On 15th of September, the Pixelmon team hosted a live AMA on their Twitch channel where Pixelmon CEO Giulio Xiloyannis and lead 3D artist Cosmo Midias explored the Pixelmon modelling process from sculpting all the way to 3D animations.
Cosmo Midias is Pixelmon’s lead 3D design artist, responsible for the recent art update currently visible on both the Pixelmon website and on Opensea. He is based in Japan and has worked on multimedia projects for 10+ years with experience in both working on the projects himself and teaching 3D modelling, photography, editing and motion graphics. He has a passion for art and technology, and blends the two, which is both his focus and speciality.
During his introduction, Giulio explains why Cosmo was brought into the team and why he was so impressive as a candidate, as explained below;
‘His ability to switch between art concepts faster than any others, going to a level of detail faster than the others and to a level of quality that was frankly superior to all the other artists.’
A sample of Cosmo’s work on Pixelmon can be seen below and in further detail on the Pixelmon website and Opensea.
As mentioned above, the stream is rewatchable via the link: https://www.twitch.tv/videos/1591117104. Below is a selection of some of the questions and answers from the stream; however, many have been left out due to how visual the stream was. If possible, we recommend watching the stream for full effect.
Cosmo: Blender is the main pipeline used to create the animations and artwork. It is a free and open-source tool. Due to how detailed and thorough the software is, there’s no need for any other software as it can handle many of the tasks needed for the project. Blender is the ideal way to get started in all aspects of 3D creation.
Cosmo: If you want to be a sculptor, the recommended tool would be ZBRUSH. Nothing can beat its performance in terms of polygon count.
Cosmo: When I started, Blender wasn’t a great tool. I learnt to model in Blender and learnt a box modelling technique. In this process, everything starts from a cube and develops outwards. (This is a very visual process, and best to refer to the stream for detail). It’s a very old-school method, but one that works well.
Cosmo: The length of the modelling process varies a lot depending on the complexity. The Pixelmon models take around 3–5 hours, with Bublore, for example, taking 5 hours.
Cosmo: There is currently no target for the polygon count, as the current importance is placed on high-quality NFT art instead of being game-ready. The wireframe of some of the sculpted models, for example, the air tribe Dragon, is very dense, making them more challenging to fully animate.
Cosmo: It’s not difficult, just boring. When transferring to a game-ready, less polygon-rich model, you have to optimise the model’s topology through a mesh editor that can be used directly on a high poly mesh. You need as few polygons as possible to optimise the model for a high FPS game such as Pixelmon.
Rigify provides some templates, such as a horse, that can be modified to the models to speed up the rigging process.
Cosmo: To keep the workflow process clear, we try to limit the use of different tools and apps. The process of rigging is fundamentally creating a skeleton for your model. It starts by creating a bone-like structure called a metarig that lives inside the mesh. Once this is done, you can generate a control rig that the animators will use to create the model’s moves.
Cosmo: The process starts by defining the main pose for the animation. The animation begins with layers, with the base layer being a more simplified animation between a few key poses. Then, other complexity layers are added to the poses to convey a feeling of weight and more fluid movement based on Pixelmon’s specific traits, qualities and quirks.
Cosmo: I think that going forward, there will be a lot more in the AI art and generation field. There are a lot of both interesting and destructive innovations that will change the industry. I think that in animation, this shift will happen soon, resulting in a simplified way to give commands to models while the AI generates the movements and actions. I think this is coming very soon; the space is changing quickly, making animation a lot easier and more accessible.
Cosmo: I recommend reading ‘The Animator’s Survival Kit’ by the creator of ‘Rodger Rabbit’, Richard Williams. It’s an amazing course that shows a lot of animating techniques vital to 3D animation.
Cosmo: We target the procedural weight of texturing. In the beginning, these models had no texture. After hearing the community and the team’s feedback for further detail, we moved to procedurally generated textures through the procedural workflow available in Blender, which starts from noise. Once we had these initial textures, we refined them and edited them quickly to improve the models to the project’s needs.
Cosmo: It all started with video games. I played Sims, got into the internal files and started looking at the models and wondering how they were created. This interest led me to the internet and eventually Blender. Diving into the rabbit hole of modelling, I got into this field and got my first job in art in 2010 after self-learning from the information on the internet in augmented reality. Everything happened naturally, and it’s a career I’m incredibly passionate about.
This sneak peek is available at the linked timestamp below of the Twitch VOD: https://www.twitch.tv/videos/1591117104?t=1h27m35s