quixel I first created a Material ID map in 3ds Max. The Material ID is created by assigning each different part (hat might be one, skin another, etc) of your model with a bright one hundred percent self illuminated colour and then rendering it out using the UV Channel I created before. This part is more or less the core of making the quick and dirty UV map I created work well. One problem I did encounter when creating the Material ID was that I didn’t have enough padding set when rendering the ID. Also me being the incredibly talented person I am, I had deleted the save file with all the different colours applied to the model. This gave me two options:
- Reapply all the materials and re render with higher padding
- Salvage it in Photoshop
The second method was the one I went with, and thanks to the selection tools Color Range and Expand Selection I was able to quickly fix the problem I had created. Next up was importing it into Quixel and going to town. Quixel as a program can be super hit and miss. You either find that you are running into importing issues or something of the like, or it is running seamlessly. Luckily for me, I had a fairly smooth run with Quixel this time around. I added the smart materials that I wanted and played with their texture intensity/colour/opacity. This is a quick way to create a fairly good texture, although you can definitely get much better results if you go into texture painting etc. I more or less just added the smart materials and played with the sliders, however taking advantage of Quixel (other than abusing its really quick ability to make good textures) is definitely something I have in mind for future character creation (and assets in general).
The Joys Of Skinning and Rigging
Thanks to my amateur character model I created a huge nightmare for the rigging and skinning phase. When modelling, I thought it would be a great idea to create thickness to my clothes by adding things like skin modifiers, buttons, belts, etc. However, when all these accessories travel across multiple bones, lo and behold you are going to encounter the worst clipping you could imagine. Add to the fact that a lot of this is all in different sub meshes, and you have pure chaos. What else could I do to make this process harder? Pfft a CAT Rig will do just fine (oh god I really need to manage my time better next time).
This phase was a serious nightmare though. Making sure that everything around the thigh bones and hip bones was all skinned the same proved to be a pretty big task, and consisted of many tedious hours of selecting vertices, looping them, deselecting them and weighting. Rinse and repeat for all the different parts of the waist, and anywhere else the clothes intersected aka everywhere. Most of my rigging and skinning problems more or less stemmed from things I did wrong in the modelling phase. I am thankful that I have learnt not to make the same mistakes again, although it was a pretty brutal way to learn.
Time to Animate
Being the reference nut I am combined with my love for Bloodborne, I spent quite a while studying (could possibly be read as procrastinating (the lore is just really good okay)) the different move sets of the weapon I had recreated. This involved a lot of watching the character move around in game with it, for example in the clip below it is held over the shoulder and then swung from there when attacking. Animation reference also may or may not include me attempting to do those poses and walks in and around my house. This is more so I can imagine how the body would twist/turn and where each position would hold. Doing these poses is actually extremely helpful for me when I am key framing an animation as I am able to recreate my pose, and then exaggerate it even further because it is more aesthetically pleasing that way.
This video actually served as amazing reference, as it had not only move sets and attacks, it also had a very basic walk cycle that I could recreate. Alongside the in game footage I also heavily used basic 2D walk cycle tutorials, as they more or less transfer fairly well into 3D animation. This was a great source that I looked back on quite a lot while animating. One thing that did disappoint me was how poor the animations I made ended up being. A combination of a shit rig, and rushed animation ended up leaving a very mediocre result, and a horrible issue with the feet in the final sequence.
Sadly, due to the nature of the sequence, I wasn’t able to incorporate anything such as emotes, which is arguably one of the best parts of the Soulsbourne series. Being able to incorporate something like this would have been an awesome thing to do, however we were aiming for more of a serious sequence rather than showing the beauty of this series. Sometime in the future I would love to be able to pay homage to this in someway, but for now who knows.
I Didn’t Forget About The Sword
The sword followed a fairly similar creation process to the character, although due to my time constraints I didn’t actually get around to texturing it myself. Harley, my fellow team mate, made an awesome texture for it in Substance Painter which really helped make the character look much coolers, and provided for some awesome renders. Alongside the sword he also created the environment that our characters feature in, rendered and compiled the whole thing. This level of help aided me in catching up and I wouldn’t have been able to finish the project otherwise.
Header Image – Retrieved from Bloodborne The Old Hunters
Character Creation for Videogames. (2017). Art of Luis. Retrieved 9 May 2017, from http://www.artofluis.com/character-creation-for-videogames/
Character Modelling Pipeline research. (2017). polycount. Retrieved 9 May 2017, from http://polycount.com/discussion/115231/character-modelling-pipeline-research
Game character modeling pipeline w Maya and ZBrush. (2017). polycount. Retrieved 9 May 2017, from http://polycount.com/discussion/172336/game-character-modeling-pipeline-w-maya-and-zbrush
Game character production pipeline. (2017). polycount. Retrieved 9 May 2017, from http://polycount.com/discussion/140586/game-character-production-pipeline
Limb Topology – polycount. (2017). Wiki.polycount.com. Retrieved 9 May 2017, from http://wiki.polycount.com/wiki/Limb_Topology
Stylianos, J., Staff, S., Staff, S., Staff, S., Staff, S., & Staff, S. (2017). Body Shape: Pear. Shop Your Shape. Retrieved 8 May 2017, from https://shopyourshape.com/body-shapes/pear-shaped-body/
The Ultimate Color Combinations Cheat Sheet. (2017). BrightSide — Inspiration. Creativity. Wonder.. Retrieved 8 May 2017, from https://brightside.me/article/the-ultimate-color-combinations-cheat-sheet-92405/
tutorial-2 : walk cycle |. (2017). Angryanimator.com. Retrieved 9 May 2017, from http://www.angryanimator.com/word/2010/11/26/tutorial-2-walk-cycle/
ZBrush Tutorials > High-Resolution Game Character Creation Pipeline in ZBrush and Maya Tutorial > Pluralsight. (2017). Digitaltutors.com. Retrieved 9 May 2017, from https://www.digitaltutors.com/11/training.php?pid=1361