Tsubasa Nakai has told us about the production process behind the Lucifer project, explained the workflow in Maya, and talked about capturing facial expressions using Rokoko.
In case you missed it
You may find these articles interesting
Hi, I’m Tsubasa Nakai, a Creative Director from Japan. I worked for Shirogumi and Marza Animation Planet before. I started my personal work about 4 years ago and I have created many tutorials on YouTube or websites like 80 Level and ArtStation.
Since my previous 80 Level Article, I have been making personal works using mainly Unreal Engine and also animating characters with a help of a motion capture device. I really like the workflow that allows me to create works that are more interactive than just a simple pre-render.
The Lucifer Project
I started modeling in early September and the production lasted for about a month. The goal of this project was to express Lucifer, the devil and the fallen angel, in my style.
I wanted to express a dark taste in POP so I was very particular about the facial expression.
Face and Body
Regarding modeling, I sculpted the model with ZBrush and inflated my imagination without worrying about the number of polygons. After finishing the sculpt, I wrapped the production model with a clean topology to sculpt model.
This time I used the MetaHuman model as base mesh because I wanted to reuse BlendShapes made by a MetaHuman rig. I tried to make a slightly stylized taste, so I had a hard time adjusting the balance of the face.
I planned the best way to achieve my goal. Accessories such as jewels are very effective in improving reality, but it takes time to make them. These days, You can get these props easily as KitBash on sites such as Artstaion, etc, so there is no way not to use them. How quickly I can output my image is an eternal theme for me.
Hair and Fur
Hair and fur are generated and adjusted with XGen in Maya, and then brought to Unreal Engine via Alembic. In this project, I used the hair shader used in MetaHuman and it was able to express beauty scatter from the backlight. However, if it is raytraced, there was some flickering. I think it is effective for animation if you can avoid flickering with a denoiser.
The facial capture is really simple. After taking a facial capture with Rokoko, I just exported the facial template mesh to Maya and copy and paste the animation key to my base mesh with 52 blendshapes used in ARkit. The blendshape was extracted from MetaHuman. I fine-tuned the mouth opening condition with Maya, but I didn’t need a big adjustment. I want to try to be able to achieve beautiful animation in real-time with a stylized character.
All textures were created with Substance Painter. I don’t spend much time making textures. This is because you can get very good textures or materials on Substance Source and ArtStation and you will get good results with them in an instant. The important thing for material is if the model is good or not.
Are the edges properly beveled to get the beautiful highlights? Is there a three-dimensional effect for good shading? These two are very important to get good results for materials. Creating a number of material layers for one mesh can express more realistic and complicated information. Think about the history of the assets and making many material layers.
Rendering and Lighting
To achieve the desired look, I had to be very careful with the lighting. I adjusted it so that there were beautiful highlights in the hair and powerful eyes. Raytracing is used for shadows and GI, but even if it does not do, I think that a near picture comes out.
As for post-processing, I don’t make many adjustments in postProcessVolume and I mainly adjust the look with LUTs. Here's a great tutorial on how to use LUTs in UE.
The quality of images or videos created by individual artists is improving year by year, and I’m looking forward to making the leap forward in personal production.
You may find these articles interesting